Subjects compared the pitches of two tones which were separated by a retention interval during which six extra tones were interpolated. The effects were studied of varying the sizes of the melodic intervals formed by the successive tones of the interpolated sequence. It was found that error rates were lower when the interpolated sequences were composed of smaller melodic intervals; and it was argued that such sequences formed a more effective framework of pitch relationships to which the test tones could be anchored.