1. I'd mention how non-parallel drops will bow a hub axle & 'cone' out the bearing seat(s). I went through a couple hubs before I realized this. -WeedBMX
2. Some bearings just simply have play in them too. I've cured some wobbly hubs with new, higher quality bearings. I've also cured some hubs that are way too tight with once wobbly bearings. Although very very tight, bearing manufacturers also have a tolerance to work with. - Couch of Zodiac Engineering
3. It is very important to distinguish between bearing arrangements that may LOOK similar, but behave in different ways.
With a crank:-
The load acts in a relatively constant direction relative to the OUTER race, so the INNER race "sees" a load which acts radially but at a varying direction.
ie. if you mark a white dot on the inner race of the BB bearing, then when the dot is at the bottom, the load acts to push the axle INTO the dot.
But when the dot is at the top the load acts to pull the axle away from the dot.
So it is important that a crank axle is a fairly close fit inside the bearing to prevent the axle "rolling around" inside the race and causing wear.
With a hub:-
The load acts in a relatively constant direction relative to the INNER race, so the OUTER race "sees" a load which acts radially but at a varying direction.
So with a hub, it is more important that the OUTER race is a good fit in the hub shell and the inner race can be a looser fit on the axle (which helps allow the bearings to align with each other).
There are a number of standard bearing "arrangements" based upon whether load is radially oscillating or constant; and that "fully retain" different races. For the record it is "incorrect" to fully retain all the races. Bearing mountings should allow for differential thermal expansion of different components. - G of G-sport
