The first time the U.S. tried to study this, they asked for a sample of parts, specifying a precision range, noting that it was acceptable for up to 10% of the parts to be at the extremes of said range.
The Japanese company sent them 3 bags of parts noting: The bag with the full quantity is made to match the provided blueprint, the other two bags contain 1/10th as many parts which are either larger or smaller as requested. May we ask what the purpose of these other parts is?
When they started measuring the parts, they were all identical --- part-way through, they decided that their micrometer must be broken and sent for another --- it revealed that all parts where _exactly_ the right size per the specifications of the drawing.
I've heard this story ever since I was a kid, but never been able to find a credible source. I'm compelled to believe it's no more than a story, at this point.
It was in an audiobook narrated by Pat Morita that I heard it.
The context was auto transmissions --- it was found that one source of premature failures was when all parts in a transmission were at the extreme of too small or too large --- on average, the large/small would balance out, but when it didn't that was when there were early (and catastrophic) failures.
The Japanese company sent them 3 bags of parts noting: The bag with the full quantity is made to match the provided blueprint, the other two bags contain 1/10th as many parts which are either larger or smaller as requested. May we ask what the purpose of these other parts is?
When they started measuring the parts, they were all identical --- part-way through, they decided that their micrometer must be broken and sent for another --- it revealed that all parts where _exactly_ the right size per the specifications of the drawing.