For a roll test I like to do six tries for each ball with a different axis to the left on each roll. I use a small arrow marked on the ball so I can control the orientation.Four times; same results.
It's also good to do a separate test for roundness which the roll test is not guaranteed to find. I have gauges for this -- round holes in steel plates.
I have seen a tool for this also -- the ball sits on three bearings and a dial indicator gauge touches the ball at the top. The ball is rotated in various positions and changes in the dial are noted. I suppose you could calibrate the dial to tell you diameter but it's not exactly a diameter. More expensive commercial machines can measure sphericity to one microinch or better.
Trivia: it is possible for something that's nearly a sphere to have all diameters equal but not actually be a sphere.
Last edited: