It is suggested that there are two families of ratio scales, the one obtained by marking off equal intervals and the other, by marking off equal ratios. Data are presented which suggest that the latter procedures seem to be operating in direct psychophysical scaling and in observational measures where frequency counts are made of behavior. The data suggest that both families of scales function comparably in mathematical experiments, although the parameters in the resulting equations differ in predictable ways. It is also suggested that ratio measurement is always relative to the origin of the relationship under investigation. This point is illustrated via experiments on the utility of wages.