Key points are not available for this paper at this time.
Subjects searched a rapid sequence of computer-produced letter arrays for the presence of a numeral in one of the arrays. The subjects' scanning rates were computed from their precentage of correct detections of the location of the numeral. Scanning rates were very high and approximately the same for a wide variety of conditions; the highest scanning rates (125 and 75 letters per second for two subjects) occurred when there were 9 or 16 letters in each of the arrays and when new arrays were presented every 40 to 50 milliseconds. Giving the subject advance knowledge of the numeral to be presented made little difference in the scores.
Sperling et al. (1971) studied this question.