Lacking sufficient reputation on this site to comment on posts, I'm going to make this an answer, and you know it's a true urban legend if you read the posts above. Once, at a Princeton physics exam, a group of the senior Princeton physics faculty were trying to figure out why, when you shake a bunch of rods in a container with certain asymmetries in the geometry of the container, the rods assume a "more ordered" state (they tended to concentrate on one side), and they couldn't figure out why this did not contradict the second law of thermodynamics! In an amazing twist, they speculated that the result had to do with finite size effects.....
(for those who aren't in on why this is just too crazy to believe: the shaken container is not a closed system, so the second law doesn't apply. Further, the forces on the rods, a combination of shaking and frictional forces, do not correspond to thermal noise and dissipation, so there is no reason for the system to go to thermal equilibrium. It's like asking why, when there is a baited mousetrap and a live mouse in the room at time t=0, is it the case that after a certain amount of time the entropy decreases in that the mouse is more likely to be in the mousetrap than not. I think the most clever answer for the student is: "I notice that this exam has gone on for 30 minutes already and you are still walking and talking. Why are you not relaxing to thermal equilibrium? Perhaps the food you ate this morning is helping keep you out of equilibrium?")
:-)
