At half past eight on the morning of Friday, 14 September 1962, a field telephone was cranked in a red tent pitched on a glacier 130 metres under the Marguareis massif, on the border between France and Italy. The man who cranked it was twenty-three. He had just woken, and he did what he had done at every waking for two months: he gave his pulse, counted aloud to 120 at what he believed was one number a second, and told the man on the surface what he thought the date was. He thought it was about 20 August. The man on the surface, a speleologist named Pierre Nicolas, told him it was 14 September, and that the experiment was over.

The conversation that followed was taken down, and it is printed in the book the man below wrote afterwards. “It’s Friday, September 14,” Nicolas says, “at 8:30 in the morning.” The man below asks about his days. “Yes, your days,” Nicolas says. “My ‘days’ were short,” the man below answers. Nicolas tells him that “something else is still more amazing”, and that as soon as he wakes tomorrow he must start getting ready to come up. “No!” says the man below. “Yes,” says Nicolas. The man below hung up and went to bed at ten past ten in the morning.

He had been underground fifty-eight days and seventeen hours, 1,409 hours by the surface log, and he had lived them, by his own count, as about thirty-five days. He was twenty-five days behind the calendar. The machine that had kept the true count was not in the tent with him. It was a stack of paper on the mountain above, on which, at each of his calls, a policeman or a caver had written the real time beside the time he guessed. That stack was the instrument. The man below, Michel Siffre, a geologist from Nice, had designed it, and he was the one thing in the experiment it could not be shown to.

Who ran it

Michel Siffre was born in Nice on 3 January 1939 and went into his first cave at ten. He trained as a geologist in Paris, and his teacher there, the marine geologist Jacques Bourcart, presented his notes to the Académie des sciences from the time he was seventeen; the Académie’s Comptes rendus for the summer of 1962 carry a note of his, presented by Bourcart, on tracing the underground streams of the Marguareis with dye. Bourcart, according to a profile of Siffre in Le Monde in 2004, saw in him “a young Darwin, the one of the Beagle years”. In 1962 Siffre founded an Institut Français de Spéléologie, of which he was president. He was a spéléologue: a caver, and in his own account a professional one.

In August 1961 he led an expedition from his Nice club, the Club Martel, to the Marguareis, a limestone massif of the Maritime Alps whose caves open at 2,000 metres and run down for hundreds. On 22 August 1961 the party, which included the caver Philippe Englender and two men of the CRS, the French riot and mountain-rescue police, Lafleur and Canova, found, at a depth of between 340 and 430 feet in a shaft called the Scarasson, a great cave floored with ice: a small glacier, underground, and fossil, as a later book of his puts it. It was so cold and so wet in there, Le Monde records, half a degree below freezing with the air at 98 per cent humidity, that the explorers stayed an hour.

Siffre decided to go back and camp on the ice to study its movement. Fifteen days, he thought at first. “Well, fifteen days is not enough,” he told an interviewer in 2008, remembering the thought. “I shall see nothing.” So two months. And then, as he said, “this idea came to me—this idea that became the idea of my life. I decided to live like an animal, without a watch, in the dark, without knowing the time.” He had read some biology in his student room and found a science that barely had a name yet, the study of the body’s rhythms, and had read of the rats whose activity kept a daily cycle in constant darkness. Nobody had tried it on a man for two months.

The people who ran the experiment with him were cavers and policemen. The surface camp was led by Abel Chochon; three speleologists, Jean-Pierre Mairetet, Jacky Meunier and Pierre Nicolas, kept the watch with the CRS men, Chief Sergeant Lafleur, Sergeant Canova and Sergeant Sprenger; the Protection Civile of the Alpes-Maritimes lent a helicopter. There was no scientific institution behind him. The Spelunca, the bulletin of the French caving federation, reported the next spring that he had tried to sell the story to the press but had wanted to control the texts and the headlines, so no deal was done; that he had no subsidy, and that he was helped a great deal all the same, by the helicopter and by up to sixteen CRS of the mountain rescue. In Paris, Le Monde says, people laughed in his face, and he went into debt to the neck. His own book says his friends thought him a madman.

The scientists came afterwards. Franz Halberg of the University of Minnesota, who had coined the word circadian a few years before, took the surface logs and analysed them with Siffre, Alain Reinberg and others in a paper to the Académie des sciences in January 1965; the analysis, Siffre said in 2008, was paid for by NASA. The psychologists Paul Fraisse and Geneviève Oléron took the time estimates. For the second experiment, ten years later, the sleep physiologist Michel Jouvet of Lyon read the brain-wave records, and a NASA scientist, Malcolm Smith, arranged the subject’s food.

What they were trying to find out

There were two questions, and they were not of the same kind.

The first was about survival. Siffre wrote in his book that he wanted to “extend the physiological limits set for human endurance”, and he spelt out why that mattered in 1962: an age in which the atom might oblige mankind to take refuge in underground shelters, and a dawning age of space travel in which men would live without any points of reference in time or space. He called the whole thing, later in the same book, “my survival experiment”. The newspapers took it the same way: the Reuters report of his descent said he “wanted to study the psychological effects of isolation and its applications in the atomic era”.

