Hello all,

I've posted a new "development" version at

http://www.projectpluto.com/pluto/devel/find_orb.htm

The main improvement is in the error model. Uncertainties should be more
"realistic" than was previously the case, especially for fast-moving objects.

This really should have been released about a month ago. The delay resulted
from my trying to document the new methods for setting uncertainties in position,
time, and magnitude. The documentation became a twisted mess, persuading me
I must have made the new system much more confusing than it ought to be, so I
had to revise it heavily. The current version is, I think, a _lot_ easier to
figure out.

I'd highly recommend switching over to this version. Despite the term
"development", it's probably more stable than the "release" version. As is
described at the above link, switching from "observations have weights" to
"observations have uncertainties" represented a considerable change in
mind-set and code. The result probably zapped a few bugs in the process.
I know of at least one: when first loading an object, Find_Orb would give
unrealistically large sigmas for the mean anomaly and time of perihelion.
You could click on "full step" and see a sudden decrease in both uncertainties.

Usual request: if something is broken or confusing, please let me know
about it. Once I get some feedback, I'll move this to the "release" version.

-- Bill