I only asked about the choice of bearing. Any one else using cheap ball race on a shortstroke motor?
I would imagine with a ball race fitted it would be difficult to measure between the timing bush face and the crank, (there will be no end float and access difficult) but as long as the crank pokes through the bush as usual and the cases bolt together without the crank binding when cold, then sufficient axial clearance should exist between the timing bush face and the crank when the motor is hot, allowing lateral thermal expansion of the crank towards the bush.
The problem comes with those sideways forces, and I can see that a ball race firmly clamped to the crank will either restrain the crank, or be moved bodily sideways, all dependant on the forces holding the bearing outer race to the crankcase. Maybe this is the reason for the change from ball to roller.
As for the bearing fit in the case, the coefficient of thermal expansion is known for the differing metals, we know the original size "cold". We know the expected change in temperature so can calculate the change in size with increasing temperature for a hot motor. The trick is to make sure an interference fit still exists at the higher temperature and the alloy case is originally machined to accommodate this. All fine and dandy, but 60 years later don't worry about it too much, just reach for the Loctite Bearing Fit if the case is a bit slack.
Bergs fit I would describe as a shrink fit, but either way we are relying on the friction between the two components.
Looks like I've got enlightenment as well as "O" level physics.
Swarfy