The second question was the one that lasted. A man cut off from every cue to the time, from the sun, from clocks, from other people’s hours, will sleep and wake. Will his sleeping and waking keep to a day? If so, what length of day, and set by what? Siffre’s protocol for answering it was, as he said, simple: he would call the surface when he woke, when he ate and before he slept, and the surface would never call him, so that nothing could reach him from the world of clocks. At each call he gave his pulse, taken at the wrist or the carotid artery between two signals, and counted to 120 at a number a second; the team timed the count. And at each call he said what time and date he believed it to be. The real time was written beside the guess.

In 1972 the question had moved. By then Siffre had sent a dozen other people into caves, and nearly every one of them had, after some weeks, slipped into a day of about forty-eight hours, thirty-four to thirty-six hours awake and twelve to fourteen asleep, without noticing. He had not, in 1962. “All of the other people I had put underground caught a forty-eight-hour sleep/wake cycle, except for me,” he said in 2008. “I decided I would stay underground for six months to try to catch the forty-eight-hour cycle.” He also wanted to measure what ten years had done to his sense of time. The Associated Press, in its report from Del Rio on the day he went down, put the hypothesis as he gave it: “that man can live without undue fatigue on a rhythm of 34-36 hours activity and 12-14 hours of natural sleep, instead of the normal 24-hour cycle lived in the outside world.”

What they built

Scarasson, 1962. The camp was a tent on the ice. The air in the cave never rose above freezing and was saturated with water; Siffre’s book says the humidity was always 100 per cent, and that he washed his dishes “standing on a floor of ice”. Everything condensed; the groundsheet, the camp bed and the sleeping bag were wet from the start. He had a small stove he did not dare run at night for fear of carbon monoxide, remembering that Admiral Byrd had nearly died of it alone in the Antarctic in 1934. He had books, a tape recorder and a pile of 45-rpm records. He had no watch. On 16 July 1962 he handed it to Sergeant Canova at the surface camp, a small tent at 2,000 metres, and went down at two in the afternoon. At his own request the ladder was then pulled up, so that he could not climb out on an impulse.

The one instrument that mattered was the telephone. It was a hand-cranked field set on a line to the surface, and it was one-way: he could call them, they could not call him. The Reuters correspondent at Nice noted the arrangement in both his reports, going in and coming out, as the thing that made the stunt an experiment. The surface team kept the log of calls against real time, and one of the CRS men, with time on his hands, began to draw it as a chart.

Midnight Cave, 1972. Ten years later the camp was, by comparison, a laboratory. Midnight Cave is a limestone cave on Tom Harding’s ranch on the Edwards Plateau, about fifty miles from Del Rio, Texas, near the Mexican border; the Associated Press reporter who drove out for the ending said the last twenty miles were a rough dirt track and the base camp was reached on foot over the rocks. Its air stayed at 20 or 21 degrees Celsius the year round. Siffre’s team built a wooden platform in a chamber about a hundred feet down, at the end of 700 feet of passage with crawls in it, and pitched a nylon tent on it. The tent had electric light, controlled from above, a freezer, cooking heat and furniture. A field telephone linked it to the surface, one-way as before.

He was wired. A rectal probe recorded his body temperature, electrodes on his chest took his heartbeat, and miniature Beckman electrodes glued to his head and face recorded the stages of his sleep; the cables plugged into a belt and the belt into a line to a laboratory tent on the surface, where four Frenchmen kept watch round the clock: Jacques Chabert, Gérard Cappa, Raymond Valente and a fourth whom the Associated Press named as Englender in February and as Jean-Pierre Maison in September. The agency’s tally at the end was a body temperature every five minutes for 4,920 hours, a heart rate for 160 days, brain waves on 140 nights and more than 10,000 tests. His food was frozen meals identical to those the Apollo 16 astronauts ate just before and after their flight, supplied by Marriott’s airline-catering arm through Malcolm Smith of NASA, a man with, as Siffre put it, “an eye cocked toward lengthy space flights of the future”. He had books and records, though by May, he wrote, he had only the books, and a pellet rifle, with which he shot at a target to gauge his coordination. He had no watch and no radio.

The money was a different matter. The experiment was backed, the Associated Press said, by his own institute, the French Ministry of Defence and NASA; Reuters named the United States Cave Research Foundation as well. American firms lent computer equipment. But Siffre had left Paris for Texas in the third week of October 1971 expecting to go down on 20 November, and he went down on 14 February 1972, three months late, because the money came slowly. “At one time we barely had enough to eat up at the base camp,” he told the agency. He estimated afterwards that there was perhaps half a million dollars’ worth of equipment in the tents above him, and he came out, he said in 2008, 100,000 dollars in debt.

The twentieth of August

What happened in the ice cave in 1962 is best read in the surface log, and the surface log is best understood by what it did not contain.

