Tautology or not tautology?

It has been suggested that the good correlations found between carcinogenic potency in mice and in rats could be an artifact. The artifact suggested arises because there are four correlations to consider--interspecies correlations of toxicity, interspecies correlations of carcinogenic potency, and two intraspecies correlations between toxicity and carcinogenic potency--and the existence of any three implies the fourth. It was argued that the intraspecies correlations between toxicity and potency were due to criteria for data selection. Here we discuss the correlations in detail and show that they are principally due to the experimental observation that there are few cases where most or all animals in a bioassay get cancer. We conclude that the correlations between carcinogenic potency are valid.