This could be bunk, but it might give some of the magnet people here some insight for things to try. A couple of quotes:
"Pulling two magnets apart along the same path they took to pull themselves together will of course require as much (or more) energy as the amount generated by the magnets when they come together. However, it has been discovered recently that two permanent magnets of a particular shape can be pulled apart along a prescribed path that requires less work compared to the amount of work produced when the magnets come together along a different path."
and,
"The paths that must be followed came as a surprise and were not intuitive. For example, it has been discovered that two cube-shaped neodymium magnets (measuring .75ââ,¬Â, 43 pound pull-force, Grade N38, 1.8 ounces) are capable of generating 7.46 inch-pounds (work) when they pull themselves together ââ,¬Å"sidewaysââ,¬Â in the horizontal plane. It takes only 6.56 inch-pounds to pull the magnets apart along a vertical path that is perpendicular to the path they followed when they came together. This leaves a .90 inch-pound net-yield of mechanical energy (work) which can be used for example to turn an electric generator."
tak
http://www.kedroncorp.com/abstract.html (http://www.kedroncorp.com/abstract.html)
message deleted.