Abstract
Experiments were performed upon groups of three or four human subjects in an isolation chamber (total n = 14). Subjects lived initially on a conventional lifestyle and then delayed their hours of sleep by 8 hr (so mimicking some aspects of nightwork) for 2 or 5 days. They also performed two constant routines—protocols designed to minimize any effects due to the environment, mealtimes, and activity. Regular samples of urine were taken when subjects were awake, and were analyzed for sodium, potassium, and chloride; rectal temper ature was measured and logged at 6-min intervals throughout. Shifts in circadian rhythms produced by the change in sleep time were assessed by cosinor and cross-correlation tech niques. The protocol enabled these assessments to be made on days when sleep was allowed and under constant-routine conditions, so that masking and behavioral effects could be in vestigated also. The results confirmed that adjustment to the change in sleep time was slow and only partial, and that assessments made on days when sleep was allowed overestimated this adjustment. Furthermore, it was concluded that, whereas cosinor and cross-correlation techniques using only one shifting component were equally useful in describing the observed changes, both were inferior to a cross-correlation technique that made use of two shifting components. Some practical and theoretical implications of these results are discussed.