> Second, I might attempt to write code that computes positions for
> these objects by interpolating the ephemerides. I dislike this idea
> (compared to just using elements, it's a horrible kludge), but it may
> be the only option.

This isn't the ONLY option. Another option would be to use e.g.
Gauss' method to compute the orbial elements from three ephemeris
positions. If you want to be fancy, you could, after having obtained
preliminary orbital elements from Gauss' method, even refine these
elements from a least-square analysis using all ephemeris positions.
In this way the orbital elements used to compute the ephemeris ought
to be recreated.

----------------------------------------------------------------
Paul Schlyter, Swedish Amateur Astronomer's Society (SAAF)
Grev Turegatan 40, S-114 38 Stockholm, SWEDEN
e-mail: pausch at saaf dot se or paul.schlyter at ausys dot se
WWW: http://hotel04.ausys.se/pausch http://welcome.to/pausch