Slews which are very curved can produce exhorbitantly large images
which do not maintain the flat projection and hence contain points with
increasingly inaccurate astrometry. This happens in about 1% of slews.
AT points of inflection the interpolation of the curve fails a
bit. This produces errors of arcmin at the worst.
NB: The slew images have very few photons (typically about 100
in half a square degree). For reasons which are not fully understood
the more recent versions of the source detection chain produce a large number
of spurious sources from these images. At the moment it is recommended
to use SAS 6.1 to source search slew images.