The subjective relation ``octave of'' is demonstrated to be a valid interval for determining scales of musical pitch for pure tones presented successively. Octave judgments of trained musicians have a standard deviation averaging about 0.6 percent. Inter-observer variability is 2 to 5 times as great, increasing with frequency. Judgments vary significantly from day to day, but as the direction of shift at different frequencies for a single observer is random, the shifts cannot be attributed to changes in the octave criterion. Instead, this variability, and also (1) differences between right- and left-ear judgments of a given observer, (2) the change in diggerence between subjective and physical octaves as a function of frequency, and (3) the high inter-observer variability, all confirm the bais instability of pitch-frequency relations implied by the facts of binaural diplacusis. Individual and group scales of musical pitch are deduced. In these scales, the average rate of change of musical pitch with respect to frequency level is less than unity by a small but significant amount. Although this discrepancy is not explained, tests show that it is not an obvious artifact of method. Simiultaneous presentation raises variability, but affects means only slightly. The relation between the peculiarities of individual scales and binaural diplacusis is discussed.