The task of setting punitive damages most closely resembles a particular psychophysical method, which we have labeled magnitude scaling without a modulus. As was noted earlier, subjects in most applications of magnitude scaling are instructed to assign a particular number (the modulus) to a specified standard stimulus, and to judge other stimuli in terms of ratios of subjective intensity relative to that standard. The modulus is both arbitrary and inconsequential: the same ratio scales will be obtained regardless of whether the standard stimulus is to be judged 10 or 100. However, magnitude scaling can also be conducted without a modulus, by instructing subjects to ensure that their judgments of different stimuli should reflect the ratios of the intensity of subjective experience. In the context of psychophysical scaling (e.g., of loudness), the procedure yields scales that are quite similar to those that are obtained in the standard method, though somewhat noisier and more skewed. It appears that a subject in such an experiment spontaneously adopts a modulus—and different subjects adopt different moduli. Naturally, this procedure greatly increases the variability and skewness of the numerical responses to any given stimulus. The effects of individual differences in the size of the modulus are easily eliminated in the statistical analysis of psychophysical data, where each observer judges many stimuli. Under current legal practice, this remedy is not available, since each jury considers only one case in isolation, and indeed is explicitly prohibited from comparing it to others.