As the night went on and I ran more tests, the star
shapes got worse and worse. This is a comparison
of the four corners from when I first thought I had
arrived at good collimation, and the four corners
from an image 2 1/2 hours later. The latter were taken
in about the same part of the sky as the more northerly
set in the previous image, and is clearly worse than
any of those earlier images. So, now I'm wondering whether
the shifts are due to the angle of the scope by itself,
or if there is something that is shifting more and
more as time goes by.