Sleep, 1(2):191197 1978 Raven Press, New York The FortyEight Hour Day Wilse B. Webb Department of Psychology, University of Florida, Gainesville, Florida Summary: Four normal young adult male subjects were evaluated in a systematically imposed regime of 32 hr of wakefulness and 16 hr of sleep time in an environment free from time cues, Electroencephalographic and electrooculographic recordings were made continuously during the experiment, which lasted for 10 complete cycles, Sleep efficiency was assessed by determining the percentage of sleep time during the assigned sleep period, The average sleep efficiency for the experimental period was 77%, Results, in general, conformed to earlier findings of non24 hr schedules of sleep and waking: the overall sleep system remains relatively stable across a variety of scheduled variations; however, utilization of the sleep period becomes less efficient as the schedule increasingly deviates from the normal approximately 16 hr wakefulness/8 hr sleep schedule, Key Words: Circadian rhythmssleep schedulessleep stagessleep efficiency, Previous studies have measured sleep in a range of sleepwake schedules other than 24 hr. In shorter schedules, Carskadon and Dement (1975) imposed a 90 min schedule (30 min sleep time60 min awake); Weitzman et al. (1974) reported on a 3 hr schedule (1 hr sleep time2 hr awake); Moses et al. (1975) used a 3Y2 hr schedule (1 hr sleep time2y2 hr awake). In longer schedules, Jouvet et al. (1974) imposed a48 hr schedule on one subject (14 hr sleep time34 hr awake). This paper reports on a 48 hr schedule (16 hr sleep time32 hr awake). This is an extension ofa series of schedules reported previously (Webb and Agnew, 1975; 1977); 9 hr (3 hr sleep time6 hr awake), 12 hr (4 hr sleep time8 hr awake), 18 hr (6 hr sleep time12 hr awake), 30 hr (10 hr sleep time20 hr awake), and 36 hr (12 hr sleep time24 hr awake). METHOD Four male subjects aged 18 to 29 years old (mean = 21.1) were selected to serve in the experiment as paid volunteers. Applicants were screened with the University of Florida Sleep Inventory to assure that the individuals selected normally slept 7 to 8 hr, were free from sleep disorders, and were free from the habitual use of drugs including alcohol. The subjects were informed that they would be in an Downloaded from https://academic.oup.com/sleep/articleabstract/1/2/191/2749546 by guest on 05 September 2018 Accepted for publication November 1978. Address reprint requests to Dr. Webb, Department of Psychology, University of Florida, Gainesville, Florida 32611. 191
192 W. B. WEBB experiment which altered the length of the normal sleepwakefulness cycle, but they were not informed of the exact regime on which they were placed. Immediately preceding the beginning of each subject's experimental run, four consecutive nights of sleep were recorded between 11 p.m. and 7 a.m. The regime of these subjects was a waking period of 32 hr and a sleep period of 16 hr. The sleep periods began at 12 noon and extended to 4 a. m. each "day." The first experimental sleep period began 29 hr after the last baseline sleep period. There were 10 complete cycles. The subjects lived separately in rooms especially constructed to eliminate all cues as to time of day. Fifteen minutes before the beginning of a scheduled sleep period, the subjects were sent a typewritten note instructing them to get into bed and attempt to go to sleep. The lights were off during the sleep period whether or not the subject was asleep. When the lights were turned on at the end of a sleep period, the subjects were instructed to get out of bed immediately and remain awake until the next sleep period. If a subject attempted to nap during the wakefulness period a loud bell was sounded until he roused himself. During the wakefulness periods the subjects were free to occupy their time as they pleased, except for napping. Typically they listened to recorded music, read books, played musical instruments, and worked on hobbies. Continuous electroencephalographic (EEG) and electrooculographic (EOG) recordings were obtained and were scored according to procedures detailed in Agnew and Webb (1972). While awake, subjects were continuously monitored by both EEG recording and TV. EEG electrodes were refixed when required by evidence of poor recording properties. RESULTS The basic results of this experiment are displayed in Figs. 14. Each row designated a 24 hr period extending from 12 midnight to 12 midnight. The solid line indicates sleep. Awakening intervals of