Siffre’s body kept time. From the first days his calls came in a pattern: a waking, meals, a sleep, a waking. When the log was analysed, the pattern ran on a period of twenty-four hours and thirty-one minutes, and in the book he says it “remained unbroken” but for a few days. A man with no sun was living a day half an hour longer than the earth’s, and living it with a regularity nobody had expected. By the tenth day, he wrote, his rhythm was “completely reversed”: he was sleeping through the daylight hours of the world above and active through its night, without knowing it. He drifted on round the clock. On the day he was told the experiment was ending he had come full circle and was reversed again.

Siffre’s mind did not keep time. He slept, early on, twelve or thirteen hours at a stretch; later he believed he was sleeping six or seven. He sat up what felt like a few hours and had been awake fourteen or eighteen. When he lay down for what he took to be a siesta of an hour or two, and then rose, spent an “afternoon” and ate “dinner”, he had in fact completed a second full cycle of sleep and waking. Each such cycle was a day to his body and half a day to him. He counted his awakenings in his diary and still could not make the count come out. Near the end he wrote that he had wakened forty-one times, “which means I have passed forty-two physiological days here. Yet sometimes I conclude that I have been here only twenty-seven days of twenty-four hours each.” He tried averaging the two: “If I add 27 to 43 I get 70. Dividing that by 2, I get the mean time, or thirty-five days. I came down July 16, so it must be approximately August 20.” The log on the surface said fifty-eight. He was sixteen days short by one of his counts and thirty-one by the other, and the average he trusted put him twenty-five days behind the calendar. The 120-count, meanwhile, had been telling the surface the same thing at every call: he took five minutes to count what he believed was two.

The telephone could have told him. It did not, because he had designed it not to, and because the men above kept to the design even when, in the last weeks, he was plainly failing. The CRS man’s chart showed them a rhythm of remarkable regularity; it showed them that when their man was awake longer his night was shorter, and that the whole came out at twenty-four and a half hours. It also showed them, against his guesses, the gap opening day by day. They logged it and said nothing. That is what makes the experiment what Charlie Buffet, writing in Le Monde in 2004, called the only “pure” one of its kind: the subject did not know he had an internal clock, and so could not be influenced by the knowledge. Everyone who went down after him knew.

The ending was managed as the design required. Nicolas told him on the Friday. Siffre’s first thought, he said afterwards, was that they were lying to get him out early; he believed he had a month to go. A conversation he judged to have lasted five minutes lasted twenty. A team came down for him at the weekend, and on 17 September, at 7:30 in the morning by the Spelunca’s account, he began the ascent of the shaft, in good form at first. He was not in good form by the top. “Like a disjointed marionette,” he wrote, “I am hoisted to the surface. September 17, 1962, 11 a.m.” He blacked out twice on the way up; the next he knew, he was on a stretcher under a blanket. The police put a canvas hood over his head against the light, two friends carried him on their backs across the mountain to the helicopter, and he was flown to Nice and on to Paris for examination by doctors and psychologists. The Reuters man at Nice heard him say, as he came out, that it was good to see the sky again, and that alone at the bottom he had “cried and sometimes wept uncontrollably”. A press photograph a week later showed him being helped down the steps of an aircraft at Orly, barely able to walk, in dark glasses.

The newspapers counted sixty-two days and called it a world record; the Italians who held the old one had done 700 hours, under thirty days. The British caving press read the reports as a warning. Geoffrey Workman, who went down Stump Cross Caverns in Yorkshire the next summer to beat the record, recorded that the papers “suggested that he was in a state of physical and psychological breakdown at the end, and he had to be helped out of the cave”, and that his caving friends took two months to be the limit of what a man alone underground could stand. Workman stayed 105 days and came up fit, having known the time throughout; the limit, he concluded, was in the equipment and not the man.

What it looked like from inside

The record of the inside is the diary Siffre kept with a 4.5-volt lamp, an arm out of the wet sleeping bag, and published a year later as Hors du temps, and in English as Beyond Time. It is a record of a mind trying to find purchase on a thing that had no edges.

“I had the impression of being motionless,” he wrote in it, in the passage Le Monde chose to quote forty years on, “and yet I knew I was being carried along by the uninterrupted flow of time.” Falling asleep became the one pleasure and waking the worst moment: he would lie with his eyes open in the total dark wondering whether he was asleep or awake, hoping to be still asleep, and then, resigned, press the switch that broke “the unity of the night”, and crank the telephone. His memory failed him in a way that frightened him. He put the same side of a record on for the third time without knowing it, and after what he numbered his fifty-first awakening recorded an “unprecedented lapse of memory”; the CRS men, who sometimes listened in, told him afterwards that he had played the same Luis Mariano song as many as ten times running. “After one or two days,” he said in 2008, “you don’t remember what you have done a day or two before. The only things that change are when you wake up and when you go to bed. Besides that, it’s entirely black. It’s like one long day.”

