widderjoos said:
Thanks, but how do they measure the velocity distribution without significantly heating the system up?
I'm more familiar with measurements in the μK range that were common 20-or-so years ago. I'm not 100% sure if they still work in the nK range -- I believe they do, or some variation of them -- but here is a brief explanation:
After cooling and trapping atoms the optical trap is shut off, releasing the atoms that had been trapped. The collection of atoms then expands, owing to the different velocities of the atoms. Loosely speaking, the amount of expansion of this "cloud" of atoms is measured some time later. The expansion rate is a measure of the velocity distribution, from which the temperature can be inferred.
For more details, Bill Phillips has nicely described several methods, all relying on the expansion of atoms after shutting off the trap:
http://prl.aps.org/files/RevModPhys.70.721.pdf
One method described on p. 730 (p. 10 of the pdf file), in the paragraph that begins "Using the techniques for chirp cooling, ..."
A second method is described starting at the bottom of p. 731, in the paragraph that begins "In this time of flight (TOF) method,..."
Two more methods are described briefly, on p. 732, in the paragraph that begins "Another method was the 'fountain' technique..."
Hope that helps.