I think by now that you can see where this is going. As we reduce the average gap between the two components, by shifting the average bearing radius down, the proportion of assemblies which are too loose goes down, while the proportion of random pairings which will not fit together goes up. Therefore, the figures for the next two attempts will be given in a more condensed form. You might like to produce tables like the ones for the previous attempts to verify the results for yourself.

Attempt 4

If the average bearing radius is set at 10.02:

• About 32% of the time a randomly selected shaft and bearing will not fit together.
  
  
• About 3% of the assemblies will be too loose.
  
  
• About 29% of the assemblies will be ideal.

Given these percentages, we will have to go through the assembly process about 1471 times to produce 1000 assemblies, of which 956 will be acceptable, and only 427 will be ideal.

Attempt 5

If the average bearing radius is set at 10.01:

• About 41% of the time a randomly selected shaft and bearing will not fit together.
  
  
• About 2% of the assemblies will be too loose.
  
  
• About 24% of the assemblies will be ideal.

For these pecentages, we will have to go through the assembly process about 1693 times to produce 1000 assemblies, of which 966 will be acceptable, and only 406 will be ideal.