In article <4cho2j$1q3@cello.hpl.hp.com>, jacobson@cello.hpl.hp.com says... (most, about AF error at short end of a zoom, deleted, alas)
>There is really no way for a particular lens to be very much off on AF
>if the above test passes.

Just tossing this out: What if the lens has serious field curvature -
it can sure mess up split-image RF focus, so what about AF, if it also
uses lens edges for focus information?
Hope this Helps