The cave was not quiet. Twice the tape recorder running during his calls caught the crash of rock and ice coming down, and one morning a series of what he called immense cave-ins brought blocks of stone and ice down a few metres from the tent. “I begin to be really terrified,” he wrote. He later believed the fright had saved him. Early in the stay he had taken his temperature, read 36 degrees, decided the thermometer was broken, and stopped; in the book he says the mercury would not rise above 97 degrees Fahrenheit and that this was not the thermometer. “I was in a condition of semihibernation, what the doctors call hypothermia.” In 2008 he said his temperature had got as low as 34 degrees Celsius. The figures differ; the lethargy they describe does not. He had amoebic dysentery, brought back from an expedition to Ceylon, and it came in bouts that made him fear paralysis. He danced the twist alone in the tent, “like a bearded madman”, he wrote. He dreamed, he said in 2004, violent dreams in which he settled accounts with the people who had refused to fund him.

Ten years later, in a warm cave with electric light, the inside was worse. The first weeks in Midnight Cave were almost cheerful; he had food that delighted him and luxuries undreamed of at Scarasson, and he wrote in the diary he kept for the National Geographic of “the longest beyond-time experiment in history”. He measured his life in cycles, from awakening to awakening, and on the fifth day, 18 February, noted that since the sixteenth he had been living by cycles rather than days. On the seventy-seventh day, 30 April, cycle 63 by his count, he wrote that he had passed a significant mark: “In 1962 I spent exactly 63 days in Scarasson, emerging in acute physical and emotional distress.”

Two days later it began. “79TH DAY (MAY 2, CYCLE 65) My crack-up commences with a few lines in a book by French speleologist Corentin Queffélec.” Mildew was spreading over everything; he had counted on music to sustain him and now had only the books; dust from the cave floor made him wonder whether six months of it would infect his lungs. He stumbled, cycle after cycle, to a particular rock at the farthest limit of his chamber and back, “overcome with despair, with feelings of overwhelming self-pity”. Then, he wrote, “the perfect solution occurs to me: suicide. I will, of course, make it appear an accident.” He did not act on it. He said down the telephone, through clenched teeth, “J’en ai marre!—I’ve had enough!”, and his old friend Gérard Cappa, on the other end, did not reprove him: “Yes, yes,” he said. On 17 May, cycle 79, Siffre disconnected the probe and the electrodes and gave himself a ten-cycle rest “to rebuild my confidence”, and was then overwhelmed with guilt in case he was living forty-eight-hour days at that moment “and robbing my colleagues of vital data”. A journalist who met him in 2018 recorded one more hazard of the wiring: lightning came down the cable during a storm and gave him a violent jolt.

In the fifth month he found company. Early in the stay he had heard mice in the dark and believed he had wiped out the colony. Then a mouse came foraging among his supplies; his heart leaped. He named it Mus, and by the 163rd day, 25 July, he was spending his days studying its habits. He tried to catch it under a casserole dish propped on a book above a smear of jam, firing the pellet rifle at the book to drop the dish; Mus scampered away. He waited with the dish in his hands. On the 170th day, 1 August, “my patience prevails”: the mouse edged up to the jam, he brought the dish down, and the rim caught it on the head. “I stare at him with swelling grief. The whimpers die away. He is still. Desolation overwhelms me.”

Nine days later the telephone rang. “179TH DAY (AUGUST 10, CYCLE 151) … Gérard tells me that it is August 10, a stormy day, and the experiment has concluded. I am confused; I believed it to be mid-July.” Then, “as the truth sinks in, comes a flood of relief.” He had been out of time for 179 days and had lived them as 151 cycles, and his sense of the date was, again, about a month behind. He stayed underground nearly four more weeks for other tests, knowing the time now. In the last week of August his wife Nathalie, whom he had married a year before going down, climbed 120 feet down the steel ladder and crawled the 700 feet to the tent for a two-hour visit; “I found him very much changed, to be sure,” she told the Associated Press, “yet still the same.”

What happened

The clock. In 1962 the body kept a day of twenty-four hours and thirty-one minutes for two months in the dark, and the man who owned it had no idea. The first published analysis, Halberg’s in the Académie’s Comptes rendus of January 1965, treated the pulse, the alternation of waking and sleep and the estimates of time as rhythms running free, and found, as a review of the cave studies summarised it half a century later, that they had stayed in step with one another while drifting away from the clock. Jürgen Aschoff, who had been shutting volunteers into an underground bunker at Andechs in Bavaria since 1962, reported in Science in June 1965 “the surprising fact that 85 percent of all subjects studied so far had periods considerably longer than 24 hours,” and added: “This is in agreement with Siffre’s records on his own activity cycle in a cave, and with the data published recently by Mills”. John Mills, a Manchester physiologist, had analysed Workman’s urine at Stump Cross in 1963, found his sleep and his potassium excretion keeping a day of a little over twenty-four hours together, and remarked that Workman, who knew the time, and Siffre, who did not and “made such erroneous estimates”, had cycles of similar length. By the middle of the decade the result Siffre had stumbled into with a telephone had been found three ways: an internal clock, running free, a little slow.

