Oops - I did misread what he wrote. He was discussing the distribution of the min of the sample, not the distribution itself. Also, there may be a bias in the

Oops - I did misread what he wrote. He was discussing the distribution of the min of the sample, not the distribution itself. Also, there may be a bias in the true p-value, but there is none in the p-value bound based on h(d) and min(N1, N2).