Abstract
In this article we describe a procedure for representing multivariate time series data by means of interactive, computer-generated dynamic imagery with computer-music accompaniment. This innovation conveys the novel insights that dynamic imagery can provide; yet, the imagery is developed from principles that make the representation useful when examined either statically or dynamically. This is because the development of the dynamic representation is guided by the same perceptual and technical principles used in making a motion picture. The particular implementation we describe is evaluated by a formal psychophysics experiment in which we measure the threshold correlation that can be perceived in our dynamic representation, and in each of three different types of graphical portrayals.