The question Michel Siffre stumbled into on the Scarasson glacier had been asked before, and it would be answered, in the end, with a precision no cave could give. This is the short history of the number.

Mammoth Cave, 1938

In the summer of 1938 Nathaniel Kleitman, the University of Chicago physiologist who had spent fifteen years on sleep, took a graduate student, Bruce Richardson, into Mammoth Cave in Kentucky for thirty-two days. The cave was a tourist attraction with electricity, beds and a hotel kitchen, and the two men told the newspapers that the only drawbacks were damp sheets and the want of another couple for bridge. Kleitman’s question was whether the twenty-four-hour cycle of sleeping and waking was a response to the conditions of life or a property of the body, and his method was to impose a different day, of twenty-eight hours, and see whether the body’s temperature followed it. His did not; the younger man’s partly did. The Chicago press made the experiment famous, and historians of the subject have noted that it also made the cave the natural place to ask the question.

Scarasson and Andechs, 1962 to 1965

Siffre’s approach in 1962 was the other one: impose nothing and let the body choose. His sleep and waking, logged by telephone, ran on a period of twenty-four hours and thirty-one minutes, as he reported in his book, and Franz Halberg’s analysis of the pulse, the sleep and the time estimates, published in the Comptes rendus of the Académie des sciences in January 1965, found the rhythms staying in step with one another while they drifted from the clock.

In the same years Jürgen Aschoff and Rütger Wever were doing the same thing under control. From 1962 they shut volunteers, Aschoff among them, into an underground bunker at Andechs in Bavaria, behind a double door whose magnetic catch let only one leaf open at a time, so that a subject could neither see out nor meet whoever brought his supplies; he could communicate only by letter. The subject chose when to turn his lights on and off and when to sleep; he collected his own urine, and his body temperature and his movements in bed were recorded. Aschoff reported the results in Science in June 1965. His own free-running period had been “about 25.0 hours”. Another subject’s was 25.9. After eighteen days in the bunker that man woke at eight in the evening; he had “lost” a day and a half. “There is, on the other side,” Aschoff wrote, “the surprising fact that 85 percent of all subjects studied so far had periods considerably longer than 24 hours. This is in agreement with Siffre’s records on his own activity cycle in a cave, and with the data published recently by Mills.” He offered three possible causes: the dim light, the absence of noise and social contact, and the feedback between a subject’s own activity and the light he switched on for it.

John Mills, the Manchester physiologist, had got his data by luck. Geoffrey Workman went down Stump Cross Caverns in Yorkshire in June 1963 to beat Siffre’s record, and agreed to collect his urine; Workman knew the time, from a watch and a daily telephone call to the cave’s owner, but his potassium excretion and his sleep, Mills found, drifted together on a day a little over twenty-four hours, as if, Workman wrote in his own account of the stay, “some intrinsically rhythmic process, a biological clock, governed alike the rhythmicity in sleep and potassium excretion, and that this clock was not keeping perfect 24 hour time”. Mills found it remarkable that a man who knew the time and a man who did not had cycles of similar length.

So by 1965 the answer was in: the human body has a clock, it runs without the sun, and it runs slow, at about twenty-five hours. That number stood for thirty years. It is still in textbooks.

The forty-eight-hour day

The bunker also showed something stranger. In one of Aschoff’s subjects the rhythms came apart. Activity and the excretion of calcium kept one period; body temperature, water and potassium kept another; the two sets fell back into step every third or fourth day and then drifted again. Aschoff called it desynchronisation, and his Figure 7 shows it; another of his subjects had put in two stretches of more than twenty hours awake before his rhythm settled. A textbook of 1981 lists Aschoff and Wever’s bunker, in 1967, among the places a forty-eight-hour sleep-wake regime had been reported.

Siffre’s cavers went further. In 1966 a man who did six months underground reached forty-five hours between wakings within ten days, while his temperature kept its day. In 1968 Philippe Englender settled into an almost exactly regular forty-eight-hour day, thirty-six hours up and twelve down, and Jacques Chabert followed him. Michel Jouvet’s laboratory in Lyon, which read the brain waves, published the series in 1974 as a “bicircadian” periodicity: all the subjects, with variable latencies, had come to about thirty-four hours of waking and fourteen of sleep. NASA translated the paper in 1975. The French army was interested because a soldier who kept such a day could be used for thirty-six hours at a stretch; Siffre was interested because he had never caught it himself, and went to Texas in 1972 to try, and caught it twice in 205 days and could not tell those days from the others.

