Abstract
Metaphase analysis, the micronucleus test, and the dominant lethal assay were performed in rodents with cyclophosphamide. The variance in the results indicated that use of binomial statistics (or the Poisson or normal approximations of this distribution) would be inappropriate for determination of significance. This conclusion was reinforced by finding that negative binomial distributions best explained certain aspects of the data, as well as being theoretically more likely. Because there was no simple significance test based on negative binomial statistics and there was some doubt about the distributional form, a distribution‐free significance test seemed most appropriate. Thus a one‐sided Kolmogorov‐Smirnov two‐sample test was used. Although both the dominant lethal assay and metaphase analysis proved superior to the micronucleus test in terms of degree of response, this advantage was more than offset by the lesser difficulties and greater number of cells that could be analyzed in the micronucleus assay.