The timing of bedtime and waketime decisions in free-running subjects.

Twelve "free-running" temporal isolation studies were performed using healthy human subjects aged 20-81 years (4 males, 8 females). Circadian rhythms were measured in subjective alertness (using a visual analogue scale technique) and rectal temperature. In all 12 subjects best fitting rhythm period were found to be shorter for rectal temperature (mean = 24.3 hours) than for subjective alertness (mean = 24.8 hours). Both rhythms were predictive of bedtime and waketime decisions. Bedtime decisions tended to cluster on the falling arm of the temperature and alertness cycles (modal times = 120 degrees (temperature) and 140 degrees (alertness) after acrophase); waketime decisions on the rising arm (modal times = 240 degrees (temperature) and 300 degrees (alertness) after acrophase). When this pattern was violated, sleep episodes were significantly more variable in length and disturbed by intruding wakefulness.