I've always found it easiest to clamp the opposite end of the axle assembly (via the lock nut) lightly in a vice and then adjust the cone/locknut on the free end. Otherwise, it seems you need 3 hands (I'm usually one short it seems) to stop the axle from turning as you tighten the locknut.
It usually takes me 2-3 tries before I have it "to spec" regarding play vs binding.
Regarding torque spec, since the assembly is captured, there's little need to go nuts (pun intended) on a lock nut.
Gary Watts Vancouver, Washington USA
----------------------__________________________________________________ ______________________________________________________
I think the torque one can apply with a Campy two-ended cone wrench without injuring your hand is a representative upper limit. The function of the nut is to jam the threaded cone so it does not turn due to ball friction, and to ensure that it maintains the proper position when the QR skewer is locked. The bearing is set a skoche loose (no, I do not have an end-play spec to quantify "skoche"), in the hope that it pulls up to just touching when the skewer compresses the frame ends and hence the cone/axle subsystem. We're making the skewer (a stiff spring) compress the axle (a stiffer spring, 'cuz it has more cross-section).
I also think the multiple cone-wrench method is the best way to go, because cones don't always stay where they should when you tighten a lock-nut. Setting cone bearings can involve some art, or at least skill.
Ken Freeman
Ann Arbor, MI USA