The assumption of polyphasic sleep/wake regulation is based on the occurrence of nap-sleep at specific phase positions in the circadian cycle. Further support would be the split of the normal long major sleep episode into shorter components. Evidence for this hypothesis comes from the discovery of bimodal distribution in sleep duration. An experimental approach to test this hypothesis has been carried out by restricting sleep duration in free-running rhythms. The outcome was a biphasic distribution of sleep within a circadian cycle with sections of dissociation and synchronization of the two sleep blocks. The results show similarities with 'splitting', a phenomenon which has been found in animal studies. The relatively short duration of the different sections as well as the asymmetric distribution of the two sleep blocks in the circadian cycle leads to the assumption of a splitting of the major sleep episode and not of the circadian rhythm. Sleep split into two, relatively short sleep episodes of comparable duration contrasts with napping, which is characterized by an extra sleep episode in addition to the long major sleep.