A visual differential threshold was recorded continuously for 14 min. in each of six different sessions. Integrated scores for 1/2‐min. periods, indicative of mean threshold values, showed a marked tendency to rise between the second and the tenth minute of the experiment. This was attributed to a fatigue decrement of a central state of vigilance or alertness. The scoring of length of curve and number of control adjustments during 1/2 min. periods, indicative of the subject's working behaviour, showed no such consistent tendencies. In the first run, however, the method of working showed some training effects which seemed to mask the fatigue decrement of mean threshold values and led to a change in the criteria of judgement for the second run. All three variables showed marked and fairly consistent individual differences. No clear differences between morning and evening values could be established.