The Park tool headset reamer for standard 44mm headset is 43.95mm. This means, assuming the headset cup is exactly 44.0mm, that there is an interference fit of 0.05mm (0.002") between the bore of the head tube and the headset.

In any normal machining context, an interference fit of 0.05mm on this diameter would be considered an extreme or even impossible amount of interference. The rule of thumb that I was taught in my machining career for interference fits, was "0.0005 inch interference per inch of diameter" for a strong interference fit. That means the Park reamer is giving ~5 times more interference than I would have expected.

Clearly, the reason for this large amount of interference is that the bicycle tubing is elastic. So the rules must be different for tubing than for normal bores in castings or forgings. So, we can get away with more interference, or maybe we even need the extra interference because of the elastic tubing.

Headset tubing is also not round, either after machining or after brazing. Even after reaming, I find head tubes are often 0.05mm or more oval, because they spring back a bit after reaming. But this isn't my major concern. The tubing will stretch back to round when the headset cup is pressed in. It does mean that getting a good diameter measurement is hard and may require taking at least 2 or preferably more measurements and averaging them.

I make my own head tubes usually, and I have the freedom to bore them however I want. Yes, they don't end up perfectly round after brazing, but as I mentioned above, they aren't perfectly round after reaming either because of springback. So I'm considering stopping the reaming altogether and just boring them to final diameter when I make them in my lathe. So should I target 43.95mm, like the Park reamers? Or something else?

Is 43.95 really the best for all frames, and has anyone used different tolerances on purpose? I've seen pretty big differences in head tubes...such as thin steel road-bike head tubes with the barest hint of a reinforcement ring. Or titanium head tubes, which is twice as springy as steel. Or gnarly thick Aluminum fat bike head tubes, or thick steel BMX head tubes. It seems unlikely that they all work well with the same amount of interference. Does anyone deliberately deviate within the 43.95-44.0 range?