THE PURPOSE OF THREE EXPERIMENTS WAS TO DEMONSTRATE HOW PRACTICAL AND THEORETICAL PROBLEMS OF FACE RECOGNITION ARE LINKED. IN THE FIRST EXPERIMENT, MEMORY FOR PHOTOGRAPHS OF REAL FACES WAS COMPARED WITH MEMORY FOR COMPOSITE FACES CONSTRUCTED WITH THE PHOTOFIT SYSTEMS (PHOTOFIT FACES HAVE LINES FORMED AT THE BOUNDARIES OF THE FIVE COMPONENT FEATURES). RESULTS INDICATED THAT THERE IS AN ADVANTAGE IN MEMORY OF PHOTOGRAPHS OF REAL FACES OVER MEMORY FOR ARTIFICIAL FACES PRODUCED BY PHOTOFIT. THE SSECOND EXPERIMENT WAS DESIGNED TO TEST THE HYPOTHESIS THAT THIS DIFFERENCE IS DUE TO THE PRESENCE OF LINES IN PHOTOFIT FACES THAT INTERFERE WITH RECOGNITION, AND RESULTS CONFIRMED THAT THE PRESENCE OF LINES IMPAIRS MEMORY OF FACES. THE THIRD EXPERIMENT SOUGHT TO DETERMINE WHETHER THIS IMPAIRMENT IN FACE MEMORY OCCURRED ONLY WHEN THE LINES BROKE UP THE FACE INTO IDENTIFIABLE FEATURES, OR WHETHER RANDOMLY PLACED LINES HAD THE SAME DISRUPTING EFFECT. RESULTS INDICATED THAT THE RANDOM LINES PRODUCED THE SAME DELETERIOUS EFFECT ON FACE MEMORY AS THE SYSTEMATIC PHOTOFIT-LINES. THE PRACTICAL IMPLICATION OF THESE CONCLUSIONS IS THAT FORENSIC DEVICES FOR DEPICTING FACES SHOULD AVOID THE PRESENCE OF LINES OR OBVIOUS BOUNDARIES BETWEEN FACE COMPONENTS AND THAT THE IDENTIKIT SYSTEM (IN WHICH TRANSPARENT OVERLAYS OF SKETCHED FEATURES ARE USED TO BUILD UP A FACE) IS BETTER THAN THE PHOTOFIT SYSTEM. TABULAR DATA AND A LIST OF REFERENCES ACCOMPANY THE ARTICLE. (MJW)