What the forty-eight-hour day was has never been settled to everyone’s satisfaction. Siffre, asked in 2008 why people fell into it, said he had no theory and did not make theories; it was a fact he had observed. The bunker’s description of it as a splitting of sleep from the temperature clock, with the sleep cycle free to stretch when nothing constrained it, is the one the field has kept. The caves added a condition the bunker shared: a subject who chose his own light and had every reason to stay up.

Harvard, 1999

The light turned out to be the problem. In June 1999 Charles Czeisler’s group at Harvard reported in Science that the long periods of the isolation studies, a median of 25.2 hours for the activity rhythm and about 25 hours for body temperature, “were based on studies of humans exposed to light levels sufficient to confound circadian period estimation”. A subject who chooses when to turn on his lamp turns it on at what his body takes to be its morning and keeps it on into what his body takes to be its night; ordinary room light at the end of the biological day pushes the clock later; and so the period measured was the clock plus the lamp. Measured on the rhythms of melatonin, body temperature and cortisol under lighting controlled to prevent that, in young and old alike, “the intrinsic period of the human circadian pacemaker averages 24.18 hours”, with “a tight distribution consistent with other species”. Twenty-four hours and eleven minutes. The human clock is not slow. It is nearly right, and it was being read through the very light it was supposed to be free of.

That reading does not undo the caves. Siffre found the clock, and found that a man cannot feel it running; Aschoff and Mills found it by three independent routes in three years; the drift was real, and so was the fact that a body kept its own day while its owner lost a month. What changed was the number on the dial, and the knowledge of which part of the apparatus had been setting it. In 1962 the lamp in the red tent was a tool for reading. It was also, it turned out, the hand that moved the clock.

Sources

• Kirsten D. Hussey, “Timeless spaces: Field experiments in the physiological study of circadian rhythms, 1938–1963,” History and Philosophy of the Life Sciences 45, 2023. Kleitman and Richardson at Mammoth Cave, the twenty-eight-hour day, Siffre’s approach, Mills and Workman. europepmc.org
• Jürgen Aschoff, “Circadian Rhythms in Man: A self-sustained oscillator with an inherent frequency underlies human 24-hour periodicity,” Science 148, 1427–1432, 11 June 1965. The bunker, the 25.0 and 25.9 hours, the 85 percent, the three possible causes, and the desynchronised subject. mechanism.ucsd.edu
• G. H. Workman, “An Experiment in Underground Living,” Memoirs of the Northern Cavern and Mine Research Society 2 (3), 1973, quoting J. N. Mills, “Circadian rhythms during and after three months in solitude underground,” Journal of Physiology 174, 217–231, 1964. Mills’s words are given at second hand from Workman. nmrs.org.uk
• F. Halberg, M. Siffre, M. Engeli, D. Hillman and A. Reinberg, Comptes rendus de l’Académie des sciences 260, 1259–1262, 1965; J. Colin and colleagues, Journal of Applied Physiology 25, 170–176, 1968; G. Chouvet and colleagues, Electroencephalography and Clinical Neurophysiology 37, 367–380, 1974. Cited from their titles and from L. Zuccarelli and colleagues, Frontiers in Physiology 10, 2019, and the NASA bibliography abstract of 1975. europepmc.org
• Reuters, “48-Hour Day,” Press (Christchurch), 25 October 1968, via Papers Past. paperspast.natlib.govt.nz
• Joshua Foer, “Caveman: An Interview with Michel Siffre,” Cabinet 30, 2008. The two occasions in Texas and “I don’t make theories”. cabinetmagazine.org
• Biological Rhythms, edited by Jürgen Aschoff, Plenum, 1981, for the list of reports of a 48-hour sleep-wake regime, read in a snippet on archive.org. archive.org
• Charles A. Czeisler, Jeanne F. Duffy, Theresa L. Shanahan and colleagues, “Stability, Precision, and Near-24-Hour Period of the Human Circadian Pacemaker,” Science 284, 2177–2181, 25 June 1999. The 25.2 and 25 hours, the confounding light, and the 24.18 hours, from the abstract. pubmed