Originally Posted by
smontanaro
As "not a framebuilder," I'm curious about the magnitude these terms represent. I ask because... I recall a Dave Moulton blog post (I think) from several years ago where he described learning framebuilding from Pop Hodge.* As I recall, he described laying out the frame on the shop's brick floor. That would represent the extreme of catywumpus I suspect. (I also suspect Pop Hodge's frames were mostly built for a different clientele than today's modern builders.)
Also, if anybody has a reference for the Pop Hodge blog post, I would appreciate a link. The last time I went looking I couldn't find it.
* It still boggles my mind if I think too deeply about the time covered by the overlapping careers of Pop Hodge and Dave Moulton, probably over 85 years.
The maximum amount of error people see on tables and things is about 3mm over a distance of 500mm or so. That's about a third of a degree. If I weld a BB shell to a ST it looks absolutely square if I hold a 1 foot ruler flat to the shell and look at the gap between the ruler and the ST. But sometimes I can see a tiny error if I use a metre ruler all the way up the ST. That's going to be even less-- like 0.1 degrees or so, which over the length of the crank will not be perceptible. If you can't see it you certainly can't feel it.
A Reynolds frame tube has a straightness tolerance of about 1.5mm TIR ("Total Indicator Reading"). I think that means if you lay it down flat one end can be up to 1.5mm higher than the other. That would be about 0.14 degrees. People trying to align frames to super-high tolerances ought to be measuring this stuff or the corrections they're making may not actually be helping.
When they laid things out on the shop floor they may have just realigned the frames afterwards I guess, although this is not ideal.
These tolerances of around a mm or so over half a metre are the best you're going to get with a structure made like a bike frame, and plenty good enough. The tubes themselves are made to that kind of tolerance and you are always going to get some distortion from the heat, although this can be managed. Most of it affects the rear triangle in an obvious way and that can be corrected for. But what you can see with a good eye and measure with string is in this 1mm over 0.5m sort of range.
The lengths of the tubes and the angles are going to be a similar kind of tolerance if the frame is hand-made. I measure where I want to cut with a metre ruler, make a mark with a sharpie, and put my mitre template on. When it comes to the last tube in a loop (the TT and the SS) I start a little bit long and sneak up on the right length to make the angles right as measured with my Chinesium digital angle box. But idk how accurate that really is, and if the reading is off by a tenth of a degree or so I don't fuss about it (the display is probably more precise than the actual tool is accurate).
When you make something you try to make it as perfect as you can but TBH it can probably be off by a good deal before it will really affect the riding. But if it's
visibly off like the wheel is closer to one stay than the other etc. that will quite rightly bug the hell out of you.
There are much tighter tolerances for the facing of the HT and the BB shell and the IDs of the HT and the ST-- about 0.2mm or so. That's why you need to machine those.
You could make a much finer toleranced bike if you machined the whole thing from billet aluminium like these guys:
https://polebicycles.com/machine/. But unless you're planning on riding it around the inside of a stellerator fusion reactor I don't think it's going to make any difference in the real world (tolerances aren't the reason Pole machine their frames-- it's so they can use 7000 series aluminium which isn't weldable).