Most everything on a modern bicycle is built to metric measurements - see, for example this question. Major exceptions include bearing balls, where metric balls don't even merit mention on Sheldon's site. This results in some wacky mixed engineering, like cones that are precision manufactured to metric measurements, apart from a race curve that is precision manufactured to an imperial radius.

What is the history behind this? Why wasn't there a Continental European standard that caught on after metric became almost universal? Even French bikes from the seventies, which have metric versions of everything else, use imperial bearing balls. I guess the same question could be asked about the pedal pitch mentioned this question, but I'm guessing that's because the Brits came up with reverse threading first.

  • Imperial threads show up in a few other places too, such as 3/8" wheel axles
    – Chris H
    Nov 3, 2017 at 13:28
  • Those are a lot less common - e.g., Shimano doesn't use imperial threaded axles. As far as I know, a metric bearing ball standard doesn't even exist on bikes. Nov 3, 2017 at 14:17
  • They were common a few years ago on otherwise metric bikes (I have a bike with a metric front axle and imperial rear, for example). These are axles with nuts and dropouts, which themselves aren;t very common
    – Chris H
    Nov 3, 2017 at 15:11
  • 2
    Because we humans like to make things complicated and people tend to resist change.
    – bated
    Nov 3, 2017 at 15:13
  • 1
    @Criggie Finding metric ball bearings isn't hard at all - it's just hard in a bike shop. I'm asking why bikes are designed with imperial ball bearings when they are otherwise metric. France's attempt to make everything a metric standard in the 70's was a flop too, but they never tried to make metric ball bearings. Nov 12, 2017 at 0:09

2 Answers 2


My father worked at Fafnir Bearings for 35 years and even though it ceased to exist as an independent company about 50 years ago, the trademark is still active. In the post World War 2 era, Japan rebuilt its bearing industry using Imperial standards. Bearings are a strategic technology in a manufacturing society and any rational government will promote policies that protects and promotes important sectors. At the end of WW 2, the US bearing industry had the right processes to make the best bearings. They had the “recipe”. Bearing plants in Japan and Germany were prime bombing targets and their assets, including people were very much destroyed. This made US and to some extent English bearings the target to reverse engineer after the war. If you need to figure out someones recipe, you should stick to making the same thing. Annealing times and temperatures are dependent on the dimensions of steel. Automotive and bike bearings were nearly optimal at a low price point, in the 1950’s . The machines and tooling were well established, and an export driven economy would want to produce standard products.

I think it simply is a matter that making these sorts of bearings didn’t merit the resources to shift to a different standard.

Another factor is that as America has deindustrialized, the machines and tooling were sold off to Asian countries.

  • I have a vague memory of "Swiss bearings being flown out to Allied forces in a Mosquito, but they have enough so tonight the plane will carry two escaped POWs" possibly from the book Colditz ?
    – Criggie
    Nov 11, 2017 at 22:35

I don't have the real history of industry type explanation, which would probably involve the personalities and business leanings of people working at OEM hub/headset/BB/pedal manufacturers over the decades and so is probably lost in time, but the simple answer is that large-scale technical choices in the bike industry get made based on what's simplest to implement at the manufacturing level, what's cheapest, what works from a patenting/licensing legalities standpoint, and what solves actual problems for those with decision-making power and/or increases their profits. Confusingly, there actually is a pretty strong "if it ain't broke, don't fix it" attitude, which one wouldn't necessarily guess from the recent standards explosion, but if you dig a little deeper into the rationale of the relevant actors behind each new standard you'll see that they're at least attempting to meet the above criteria. (Note that "increases their profits" is the big catch-all behind a lot of what's happened.)

There are a lot of places on modern bikes the world over where this dynamic of split metric and imperial units exists, or exists in a veiled way with a metric dimension based on an imperial size rounded off to the nearest tenth of a millimeter.

  • Grips are almost all 22.2mmm (7/8" = 22.225mm)
  • ISO threaded bottom brackets (1.375" X 24tpi)
  • The classic frame tube diameters are all different combinations of 25.4mm (1"), 28.6 (1-1/8" = 28.575mm), 31.8mm (1-1/4" = 31.75mm), and 34.9mm (1-3/8" = 34.925mm). Manufacturers have largely moved away from simple round tubes but we still see 28.6, 31.8, and 34.9 as the clamp-on front derailer standards.
  • Handlebar clamps for non-oversize flat bar bikes are 25.4mm (1"). The industry did standardize a metric-based similar size (26.0mm) for upper end road bars, but then when oversize came out everything became 31.8mm (1-1/4"). The new 35mm standard somewhat reverses the trend yet again.
  • BMX/youth bikes usually have 22.2mm (7/8") or 25.4 (1") seatposts
  • All modern pedal threads have imperial sizes (9/16" x 20tpi or 1/2" x 20tpi), even though all their wrench flats are 15mm, 8mm, or 6mm.
  • Axle threads for many hubs are either 3/8" x 26tpi or 3/8" x 24tpi, namely coaster hubs, internally geared hubs, some low-end derailer rear hubs, many low-end front hubs, BMX fronts (even if their rears are 14mm), etc.

And there are more examples yet. Another side to this however is that in industry the imperial system is really only the metric system in disguise, since in 1930 and 1933 the UK and US respectively began using 25.4mm as the reference dimension that defines one inch. That's why all the the fractional conversions are relatively clean numbers. But it means that the high-level work of keeping track of what defines either a millimeter or an inch in terms of physical and theoretical references is the same for both systems, and is done by the same organizations.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.