In an experiment to determine the most efficient design for the commentary of an instructional film, special consideration was given to three variables concerned with the construction of commentaries: the level of verbalization (the amount of talk), the personal reference of the narrator, and the phase relationship between the commentary and the film (sound leading the picture, or sound lagging behind the picture). Naval trainees viewed several different motion pictures, each teaching basic information about knot-tying, but each varying from others on one of the three variables. After the instruction the trainees were tested on the perceptual-motor tasks, namely the tying of three difficult knots. The experiment showed that a moderate amount of verbal description of the task aided learning while too much talk hindered it. Statements using the imperative mood or the second person active were more effective (in this military group) than those using the third person passive. Finally, a "leading" commentary helped trainees more than a "lagging" commentary. (BB)