exp(entropy/2) or exp(entropy/?)

I previously stated that exp(entropy/2) seems to remain the same shape no
matter what value of N is used. This was based on curves where s=1%. Upon
closer inspection, it seems like exp(entropy/2.3) is a better formula. I'll
announce a couple of graphs showing this. Could this be because 2.3 ~=
log(10), and 10=sqrt(1/1%)? I have no idea. Anyone have any heuristic
arguments? Anyway, I guess I should try some with s=0.6% . . . .