[donovanr]

Right, but it's the 'every once in a while" part that I wonder about, i.e. the variance of that distribution. Maybe you could argue that since the Maxwell-Boltzmann distribution is narrower the lower its mean, if kT is small then it's an exponentially tiny effect.

[tiplus]

So yes, kT is small on average (does fluctuate) and it is also 1 kT per degree of freedom. So yes, in principle a few hundreds of atoms could randomly fluctuate in energy to create this effect. However, this is hyper unlikely. I would be suprisd if it happened once in the life time of the universe. Also consider that you cannot create Maxwell demons[1].

[1] https://en.m.wikipedia.org/wiki/Maxwell%27s_demon