Yap, right out of the vacuum. The reasoning is flactuations of the vacuum, justified by the uncertainty principle. I thought of it during the boring lecture I'm in (+-). There's a given flux of pair generation depending (if I'm not wrong) on the local energy density of the vacuum, but the probability to have 10 pairs generated simultaneously out of the blue (or black in this case) should be 1/10 the probability to have one if they're uncorrelated, and I guess they are strongly correlated - as 10 pairs are ten times more energetic than one - thus could survive for only ~tenth the time. So, it's unlikely to have elephants and anti-elephants generating from the vacuum, and they should last ~10^-24 times shorter than a single pair, i.e. ~10^-55 seconds if I remember right. If you can spot a virtual anti-elephant, you're wasting your time here. Actually, you're wasting your time here anyway, but that's another thing.

(Back to my boring lecture, alas)

I think it's a fundamental part of it, specifically arising from fourier transforms to momenta-space and back. Each canonical pair of variables has its typical uncertainty relation. But the graduate physicists on the herd could probably give you a better answer.

<zohar atai <"_z-atai@zahav.net.il"
<Alexander Mann <"_alex.mann@netcologne.de"