Rectangles may appear to reverse like trapezia when they rotate at an uneven rate.

A 1983-1985 theory by Mitchell and Power predicts that, when rotating rectangles undergo certain kinds of speed fluctuation, they should appear to reverse just as trapezia do. The prediction is partially confirmed. One of two 'mimic' rectangles underwent apparent reversals more often than a control rectangle undergoing even rotation and in the same places as rotating trapezia. However, its reversal frequency was less than those of the trapezia, and a second 'mimic' slowed an inappropriate distribution of reversals round the cycle. These anomalies call for some modification to Mitchell and Power's theory, but minor qualifications may be sufficient.