Bill, how does Find_Orb select which observations it will use in
getting the orbit? I ask because I used the same data you did
except I excluded everything before 1997 as the codes indicated that
most of it was approximate or had other problems. This left 296
observations from 1997-2003. I clicked on Auto Solve and Find_Orb
gave me the following orbit which is nearly identical to yours:

Orbital elements:
107
Perihelion 2006 Dec 18.376379 TT
Epoch 2004 Sep 5.38 TT = JDT 2453253.88
M 233.31256 (2000.0) P Q
n 0.15190406 Peri. 309.69832 -0.54377796 -
0.83897086
a 3.4787491 Node 173.14678 0.81346035 -
0.53301886
e 0.0790285 Incl. 10.04755 0.20636805 -
0.10963015
P 6.49 H 7.2 G 0.15 q 3.2038286
From 296 observations 1997 Jul. 12-2003 Oct. 8; RMS error 0.795
arcseconds

My result seems to be using all the data points, but it seems like
if I click on Herget or Full Step not all the points are used
(perhaps this is only initially). Please explain the process as to
how points are selected and then how the range is expanded as the
solution is obtained.

Also, as a working rule, should one exclude those observations
marked with A or T or other codes which indicate that they have
potential errors? Of course doing so cuts out nearly all the old
observations from many years back. Also, I noticed that you
excluded all but 3 observatories. Which are the "good ones" that
should NOT be excluded. I assume that those you did exclude don't
have such good equipment for very percise measurements.

--- In guide-user@yahoogroups.com, Bill J Gray <pluto@p...> wrote:
> Hi John,
>
> Find_Orb is expecting the Minor Planet Center "standard"
format:
>
> http://cfa-www.harvard.edu/iau/info/OpticalObs.html
>
> This is the standard format produced by astrometry software,
and
> under the normal use by observers, it's just fine. It does have
some
> lacks, such as no way to specify error estimates... which is
probably
> part of why the folks at AstDyS came up with their RWO format.
(MPC
> is in the process of adopting a new observation format.)
>
> I was able to convert the .RWO file to the MPC format without
too
> much trouble, and the result is at
>
> http://www.projectpluto.com/107.ast
>
> Note that only observations from three observatories are
included:
> (673) Table Mountain, (689) Flagstaff, and the transit-instrument
> based astrometry from (950) La Palma (the Carlsbad Meridian
Circle).
> It's _really_ important here to use only _good_ observations.
Throw
> in the stuff from the surveys (which emphasize getting lots
of "pretty
> good", but not "insane accuracy-freak good" data), and your track
> will be thrown off horribly.
>
> Using the above data in Find_Orb, setting an epoch of 5
September
> 2004, I got the following orbit:
>
> Orbital elements:
> (107)
> Perihelion 2006 Dec 18.377178 TT
> Epoch 2004 Sep 5.0 TT = JDT 2453253.5 Find_Orb
> M 233.25471 (2000.0) P Q
> n 0.15190406 Peri. 309.69831 -0.54378031 -0.83896936
> a 3.4787491 Node 173.14696 0.81345892 -0.53302102
> e 0.0790283 Incl. 10.04755 0.20636747 -0.10963117
> P 6.49 H 7.1 G 0.15 q 3.2038295
> From 166 observations 1996 Apr. 18-2003 Jun. 16; RMS error
0.191 arcseconds
>
> I then saved the elements to a text file, went into Guide's
> Extras... Add MPC Comets/Asteroids function, and clicked on "Add
> MPC Comets/Asteroids" at the bottom and selected the text file I'd
> just created. Guide replaced its "usual" elements for (107) with
> those above. (You can confirm this by clicking on the asteroid,
> then ask for "more info". Guide will list the epoch among that
> info.)
>
> After clicking on asteroid and star, then on "Show Eclipse",
I got:
>
> http://www.projectpluto.com/107.png (about 23 KBytes)
>
> ...which does show a little bit of a southerly drop to the
path.
> Still, remember that more data is apt to be gathered before 5
September.
>
> -- Bill