The others. With the French army now funding him, Siffre stopped being his own subject. From November 1964 he put a furniture maker, Antoine Senni, and a midwife, Josiane Laures, into two caves north of Nice for 126 and 88 days; each telephoned him, each was told nothing, and each lost about a fortnight a month. When he rang Laures on 11 March 1965 to say it was over she thought it was 25 February and said, Reuters reported, “It is not true. You must be lying. Have I really been here three months already?” A third man did six months in 1966, and within ten days his cycle had stretched to forty-five hours between wakings. In the Olivier cave in 1968 Philippe Englender fell after thirty days into an almost exactly regular forty-eight-hour day, thirty-six hours of work and twelve of sleep, and Jacques Chabert followed him into it in December; Siffre told Reuters the tendency had been suspected before but never shown so clearly. The army wanted to know whether a soldier’s waking day could be doubled. The other subjects, and what became of them, have a piece of their own, The speleonauts.

Midnight Cave. The record from Texas is the one the physiologists had wanted: six months of a man’s sleep on paper, with his temperature every five minutes beside it. It did not show a forty-eight-hour day. For the first thirty-five days Siffre’s body kept a regular day of about twenty-six hours, and Chabert told Reuters in June that he had settled to a twenty-six-hour day with about eight hours’ sleep. On the thirty-seventh day, the mathematician Steven Strogatz found when he worked through the published record in 1986, Siffre “unknowingly skipped his usual bedtime” and stayed up for hours; what Siffre himself called “wild variations” followed, then nine more weeks of steady twenty-six-hour cycles, then, from about the 130th day, wild variations again in a seemingly random pattern. Over the 205 days his average cycle was about twenty-eight hours, 175 cycles in all; some of his “days”, a school science weekly told American children that November, lasted fifty hours. The Associated Press put his longest waking stretch at thirty-six hours; he said himself, later, that he had sometimes slept eighteen. “I did tend toward a 48-hour rhythm on two widely spaced occasions,” he wrote in 1975, and in 2008: “but not regularly. I would have thirty-six hours of continuous wakefulness, followed by twelve hours of sleep. I couldn’t tell the difference between these long days and the days that lasted just twenty-four hours.”

Jouvet’s group published the sleep records from Siffre’s programme in 1974, in a paper whose title announced a “bicircadian periodicity” of the sleep-wake cycle under timeless conditions; its summary said that, with variable latencies, all the subjects had reached a two-day rhythm of thirty-four hours awake and fourteen asleep. NASA had the paper translated into English within a year. Siffre was still saying in 1988 that five of his seven long-stay subjects had abandoned the circadian period for a “circabidian” one. What the two-day day meant, and what the laboratory finally made of the twenty-five-hour one, is in The length of a day.

The exit. Siffre came out of Midnight Cave at eleven in the morning on Tuesday, 5 September 1972, the 205th day, having shaved off his beard, to his wife, his four companions and what his National Geographic caption called a regiment of newsmen. “It is over. I emerge into the clear air of southwest Texas,” his diary says. That day the Olympic Games at Munich were under siege, Israeli athletes held hostage by Palestinian gunmen, and the Frenchman out of his cave was not the story; Le Monde records that, for the first time, his motivation left him.

Miles of tape, he wrote in 1975, had been delivered to NASA’s laboratories, where they awaited computer processing. He could not afford to follow them. “When I came out of Midnight Cave in 1972, I was 100,000 US dollars in debt,” he said in 2008. “I had badly underestimated the cost of bringing my experiments from France to Texas, and I had to leave the field of chronobiology. Much of my data from that experiment has yet to be mathematically analyzed.” In an interview with L’Express in February 1974 he said that in the absolute solitude he had suffered a serious nervous depression, and that it had taken him a year to come out of it. In March 1975, three years after the cave, he wrote in the National Geographic: “I still suffer severe lapses of memory. My eyesight has weakened”. He went back to caves of another kind, in Guatemala, where he found Maya paintings; his book on them came out in 1993.

After

On 10 August 1988 a Frenchwoman, Véronique Le Guen, went down the Valat-Nègre cave in the south of France under Siffre’s direction and stayed 111 days, a world record, the American magazine In Health said in 1991; a week after descending she began a “day” that lasted seventy-four hours, forty awake and thirty-four asleep. She came out in November 1988, and, as the magazine put it, the darkness stayed with her. Within two years she had killed herself. Her partner, the diver Francis Le Guen, told the magazine’s writer that he did not blame Siffre. Asked by the same writer whether he had ever contemplated suicide in Midnight Cave, Siffre, who had written in 1975 that he had, answered: “Oh, no, never.”

In 1999, when John Glenn went back into space at seventy-seven, Siffre decided to measure what age had done to his clock. He went down the Clamouse cave in the Hérault on 30 November 1999, aged sixty, with electricity, electrodes, equipment lent by NASA, and The Three Musketeers, which bored him furiously. He saw the millennium in three and a half days late, on 4 January 2000, and missed his sixty-first birthday. He came up on 14 February 2000 after seventy-six days believing it was 5 February. He was nine days behind the calendar this time, not twenty-five; this time, as Le Monde put it, he knew he had an internal clock to trust.

