Here I demonstrate an extremely interesting phenomenon. A magnet is suspended by Meissner Effect levitation over a toroid of high-temperature superconducting ceramic cooled in liquid Nitrogen. The magnet spontaneously begins to rock back and forth on its magnetic axis. The magnitude of the rocking gradually increases, until the magnet rocks all the way over and continues to spin in one direction, indefinitely, without the addition of any external power.
There have only been a few peer-reviewed papers in the literature concerning this effect. The conventional explanation is that it is due to a thermal gradient across the magnet and this gradient affects the magnetic field strength. I question this explanation, as I have performed this experiment with magnets encased in copper foil which should have evened out such a gradient--these magnets rotated just as well as unencased ones. I don't, however, have a viable alternative explanation.
The magnets shown are NdBFe and Samarium Cobalt; the superconducting donut is Y-B-C-O, or yttrium-barium-copper-oxide, high temperature superconducting ceramic, melt-textured and sintered, according to the formula of Evgeny Podkletnov. The liquid is Liquid Nitrogen, and the container is styrofoam.
References:
Ma, Liu, McMichael, Bruce, Mims, and Chu (1991). Spontaneous and persistent rotation of cylindrical magnets levitated by Y-Ba-Cu-O superconductors. J. App. Phys., v. 70, pp.3961-3963.
Hirsch and Hirsch (2003). Spontaneous spinning of a magnet levitated over a superconductor. arXiv:cond_mat/0303574v1 27 Mar 2003 (less info)