more than 10 min may be seen as breaks in these lines. The exception in these figures is seen in subject A (SA). His pattern of arousal within sleep was characterized by brief awakening (stage 0) which did not reach a criteria of 10 min. These periods are indicated by hatched marks. The efficiency of sleep in schedules of the earlier study was assessed by determining the percentage of time of sleep that occurred in the assigned sleep period. This was the result ofthe total available time, in this experiment 16 hr, minus sleep latency and the awake time (stage 0) after the onset of sleep (A WI). For this regime the sleep efficiency was 77% (746 min/960 min). This compares with the other schedules as follows: 3 hr6 hr, 80%; 4 hr8 hr, 84%; 6 hr12 hr, 87.5%; 8 hr16 hr, 93.4%; 10 hr20 hr, 91.5%; and 12 hr24 hr, 87.8%. This more extreme extension of the circadian day (8 hr16 hr) resulted in a yet lower efficiency. This extension, however, did not reach the reduced efficiency ofthe shorter regimes of Weitzman et al. (1974) (1 hr2 hr, 56%) and Carskadon and Dement (1975) (Y2 hrl hr, 60%). As we shall continue to emphasize, the average percentage of 77% obscures the large individual differences in efficiency of response under this regime: SA = 93%, SB = 81%, SC = 75%, and SD = 60%. The three subjects (B, C, Downloaded from https://academic.oup.com/sleep/articleabstract/1/2/191/2749546 by guest on 05 September 2018 Sleep. Vol. I, No.2, 1978
SUBJECT A 12M 4 6 10 2N 4 6 B 10 12 1M? 4 ~ I SUBJECT B B 10 12N? 4 6 8 10 I ~ 1"lIfllll1Il BAO RECORD BAD RECORD 111111 ~1fI"" I SUBJECT C 12 M 2 4 6 8 10 12N 2 4 6 8 10 I t f f 12M 4 6 I I f e f SUBJECT D o 8 I 0 IN 4 6 8 I 0 I Downloaded from https://academic.oup.com/sleep/articleabstract/1/2/191/2749546 by guest on 05 September 2018 FIGS. 14. Sleep and waking in ten 48 hr cycles. Solid lines indicate sleep episodes. Each assigned sleep period began at 12 noon and terminated at 4 a.m. Each row is a 24 hr day.
194 W. B. WEBB and D) who were responsive to the schedule had a mean sleep efficiency of 72%. In earlier series it was noted that the sources of inefficiency differed for long and short regimes. The shorter regimes were characterized by longer latencies as well as substantial A WI (stage 0 after sleep onset). In contrast, the longer regimes had short latencies and large amounts of A WI as a result of the inability to sustain sleep continuously beyond 810 hr. For example, the longest previous schedule (12:24) had a mean latency of 8 min and A WI of 80 min. The present data were in line with the earlier findings, but again there were individual differences. Three subjects displayed latencies shorter than the 12:24 regimes (7.4 min, 5.7 min, and 2.2 min) but one had a mean latency of 29.6 min. Three subjects displayed sharply increased difficulties in sustaining sleep (A WI), (M = 179 min, 235 min, and 372 min) while one had only a mean amount of 62 min of A WI. However, there was accord with one other response measure. The earlier paper found the A WI in the first 2 hr after sleep onset to approach nil in the longer regimes. In the present condition only four records of the 339 available showed stage 0 in the first 2 hr, two of 1 min duration and two of 2 min duration. In the earlier study, an attempt was made to assess the tendencies to terminate sleep by assessing the stage 0 in the last 2 hr of each sleep period in the longer regimes. Figures 14 show that this is an inappropriate measure in this regime since sleep was often terminated and reinitiated. Clearly the maintenance of sleep across the 16 hr period was difficult. The following number of periods were interrupted by at least one consecutive hour of wakefulness terminated prior to the end of the sleep period for each subject: A, 0 of 10; B, 5 of 9; C, 9 of 10; D, 10 of 10. Even in subject A there were significant "bursts" of light sleep characterized by alternations among stages 0, 1, and 2.. An analysis of the sleep structure in terms of sleep stages reaffirms the remarkable stability of the overall sleep system across a variety of schedule variations (cf. Webb et ai., 1971; Webb and Agnew, 1974). Table 1 compares the sleep stage percentages obtained under the experimental schedule with the four baseline nights. This is derived from the sum of stages obtained by the subjects during the entire experimental period and during baseline sleep. Clearly, stage 2 continues to be substantial, stage 4 is sustained, and stages 1 and REM are somewhat enhanced. A )(2, testing the difference