He had been right about where the work would go. The bunker Aschoff built in Bavaria, and the laboratories built after it, confirmed the free-running clock on more than a hundred subjects under conditions he could never afford. They also, in the end, read his instrument differently. In 1999 Charles Czeisler’s group at Harvard reported in Science that the long periods found in forty years of isolation studies, about twenty-five hours, “were based on studies of humans exposed to light levels sufficient to confound circadian period estimation”: the subjects, by switching their own lights on late and off late, had been pushing their own clocks back a little every day. Measured with the light controlled, the human period averaged 24.18 hours, young or old. The clock Siffre discovered is real. Most of the half-hour he measured on it was, by the later reading, his lamp.

He said in 2008 that the cave experiments were finished: doctors today would have had to wake Senni at thirty-three hours. He said he had no theory of the forty-eight-hour day and did not make theories; it was a fact he had observed. He lived out his last twenty years in Nice, where a visitor in 2004 found his flat stacked to the ceiling with the boxes of a life, and he died there on 25 August 2024, aged eighty-five.

What it taught

Start with what was found. A man alone in the dark for two months slept and woke on a day of about twenty-four and a half hours, kept it as steadily as a clock, and had no idea; the same man, ten years later, with his brain wired, kept a day of twenty-six hours for five weeks and then lost the shape of his days altogether, and had no idea of that either. Both times, when the telephone finally told him the date, he was about a month behind. The first result founded a field and has held, with the correction that the half-hour was the lamp’s and not the body’s. The second was never fully analysed. Both were bought with one man’s health, at a price his later denials suggest he did not like to count.

The subject was the instrument, and an instrument cannot read itself. Siffre built an experiment whose only sensor was his own judgement of elapsed time, and whose only recorder was a log he was forbidden to see. The design was sound; it is the reason the result is clean. But it meant that the instrument ran for two months with nobody able to tell it that it was reading wrong, and that when it was finally shown its error, in a twenty-minute telephone call it judged to have lasted five, the error was twenty-five days. The surface log was right because it was not his. Any experiment in which the observer is also the thing observed needs a second record, kept out of his reach, and should expect him to disbelieve it when it is shown to him.

Felt time and body time are two clocks, and the gap between them is itself the measurement. The body kept its twenty-four and a half hours without help. The mind, with every waking and every meal to count, lost a day in every two. Siffre’s arithmetic on his last morning, averaging twenty-seven and forty-three to get thirty-five, is the record of a careful man trying to reconcile two readings and trusting neither; both were wrong. His own explanation, that in the dark memory “does not capture the time”, is still as good as any. Anyone who plans to rely on a human sense of duration, on a submarine, in a shelter, on a long flight, should know that it can halve.

A free-running clock is only as pure as the darkness around it. For thirty years the length of the human day was taken to be about twenty-five hours on the strength of Scarasson, Andechs and their successors. Then it was found that the light the subjects turned on for themselves had been setting the very clock being measured. The caves were not dark; they had a lamp, and a lamp is a cue. Siffre’s forty-eight-hour days, which he insisted on to the end, turned up in the bunker too, in a few subjects whose sleep came loose from their temperature rhythm, and they look in that light less like a second human rhythm than like what a clock does when its owner’s lamp and its owner’s exhaustion pull it in different directions. The second reading did not make the first one worthless; it told the field what the first had been measuring.

One subject at a time is a case history, not a sample, and the subject pays the whole bill. Siffre said in 2008 that the people he sent underground were cavers because nobody else was motivated enough to stay. The motivation was the problem. A caver who wants his record will stay through a crack-up that a volunteer would not, and a director who was once that caver will let him. Two months on the ice cost Siffre a collapse at the surface; six months in Texas cost him a year’s depression, his memory, his savings and the field he had founded; a programme run on the same terms had, by 1991, a death in it. The free-running clock could have been found, and was found, in a bunker with a door the subject could open. What the caves added was the record of what months of it do to a mind, and that record exists because nobody was in a position to stop it being made.

Check the story before you learn from it. He was underground sixty-two days by the newspapers, sixty-three by his own count ten years later, and out of time for fifty-eight days and seventeen hours by his book. He thought it was 20 August; Reuters remembered mid-August. His cycle was twenty-four hours thirty-one minutes in the book and twenty-four hours thirty in Le Monde. The cave was at minus half a degree, or at 23 degrees Fahrenheit; his body was at 34 degrees Celsius, or 36, or 97 Fahrenheit. Midnight Cave was 100 feet down or 200, fifty miles north of Del Rio or seventy kilometres east, and he was in it 205 days, of which 179 were without time; his wife was twenty, or twenty-five. What the record supports without argument is the date he gave on 14 September 1962, the date the log gave back, the twenty-five days between them, and a man hoisted out of the ice three mornings later like a disjointed marionette. That is enough to learn from.

