How do metal balls induce stability in a tetrahedral magnet arrangement?

[gillsimo]

Hi...take six regular N/S rod magnets and arrange them as a tetrahedron.
It's fiddly to achieve...because it can't actually be done in terms of magnetic attraction/repulsion.
Each of the four vertices have 3 poles meeting & in each case the situation in one whereby one pole is attracted to the other two but the other two are not attracted to each other.
What intrigues is that one can add a plain metal ball to each of the vertices & each immediately becomes stable, all three poles now attracted to each other.
Might somebody please explain to me what is happening in terms of the magnetism induced into the balls to make this happen?
Many thanks

 

[DaveC426913]

I don't know for sure but:

notice how the bearings provide separation between the poles in the magnetic bars.
The +-+ poles aren't in direct contact as they would be without the bars.
This allows the bending of the magnetic lines of force to sort themselves out over a slightly longer distance through the body of the bearing.

Lines of force being far apart and smooth-of-curvature is an indication of more gentle forces.

 

[Vanadium 50]

One way to think about it: Say it's NNS. The N-S field lines flow through the ball. The "leftover" N now sees steel which it attracts like a magnet usually does to steel. They attract.

Another way: In your tetrahedron the magnets are not as close as they would be without the balls.