Hi Bob,

I suspect we aren't looking at the same image, and that you're
getting one with a corrupted FITS header. (I say that because it
sounds as if Guide is handling FITS images -- extracted from the
data on the second Guide disk -- just fine; it's only having
images with these STScI images.)

Just to check: I went to

http://archive.stsci.edu/cgi-bin/dss_plate_finder

and entered the coordinates you gave:

RA 2000 17 26 32.15
DEC 2000 -28 49' 41.8"

... and left the default (J2000) equinox, 15' by 15', FITS,
Save. I clicked on "Find Plates", and got a list of possible
plates. I clicked on "Second Epoch Southern (UK Schmidt)",
choosing the 1992-07-21 version. Then I clicked on "Extract
Image from Selected Plate", and saved the resulting 1.6 MByte
file as dss.fit.

The image is tilted by about two degrees clockwise (as viewed
with "chart uninverted" in the Inversion menu), which causes the
stars in the image to line up with GSC and UCAC-3 stars. (With
an error of about 1.5 arcsec, as I described before.)

I should have asked this before: is the image you downloaded
tilted by about two degrees? I'm wondering if it appears completely
"level", but as if it would match up with stars if it _were_
rotated by about two degrees clockwise? (In which case, I'd
start thinking that Guide was somehow getting the central RA/dec
and pixel sizes, but not much else.)

Another "should have asked before": what version date do you
get in Help... About Guide?

-- Bill