Jon Elson writes:

If the top of the knee is the big problem, you can make sure the
gibs allow the saddle to move all the way forward and back.
Then, rig a dial indicator to measure rocking of the saddle,
and use the table as a lever to try rocking the saddle. See how
much looser it is in the worn area.

The "shake" test is OK as far as it goes, but the knee-saddle-table system
involves no less than 24 different precision bearing surfaces. A simple
straight edge and feeler gages will map some wear, although jigs and other
standards are needed for complete characterization. For any detailed
measurements you must remove the table and saddle. Mating surfaces with
complementary concave/convex wear maintain rigidity and pass the "shake"
test while degrading positional precision.