Saccharin and Bladder Cancer-Reply

It is unlikely that many readers of The Journal "have been misled by the last sentence" of our abstract, if only because quantitative risk estimates are presented in the text of our article. In any event, Professor Gryder also seems to acknowledge that "quantitative estimates of the magnitude of an increased risk are, of course, for ideal situations that can be approximated only in a real experiment." The concerned physician or lay person may well ponder which is the more relevant: idealized quantitative estimates of cancer risk or etiologic inferences based on biologic reasoning—a knowledge of the cancer process and statistical analysis?

We agree with Gryder that "the animal-feeding experiments cannot be ignored." However, the three cited animal studies pose serious inferential problems to the cancer experimentalist.¹ For example, the Canadian rat study was designed to assess the carcinogenicity of orthotoluene-sulfonamide rather than saccharin; accordingly, only one