Well moving fairly far afield from the original topic I thought I might comment a few points to this whole diffusion of gases discussion.
CO2 and air will mix in a process called Brownian motion (technically I think Brownian motion refers to the molecules eratic course if you look at it closely and difffusion is the mixing process, but thats a bit pedantic). However it takes a little while (on the order of minutes) depending how how warm the gases are. I think most people have seen dry ice melt and the layer of vapour formed, but not thought about how long that layer will hang around.
Of course light gases will separate from heavy gases (it happens all the time in the atmosphere as helium escapes the earth and wanders off into space) but to cause CO2 and O2 to separate you've either got to have them very cold so that Brownian motion doesn't mix them back up again faster than they settle, or in a very strong gravity field (i.e. a Centrifuge). (Actually I'm not sure it would work even then because then the molecules no longer act like a gas and start to turn into a liquid which is how we usually separate them, by condensing them into a liquid out of the air)
I also wanted to comment about sathids speculation that CO2 doesn't form a protective blanket on your fermenting beer. I'm not sure if you were talking about a fermenter with the lid off (in which case you would be correct) but a fermenter with the lid on with either an airlock of just a small hole like the clingwrap method will certainly slow down the diffusion process to the point where it is insignificant compared to the amount of new CO2 created. Just remember to get it sealed up once its finished fermenting. Likewise I'd have to say that purging your bottles with CO2 works inspite of diffusion processes because the time between purge and fill and cap is so short (not that I do this, never really had a problem with the flavour of any of my bottles so I won't spend CO2 on it ;-)
--
Mesa.