Results of a photoperiod experiment show that human sleep can be unconsolidated and polyphasic, like the sleep of other animals. When normal individuals were transferred from a conventional 16-h photoperiod to an experimental 10-h photo-period, their sleep episodes expanded and usually divided into two symmetrical bouts, several hours in duration, with a 1-3 h waking interval between them. The durations of nocturnal melatonin secretion and of the nocturnal phase of rising sleepiness (measured in a constant routine protocol) also expanded, indicating that the timing of internal processes that control sleep and melatonin, such as circadian rhythms, had been modified by the change in photoperiod. Previous work suggests that the experimental results could be simulated with dual-oscillators, entrained separately to dawn and dusk, or with a two-process model, having a lowered threshold for sleep-onset during the scotoperiod.