“…In adherence to the stipulations of USP 1223 (3), the statistical processing of the data made proper use of the underlying probability distribution of the raw data, i.e., the Poisson distribution, instead of approximations to the normal distribution. In particular, two features of Poisson-distributed data were exploited in the analysis (27,29,33). Given count data, x and y, from two Poisson populations with expectation values (means) and , i.e., XϳPoisson() and YϳPoisson(), then the sum, s ϭ x ϩ y, is again a Poisson variable, SϳPoisson(), with mean ϭ ϩ .…”