between baseline and experimental percentages, does not approach significance. In the earlier analysis of sleep stages, in other than 24 hr schedules, we suggested certain dynamic characteristics of sleep stages which would tend to sustain the overall structure in the various regimes: the temporal distribution of the stages, the relations to prior wakefulness, and the circadian effects of sleep onset time (Webb and Agnew, 1977). In regard to temporal distribution it has been well established that stage 3/4 occurs predominantly in the early part of sleep and stage REM shows hourly increases across the first 7 8 hr of sleep and then probably asymptotes in an hourly amount (Webb and Agnew, 1977). Because this hourly amount beyond 8 hr exceeds the baseline averages of stage REM, the percentage increase in REM sleep in periods exceeding 8 hr may be attributed at least partially to this tendency. Downloaded from https://academic.oup.com/sleep/articleabstract/1/2/191/2749546 by guest on 05 September 2018 Sleep, Vol. I, No.2, 1978
THE FORTYEIGHT HOUR DAY 195 TABLE 1. Baseline and experimental period sleep stage percentages" Stage I Stage 2 Stage 3 Stage 4 Stage REM Base line 5 48 5 17 25 Exp. period 9 44 5 14 28 Diff. +4 4 0 3 +4 " Total stages (minusstage 0) summed across conditions (4 Ss). By temporal distribution considerations alone one may expect the percentage of stage 4 to be more sharply reduced than that seen in Table 1. This, however, is partially offset by the "dynamics" ofthe sleep response in relation to prior wakefulness. Again, it has been previously noted that stage 4 increases as a function of the length of the prior wakefulness period (Webb and Agnew, 1971). In the earlier series it was noted that the amount of stage 4 in the first 3 hr of sleep rose from 18 min for 6 hr of prior wakefulness to 64 min for 24 hr of prior wakefulness. In the 36 hr regime of this experiment, there was an average of 62 min in the first 3 hr. There is an additional contribution to stage 3/4 amounts by the apparently reliable presence of a stage 3/4 "kick" in very extended periods of sleep (Webb, 1977). An examination of the records of subject A, who exhibited essentially continuous periods of sleep, i.e., with no extensive full wakefulness periods, reveals intervals of stage 3/4 in 9 of 10 records beyond the 9th hr of sleep onset. These records contained an average of 16 min of stage 3/4 sleep after the 9th hr. This was not unique to this subject. The examination of 9th hr and beyond of the three records of subject B and six of subject C, which were characterized by relatively short periods of awakenings, shows stage 3/4 in all nine records averaging 20.5 min per record. The increase in the stage 1 amounts continues the generally unpredictable character of this stage seen in the earlier regimes. There is evidence to indicate the persistence of the 24 hr circadian effects on sleep and contribution of these effects to the particular findings of this study. Earlier studies have shown a reduction in REM latency when sleep onset occurs during the circadian "day" (cf. Weitzman et ai., 1970; Webb et ai., 1971). In this study the baseline mean latency (REM) for the four subjects was 112 min. The mean latency of REM during the experimental period was 68 min. This factor would contribute to the enhanced percentage of REM (Table 1). Increased awakening in day time sleep schedules have been reported (Weitzman et ai., 1970; Webb and Agnew, 1971). In this study, in spite of prolonged periods of prior wakefulness, substantial awakening can be seen in the first 8 hr sleep records of these subjects. Periods of awakening of 5 min or more in the first 8 hr were present in 6 of 10,6 of 9,3 of 10 and 8 of 10 records for subjects AD, respectively. DISCUSSION The sleep responses offour subjects to systematically imposed regimes of 32 hr of wakefulness and 16 hr sleep time available, although marked by individual differences, in general conform to the earlier findings of non24 hr schedules of sleep and waking. The utilization of the sleep period becomes less efficient as the Downloaded from https://academic.oup.com/sleep/articleabstract/1/2/191/2749546 by guest on 05 September 2018 Sleep, Vol. I, No.2, 1978
