I cannot believe that there are those who do not notice the rubber seal in the bottom plastic bit and that it does not cover the antenna area......
In other words: If you take a moment to investigate then you'll see that the air is being sucked in through the metal grill around the antenna through the thin gap around the whole of the bottom plate which is cleverly held in place by three standoffs. Duhhhh
So around the rim of the bottom plate? The part of the unit that is roughly 2 CM off the table already? That part? How much air do you think it sucks in that it needs to be lifted another foot off the desk?
So let's put it to the test (hence the late reply). I ran the new 2011 Mini elevated 6" above my desk using a round PVC tube (3" in diameter) for 4 hours. Idle temps were 45-47 C. Standard use (surfing the net, reading emails, playing iTunes, etc.) saw it jump between 47-54 C (while the jumps in temp seemed high and abrupt, they quickly descended just as fast as they ascended). High load saw it rest at 71-74 C (20 minute encode of a TV show using Handbrake which saw all 4 threads 100%). Ambient air stayed constant at 38-39 C for the length of the test.
And then we have the 4 hours of standard use to compare (which I just now finished). The only difference was the removal of the PVC tube which saw the Mini sitting right on the desk. Drum roll
The freakin' same! I reiterate: Temps were identical in all test cases (okay except for the light use, which varied in temperature, but retained the same high and low values throughout), whether the Mini was elevated or not.
Temperatures were collected using Marcel Bresink's Temperature Monitor (4.94) and confirmed to be the same as iStat Nano's readings before data was collected. The temperature above is collected from the CPU diode sensor only (and the ambient air temperatures were checked for consistency), which is what I'm assuming people are arguing will see a drop.
So there you go. And that's why you should always collect data rather than convincing yourself something has an effect...