We describe an experiment in which the eye movements of participants, carrying out tasks using two contrasting graph designs, were recorded by means of a remote eye tracking device. A variety of eye movement properties were measured and analysed both temporally and spatially. Both graph designs were based on specific psychological theories and established graph design guidelines. One incorporated attributes thought likely to enhance usability, the other included attributes likely to have the opposite effect. The results demonstrate that the design and location of a graph's legend and its spatial relationship to the data area are extremely important in determining a graph's usability. The incorporation of these and other design features may promote or detract from perceptual proximity and therefore influence a display's usability. The paper demonstrates that this influence is reflected in eye movement patterns, which can be readily monitored by means of a remote eye tracking system, and that a relatively simple temporal analysis of the results can give important insights as to how the usability of visual displays has been influenced.