AN EXPERIMENTAL SYSTEM WAS DEVELOPED FOR RECORDING EYE-MOVEMENT DATA. RAW DATA WERE IN THE FORM OF MOTION PICTURES TAKEN OF THE MONITOR OF A CLOSED LOOP TELEVISION SYSTEM. A TELEVISION CAMERA WAS MOUNTED ON THE SUBJECTS' FIELD OF VIEW. THE EYE MARKER APPEARED AS A SMALL SPOT OF LIGHT AND INDICATED THE POINT IN THE VISUAL FIELD AT WHICH THE SUBJECT HAD DIRECTED HIS GAZE. EYE-MOVEMENT DATA WERE RECORDED FOR A SAMPLE OF 43 FIFTH-GRADERS AS THEY VIEWED A TYPICAL EDUCATIONAL TELEVISION PRESENTATION. THE DATA WERE ANALYZED IN TERMS OF EYE-MOVEMENT INDEXES AND GENERAL DESCRIPTIONS. A SLOW SPEED PROJECTOR AND A FRAME-BY-FRAME FILM VIEWER WERE USED IN THE ANALYSIS. THERE WAS A MARKED CLUSTERING OF THE EYE MARKER ON CERTAIN AREAS OF THE VISUAL FIELD. THE MOST DEFINITE TENDENCY WAS FOR SUBJECTS TO FOCUS ON THE NARRATOR'S FACE. THE EYE MARKER OFTEN CHANGED IN FOCUS. EYE MOVEMENTS DID NOT FALL INTO THE TWO CLASSIC TYPES--FIXATIONS AND SACCADES. RATHER, A CONTINUUM OF MOVEMENT WAS OBSERVED. EYE MOVEMENTS TENDED TO BE LIMITED TO A SMALL AREA WHEN THE SCENE BECAME DYNAMICALLY COMPLEX. PATTERN DIFFERENCES RELATING TO INTELLIGENCE WERE IDENTIFIED. IMPLICATIONS AND SUGGESTIONS FOR FURTHER RESEARCH ARE MADE. (JC)