In article ,
"David J Taylor" -this-bit writes:
| "Nick Maclaren" wrote in message
| ...
|
| Believe it or not, this is a computing question! I am interested
| in finding out whether there is a generally accepted formula for
| converting TAI to solar time and, much more importantly, whether
| there are generally accepted bounds for its uncertainty. I have
| a suspicion that there isn't :-)
|
| Nick, could you point to a reference as to _which_ solar time you mean?
| Do you mean the 20minute or so variation between (for example) local noon
| and highest solar elevation? The so called "Equation of the Sun", if I
| recall correctly?

Oops. Mea Culpa. I posted before doing enough research to check
that all of the terms I used were well-defined :-(

The problem is this:

I want to be able to convert TAI to UTC/GMT/civil time at Greenwich,
for every date from the origin to the heat death of the universe.
Now, obviously, doing so to the nearest second is a failing known
as Delusion of Accuracy, so I am not making that mistake. But,
equally, simple numerical accuracy is not enough, as a historical
record may well be a 'precise' time of day.

Now, the conversion between TAI and UTC (or, rather, the one that
computer people always refer to as UTC) is defined from its start
to the present. GMT is a bit messier, but not too bad. But I
should like to know error bounds for civil solar time before GMT
and into the future.

Note that, because of the way that I am thinking of doing this,
I don't need the ACTUAL correction - what I need to define my
interface is some approximate BOUNDS on the correction. This is
to know how many bits to allow for it rather than to specify a
value. And, yes, of course I am thinking of using a sort of
floating-point format for far-flung times :-)


Regards,
Nick Maclaren.