In this video the "path" left by the ball bearings in the cone is quite wide

Cone from Park Tool video

whereas the path in my cone (on a road bike) is very narrow. I am wondering if perhaps the narrow path meant higher stresses, resulting in the pitted areas visible in the picture.

my cone

I am also wondering whether the difference in the width of the paths results from:

  1. the smaller radius of curvature of the cones closely matches that of the ball bearings in the bike disassembled in the video, but not on my bike, or
  2. the cone in the video has survived for so long that the "path" has kept getting wider, until it reached this width?

More importantly, my cones look like they need replacing. Is it correct to assume that no one changes just a pair of cones, and that one just replaces the entire axle, with nuts, washers, cones, reinforced rubber seals, and perhaps even the skewer?

my axle

If my axle is 109 mm long and has 8 mm diameter, and each of my cones has 15 mm outside diameter; 8 mm inside diameter; and 12 mm height, how do I identify the appropriate replacement?

Bonus Question

Is sanding the cones ever sensible?

  • Generally, you’d just go on Google or whatever and search up your hub model, followed by “…replacement cones”. The only way sanding would make sense is if your sander is actually a precision grinder that can regrind the cones to match a larger ball size.
    – MaplePanda
    Mar 19, 2022 at 2:51
  • 2
    …upon further thought, I think any grinding sufficient to remove a meaningful amount of material would remove the case hardened layer and therefore destroy the cone.
    – MaplePanda
    Mar 19, 2022 at 3:32
  • FWIW if you want to take a really deep dive on this question, look up the chapter in Barnett's Manual about it. There's more than an SE question could ever cover Mar 25, 2022 at 16:38
  • @NathanKnutson Very precious pointer. Thank you!
    – Sam7919
    Mar 25, 2022 at 17:34

2 Answers 2


I don't think it matters that much how the bearing balls interface with the bearing race (the rolling track)

  1. In theory, the balls are hard and will form a "point contact" with the race. In reality its a very small contact point, so the curvature isn't that important.
  2. The balls are harder than the races, but a single ball does multiple rotations to get around the race once. In my experience, the balls shed their hard chrome outer and the race gets a polished line/channel over time. If the race starts pitting or spalling, its game-over for that part and continued riding will only accelerate the damage.
  3. The more important features of the cones/cups is how it interfaces with the threads and the rest of the hub and bike
    • Threads have to match
    • Ends have to not bottom out in places they're not intended to
    • The hardest thing to match is dust seals and how the outer edge of the cone nut interfaces with the hub, along with its length and whatever bushings/spacers are needed to match your OLD.

Unless you've got an expensive name-brand wheelset or hub, you're probably best off buying a whole replacement axle with cone nuts, along with new bearing balls and assembling it. There's nothing wrong with using spacers from the old axle, to get it perfect. You can reuse the old skewer assuming its not damaged or worn.

  • 1
    I think I understand. Thanks. Can you confirm that the description here is accurate?
    – Sam7919
    Mar 19, 2022 at 14:53
  • 1
    @Sam I don’t believe car wheel bearings are standard. Do take into account the fact that people’s tolerance for mechanical faults is larger for cars because one can always just put their foot down a little more. On a bike, we care more about every little source of damage and drag.
    – MaplePanda
    Mar 19, 2022 at 18:36
  • 2
    (1/3) I know you're a 94k and I'm a lowly 3k :-), but I think your first point is not right. If we agree that "nothing is rigid" (engineering.stackexchange.com/a/48869/36266), and that each bearing ball under any load, but especially a 70 kg load inadvertently hitting a tree root (bicycles.stackexchange.com/q/67038/48599), becomes a squishy ball, albeit a tough squishy ball, ...
    – Sam7919
    Mar 25, 2022 at 17:30
  • 1
    (2/3) ... then the difference between the smaller of the two curvatures on the cone and the curvature of the ball bearings is one important factor (among many) that determines whether a cup-and-cone is well engineered.
    – Sam7919
    Mar 25, 2022 at 17:30
  • 2
    In fewer words: you do want your ball bearings to hug, and not just to touch, your cones (and your cups).
    – Sam7919
    Mar 25, 2022 at 17:31

There is a lot to this question and it is a good question.

One of the things that make a complete answer difficult to give is that there have been a vast number of cones in the history of bikes, which is pertinent because one group who would be concerned about this question is people restoring old bikes. But, another aspect to the question whose importance can't be understated is that if you're only looking at, say, cones from Asian hubs of the last 20 years, a lot of the time they're going to be from the same 3 or 4 major producers (Joytech, Formula, KT, etc) whose products are sold under many brand names. The really big producers of commodity-level hubs have major impetus to re-use the same cone designs, and playing off of that fact is a path to coming up with a matching replacement in many instances.

The most basic principles of matching cones are:

  • Get the intended ball size right, which is usually 3/16" for a front and 1/4" for a rear
  • Get the contour mostly right, but close enough is acceptable. You test this by holding new against old and finding something where the profile comes close to matching, assemble the hub with grease, put the cone on, spin it, and see where the ball track is landing. You want it somewhere in the middle band of the ground part, analagous to where the stock one was. You don't want it flirting with running off the edges. Some bias off to one side is okay, but note that it affects spacing which you'll then have to go back and correct with spacers.
  • Get the factors that affect sealing and fitment of all the parts right. This means that for example, if the original uses a pressed metal seal, the new one has to be the right OD to transplant the seal on to it. Or if there are rubber seals, the new one has to interact with them correctly, allow them to be at the right spacing away from the locknut, etc. There are various ways that a cone that was too long or thick could screw this up.

Repairing cones by sanding or grinding them is a good approach in many cases. To do it, all you have to do is secure the cone and locknut on to a a straight axle in similar position to how they would be on the assembled hub, except give the protruding end of the axle enough length to secure in a drill chuck. Get a dowel or screwdriver shaft that's a little smaller diamer than your bearing, clamp it in a vise, wrap in sandpaper (or just hold up your sandpaper or other abrasive paper/cloth to it) and start grinding. Use successive grits and/or polish if you want to be fancy. This is a way of repairing moderately damaged cones only. Deep pitting or uneven wear (such as from a bent axle) can't necessarily be fixed this way. It's an analog skill although there are ways more precision could be introduced to it. In my experience, it appears that the hardening on most cones is done in a way that goes suitably beyond the surface that one can ignore concerns about getting through a hardened layer (as in case hardening). There may be exceptions to that out in the world, especially since case hardening used to be more common in different fields as I understand it - I've been curious for a while about how this applies to restoring old cones in this way but I've had a hard time finding information about it. By and large, just resurfacing what you've got is an excellent way to go in many cases.

Your Answer

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

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