In studying apparently periodic phenomena modeled in the laboratory or simulated on the computer, we often find, upon close examination, that the periodicity is noisy. Figure la is a laboratory example. It is a trace of the temperature of a fluid at a fixed location in a rotating, differentially heated vessel recorded by Hide et al. during a 20-minute interval as a chain of waves passed by. 1 Since no fluid experiment can be perfectly controlled, one might assume that the failure of each peak in the curve to duplicate the third peak preceding it represents the experimental uncertainty. The authors have established, however, that the differences between the peaks are a real feature of the process they are investigating.