With a variable that is recorded on a scale with fixed bounds, it can be appropriate to use for the denominator of the coefficient of variation the square root of the (sign-independent) product of the differences between the mean and the two bounds of the scale. A simple illustrative example is given.