196 W. B. WEBB schedule increasingly deviates from the normal regime of an approximately 16 hr wakefulness8 hr sleep schedule. In extended regimes the limitation of efficiency is primarily attributable to the inability to maintain sleep effectively beyond some 8 hr to 10 hr. The analyses of the sleep stage data again attest to the remarkable stability of the general underlying structure of sleep defined by sleep stage amounts previously noted in timefree environments (Webb and Agnew, 1974), sleep displacement studies (Weitzman et ai., 1970; Webb and Agnew, 1971), and non24 hr schedules (Webb and Agnew, 1977). A detailed analysis reiterates the dynamics of sleep stages in relation to prior wakeful time, the temporal distribution of stages within sleep, and circadian effects in maintaining this overall structure (see earlier discussion in Results section). It could be argued that the decreased amounts of sleep obtained under this and the other non24 hr regimes were not less "efficient" but, in fact, more "efficient." In short, subjects were "getting by" on less sleep or "needing" less sleep. Unfortunately, the experimental answer to this question, length of recovery sleep, is not available. Our arguments for reduced "efficiency" center on two postulates and some data. We postulate that sleep is naturally organized in a circa 16 hr wake8 hr sleep system and any variable resulting in less total sleep across time is less "efficient." Insomnia, we argue, is not "more efficient" sleep. Secondly, we find it more satisfying (although less parsimonious) to attribute the reduced (less efficient) sleep to such factors as increased latencies in short regimes and sleep termination tendencies in longer regimes, rather than to a hypothetical increased efficiency. Finally, we know that when sleep is permitted ad libitum in freerunning schedules, that sleep, in fact, does organize itself into more efficient reduced sleep periods but significantly increases in amount. There remains the intriguing question of the bicircadian or 48 hr day first noted by Siffre as a spontaneous rhythm in the timefree environment and experimentally explored by Chouvet et ai. (1974) and Jouvet et ai. (1974). Certainly subject A effectively adapted to such a 48 hr sleepwake schedule. However, a preliminary analysis of the temperature measures indicates that this subject, while adaptively maintaining 48 hr sleep waking patterns, dis pia yed a "damped" 24 hr temperature cycle. This will be discussed in a later paper. ACKNOWLEDGMENT This work was supported by National Aeronautics and Space Administration grant NGR1O005. REFERENCES Agnew HW Jr and Webb WB. Sleep stage scoring. J Abstr Supp/ Am Psycho/ Assoc Ms. No. 292, May 1972. Carskadon MA and Dement WC. Sleep studies on a 90minute day. Electroellcephalof(r Clill Neurophysiol 39: 145155, 1975. Chouvet G, Mouret J, Coindet J, Siffre M, and Jouvet M. Periodicite bicircadienne du cycle veillesommeil dans des conditions hors du temps. Etude polygraphique. Electroellcephalogr Clill Neurophysiol 37:367380, 1974. Jouvet M, Mouret J, Chouvet G, and Siffre M. Toward a 48hour day: Experimental bicircadian Downloaded from https://academic.oup.com/sleep/articleabstract/1/2/191/2749546 by guest on 05 September 2018 Sleep. Vol. I, No.2, 1978
THE FORTYEIGHT HOUR DAY 197 rhythm in man. In: FO Schmidt and FG Worden (Eds), The Neurosciences, Third Study Program, MIT Press, Cambridge, Massachusetts, 1974, pp 491497. Moses JM, Hord DJ, Lubin A, Johnson LC, and Naitoh P. Dynamics of nap sleep during a 40hour period. Electroencephalogr Clin Neurophysiol 39:627633, 1975. Webb WB. The spontaneous termination of sleep. Sleep Res 6: 118, 1977. Webb WB and Agnew HW Jr. Stage 4 sleep: Influence of time course variables. Science 174: 13541356, 1971. Webb WB and Agnew HW Jr. Sleep and waking in a timefree environment. Aerospace Med 617622, June 1974. Webb WB and Agnew HW Jr. Sleep efficiency for sleepwake cycles of varied length. Psychophysiology II :265274, 1975. Webb WB and Agnew HW Jr. Analysis of the sleep stages in sleepwakefulness regimens of varied length. Psychophysiol Res 14:445450, 1977. Webb WB, Agnew HW Jr, and Williams RL. Effect on sleep of a sleep period time displacement. Aerospace Med 42:152155,1971. Weitzman ED, Kripke D, Goldmacher D, McGregor P, and Nogeire C. Acute reversal of the sleepwaking cycle in man: Effect on sleep stage patterns. Arch Neurol 22:483489, 1970. Weitzman E, Nogeire C, Perlow M, Fukushima D, Sassin J, McGregor P, Gallagher T and Hellman L. Effects of a prolonged 3hour sleepwake cycle on sleep stages, plasma cortisol, growth hormone and body temperature in man. J Clin Endocrinol Metab 38: 10181030, 1974. Downloaded from https://academic.oup.com/sleep/articleabstract/1/2/191/2749546 by guest on 05 September 2018 Sleep. Vol. I. No.2. 1978