After choosing a dependent variable and deciding how to measure it, the researcher is still faced with the serious problem of controlling extraneous (to him) events that may affect the "pure" response of the system under study. In many experiments, we hold all the factors we can think of constant and hope (some pray) that those unknown or forgotten will exert only minor influences. Those who deal in human research unashamedly act as if they can perform these maneuvers even when the subject may act in any number of capricious ways. Clinical research, and particularly that in which the measure is subjective, has been compared to a chess game in which all pieces are autonomous and the pawn might revolt at any time if not closely watched.1 In this report we relate our experiences in evaluating a hypnotic drug in a geriatric population. Probably we would not have dared