The authors describe a technique they developed for quantitatively evaluating experiments in which a subject manipulates a spline curve to match a curve supplied by the experimenter. The evaluation takes the form of a goodness-of-fit parameter and a measure of the time taken for the subject to create the new curve. It uses as a worked example an experiment where five types of spline are used--B-spline, Bézier, Catmull-Rom, and natural splines with natural and tangent end conditions. It comes as no surprise to those in the field that B-splines are the easiest, and natural splines the hardest, to manipulate in this sort of work. The interest of this paper lies in the methodology used, however, rather than in the results of the example experiment.