Measure the torsional stiffness by torque versus deflection. Then measure the polar moment (correct term for rotating inertia?) by hanging from a soft spring, twist and release, and measure oscillation period. Combine the two and you should be able to calculate natural frequencies.
I think so anyway.