Sources

• Michel Siffre, Beyond Time, translated by Herma Briffault, McGraw-Hill, New York, 1964, and Chatto and Windus, London, 1965 (the original, Hors du temps, Julliard, Paris, 1963). The 1961 discovery, the planning and funding, the team, the telephone protocol, the diary with its two columns of dates, the Nicolas transcript of 14 September 1962, the arithmetic of 27 and 43, the twenty-four hours and thirty-one minutes, the 1,409 hours, the thermometer at 97 degrees, the cave-ins, the memory lapses and the ascent of 17 September. In copyright; quoted in short phrases from the scan on archive.org, searched page by page. archive.org
• Michel Siffre, “Six Months Alone in a Cave,” National Geographic 147 (3), 426–435, March 1975. The Midnight Cave diary by day and cycle: the platform and tent, the probe and the Beckman electrodes, Malcolm Smith and the Apollo 16 meals, the crack-up of 2 May, the thought of suicide, “J’en ai marre”, the ten-cycle rest, Mus, the call of 10 August and the belief that it was mid-July, the 205th day, the two 48-hour occasions, the 28-hour average cycle, and the memory and eyesight three years on. In copyright; quoted briefly. archive.org
• Joshua Foer, “Caveman: An Interview with Michel Siffre,” Cabinet 30, Summer 2008. “Live like an animal”, the count to 120, the twenty-four hours and thirty minutes, the 48-hour cycles of the other subjects, Senni’s thirty-three hours, NASA’s payment for the 1962 analysis, the two reasons for Texas, the 100,000 dollars, the unanalysed data and “I don’t make theories”. In copyright; quoted briefly. cabinetmagazine.org
• Charlie Buffet, “Michel Siffre et son horloge de chair,” Le Monde, 22 July 2004, as reproduced by the Fédération française de spéléologie’s press archive. Bourcart’s “jeune Darwin”, the hour the explorers could stand in 1961, −0.5° and 98 per cent, the watch handed to Canova, the ladder, the stove, the dysentery, the Luis Mariano record, the 12-second rockfall and the thermometer at 36°, the twenty-minute call, the twenty-five days, the CRS chart at 24 h 30, the two faints and the stretcher, the Apollo rations, Munich, and Clamouse. The French passages from Hors du temps quoted in the entry are given in my translation. web.archive.org
• Agency reports, Reuters unless noted, printed in the Press (Christchurch) and the Evening Post (Wellington): “To Stay In Cave As Experiment,” 17 July 1962; “Record Stay Beneath The Ground,” 18 September 1962; “Stayed Underground For Two Months,” 19 September 1962; “Underground For Two Months,” 24 September 1962; “Isolation Has Them Back At Xmas,” 2 January 1965; “3 Months Alone In Cave,” 15 March 1965; “Pothole Champ—But Does Not Know It,” 18 March 1965; “48-Hour Day,” 25 October 1968; “Out Of Touch,” 26 December 1968; “Potholer sets himself task,” 25 October 1971; “Stay In Cave,” 26 June 1972; “180 Days In Cave,” 15 August 1972; and the Associated Press’s “Wife visits husband in cave home,” 30 August 1972. Papers Past, National Library of New Zealand. 1962, 1965, 1968, 1971, 1972
• Godfrey Anderson, Associated Press, Del Rio, 14 February 1972, “Plans to live six months in cave,” and the agency’s “‘Caveman’ will emerge,” 5 September 1972, both in the Hope Star (Hope, Arkansas). The three months’ delay, 1 p.m. on St Valentine’s Day, the 34–36 and 12–14 hour hypothesis, the four comrades and the log of estimates against real time, 20 degrees Centigrade, the Defence Ministry and NASA in Houston, “barely enough to eat”, the San Antonio generators, the exit at 11 a.m., the backers, Marriott and Apollo 16, the companions, and the tally of recordings and tests. 14 February 1972, 5 September 1972
• United Press International, Nice, 18 September 1962, in Diario las Américas (Miami), 19 September 1962, page 5. The hood, the two friends who carried him to the helicopter, the flight to Paris, and the sixty-two days. Library of Congress, Chronicling America. loc.gov
• Spelunca, bulletin of the Fédération française de spéléologie, 4th series: no. 3, July–September 1962, page 57 (“2 mois sous terre”, 1,500 hours); no. 2, April–June 1963, pages 31–38, “L’expérience Siffre” (the press negotiation, “aucune subvention”, the helicopter and sixteen CRS, 16 July at 14 h, the ascent at 7 h 30 on 17 September, the wet tent); no. 4, 1963, page 74 (the record of 62 days, after Le Figaro). Gallica; read through Gallica’s search-in-document service, the page images being behind a bot check; the French is given in my translation. 1962, 1963
• Michel Siffre, note presented by Jacques Bourcart, Comptes rendus hebdomadaires des séances de l’Académie des sciences 255, July–August 1962, 350–351, on dye tracing in the gouffre des Perdus, Marguareis. Gallica. gallica.bnf.fr
