Someone asked me this question and I was wondering how you guys would answer it...... Consider a container with some gas in it. It can be a sealed living room filled with ordinary air at typical pressure. The air molecules are in motion - colliding with each other and bouncing off the walls. Question: If this random motion of molecules around the room is allowed to proceed long enough (say, a billion years raised to the billionth power), will it at some point happen that all of the gas molecules are in one half of the room, leaving the other half in zero pressure? Is that situation forbidden by the principles of gas dynamnics, or are gas dynamics just statements about what is statistically likely to be the case at any one time as gas molecules knock each other around? Will those principles be wildly violated on statistically rare occasions?