dennis@home wrote:

The gap makes them swing straight, if you just have one the balls will
just bounce to one side and not impart all their energy to the next ball.

The motion wants to be:
(a) In-plane, and
(b) Irrotational.

Even with separated attachment points, flexible lines will permit a degree of
out-of-plane movement, which tends to grow. Probably the most satisfactory
compromise between in-plane constraint and low wind resistance would be a solid
strip (or two, to obviate out-of-plane deflection) of a rigid material, aspect
ratio say 5:1, with the long dimension in the plane of motion. If this was
rigidly attached to the ball, and hinged at the top, the various requirements
would be met quite well. A bearing or two (one each side of the strip) would be
better than a hinge, but might be overkill.