Hi Shan, here's a somewhat different way to reach the same conclusion as Scott.

scan during data collection? As I see Nidp = (2/pi)*(deltak * deltaR), there is no reference to the scan step.

But there is, an implicit one. Your deltaR is limited by (among other things) the scan step. If you halve the scan step, you can in principle look twice as far in R space. The problem is, because of the nature of the EXAFS equation, there isn't any signal beyond (say) R=15 AA that's not completely drowned by noise. So there is a point beyond which it's not useful to sample more finely in k space. Of course, below that point, the step size in k space does matter. Silvio