• F. Halberg, M. Siffre, M. Engeli, D. Hillman and A. Reinberg, “Étude en libre-cours des rythmes circadiens du pouls, de l’alternance veille-sommeil et de l’estimation du temps pendant les deux mois de séjour souterrain d’un homme adulte jeune,” Comptes rendus de l’Académie des sciences 260, 1259–1262, 25 January 1965. The first analysis of the 1962 logs; cited from its title and from the reviews below, the text not having been reached. pubmed
• Jürgen Aschoff, “Circadian Rhythms in Man,” Science 148, 1427–1432, 11 June 1965. The Andechs bunker, the 25-hour periods, the “85 percent” and the sentence on Siffre and Mills. mechanism.ucsd.edu
• G. H. Workman, “An Experiment in Underground Living,” Memoirs of the Northern Cavern and Mine Research Society 2 (3), 105–116, 1973. The British press on Siffre’s state, the “limit” of two months, Penman’s 64 days, Workman’s 105, and Mills’s remark on the similar cycles of a man who knew the time and one who did not. nmrs.org.uk
• G. Chouvet, J. Mouret, J. Coindet, M. Siffre and M. Jouvet, “Périodicité bicircadienne du cycle veille-sommeil dans des conditions hors du temps. Étude polygraphique,” Electroencephalography and Clinical Neurophysiology 37, 367–380, 1974; translated as NASA TT F-16234, March 1975. The 34-hour waking and 14-hour sleep, from the abstract as printed in NASA’s Aerospace Medicine and Biology bibliography, supplement 143, July 1975. archive.org
• M. Siffre, “Biological Rhythms, Sleep, and Wakefulness in Prolonged Confinement,” Proceedings of the Colloquium on Space and Sea, ESA SP-280, 1988: the seven experiments, five circabidian subjects, 34 to 36 hours awake and 12 to 14 asleep, some within 8 to 15 days. NASA Technical Reports Server, abstract. ntrs.nasa.gov
• Steven H. Strogatz, The Mathematical Structure of the Human Sleep-Wake Cycle, Springer, 1986. The Midnight Cave record: 35 days of regular 26-hour cycles, the skipped bedtime on day 37, the “wild variations” and their return about day 130. archive.org
• Charles A. Czeisler and colleagues, “Stability, Precision, and Near-24-Hour Period of the Human Circadian Pacemaker,” Science 284, 2177–2181, 25 June 1999. The 24.18 hours and the confounding light. pubmed
• 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 at Mammoth Cave, Siffre’s funding and his forty-two physiological days against fifty-eight, Mills and Workman, and the Cold War reading of the caves. europepmc.org
• L. Zuccarelli and colleagues, “Human Physiology During Exposure to the Cave Environment: A Systematic Review With Implications for Aerospace Medicine,” Frontiers in Physiology 10, 2019. The table of the cave studies from Kleitman to Montalbini, Mairetet’s 45-hour cycles, Laures’s cycle, and Le Guen’s 131 days of immune data. europepmc.org
• Laura Spinney, “The man who lost time,” New Scientist, 11 August 2018. Siffre at seventy-nine on his fear after Scarasson, “Two seconds passed, I perceived one”, the crack-up three months in, the sensors ripped off, and the lightning on the cable. In copyright; quoted briefly. archive.org
• Joseph Anthony, “Journey Out of Time,” In Health, March–April 1991. Véronique Le Guen’s 111 days from 10 August 1988, the 74-hour day, her death, Francis Le Guen’s words, and Siffre’s “Oh, no, never”. archive.org
• L’Express, 27 September 1962 (“Du courage utile”), 29 March 1965, 11–17 September 1972 (“Mardi, à Del Rio, Texas: Michel Siffre a quitté les ténèbres de la caverne de Minuit”) and 11–17 February 1974 (the interview: “j’ai fait une grave dépression nerveuse, et il m’a fallu un an pour en sortir”); Paris Match, 13 November 1971 (“Six mois sous terre pour dérégler le cycle de notre sommeil”). Read in snippets on archive.org. 1962, 1974
• Current Science, 8 November 1972; Senior Weekly Reader, 11 October 1972; You and Your World, 27 September 1972. What American schoolchildren were told: the electrodes and the cable, 150 feet, the 27-hour average and the 50-hour days, and that “he thought he had spent much less time underground than he really had”. archive.org
• Carl Kunath, tribute to Tom Harding, 22 March 2002, Texas Speleological Association. Harding, Midnight Cave in 1968, the gate, and the visitors Siffre’s book brought. cavetexas.org
• Clay Risen, “Michel Siffre, 85, Who Studied Mind From the Depths of Caves,” New York Times, 20 September 2024, as syndicated by the Boston Globe: the birth on 3 January 1939, the death in Nice on 25 August 2024, Clamouse’s 76 days and 5 February, and “Time was the only moving thing in which I moved.” Scintilena, 25 August 2024, for the announcement the same day. bostonglobe.com, scintilena.com