← Longevity & Supplement Guides
Who decided you need eight hours of sleep?
In the early 1990s, a psychiatrist at the National Institute of Mental Health ran an experiment that reads less like a protocol than a dare. Thomas Wehr took healthy volunteers and gave them fourteen hours of darkness every night, for weeks — the length of a winter night at the latitude of northern Europe, restored to people who had spent their lives under electric light.
Nobody asked them to sleep in any particular way. They were simply put in the dark and left there.
What came back was not a longer version of the night they were used to. It was a different shape.
A night that split itself in two
In the published account, fifteen people were recorded under both conditions — short nights of eight hours in the dark, and long nights of fourteen. On short nights they slept 7.6 hours, give or take twenty-four minutes, in a single block: the night a modern adult would recognise. On long nights they slept 10.6 hours — and the sleep usually broke into two roughly symmetrical bouts, several hours each, separated by an interval of quiet wakefulness lasting one to three hours.
The waking was not distress. It sat in the middle of the night like a landing on a staircase. And the internal clockwork moved with it: the nocturnal windows of melatonin and prolactin secretion stretched to fill the longer night, as they do in animals moved from summer to winter day lengths.
Wehr's conclusion was modest and, in hindsight, seismic. Human sleep can be unconsolidated. The single uninterrupted block is one setting, not the machine's only gear.
Four nights, one axis
What a night looks like when nobody is grading it
Long nights in the lab — 14 h of darkness Wehr, NIMH, 15 adults. Total sleep 10.6 h, in two bouts separated by 1–3 h of quiet wakefulness.
Short nights in the lab — 8 h of darkness The same 15 adults. Total sleep 7.6 h, one consolidated block.
Three societies living without electric light Yetish et al., 2015 — Hadza, San and Tsimane. Sleep duration 5.7–7.1 h, not interrupted by extended waking; onset 3.3 h after sunset.
The slogan — 8 h Robert Owen, 1817. A demand about the length of the working day. Never a measurement of anyone's sleep.
A historian found the same shape in four centuries of paper
While Wehr was darkening rooms in Maryland, the historian A. Roger Ekirch was reading his way through early modern England and finding a phrase he could not place: first sleep. Then its partner: second sleep. Not metaphors. Ordinary usage, of the kind that needs no explanation because everyone already knows what it means.
His 2001 essay in The American Historical Review laid out the pattern: people retired somewhere around nine or ten, woke from a first sleep near midnight, stayed up for an hour or so, and took a second sleep until dawn. The interval had contents. People prayed in it, talked, tended fires, visited neighbours, conceived children, thought about their dreams.
The detail that has always convinced me most is the dullest one. Physicians of the period wrote dosing instructions timed to it — take this after your first sleep — which is an early, crude form of chronopharmacology. You do not build a schedule around a moment that does not reliably arrive.
Two researchers, working in different centuries' worth of material and unaware of each other's evidence, had produced the same silhouette. That convergence is why the story travelled.
Then another historian took the evidence apart
This is where most retellings stop, and it is exactly where the story gets interesting.
In 2023, Niall Boyce published a reconsideration in Medical History that went at the segmented-sleep thesis on its own ground. He searched the early modern English corpus for "first sleep" himself and argued the phrase carried several meanings at once: an allotted shift of sleep, the first stretch of a broken and unhappy night, or simply sleeping early. Of the "second sleep" references he examined, only a minority explicitly paired both terms. Some of the famous nocturnal activity, he pointed out, was compulsory work or crisis, not a customary interval. His conclusion was careful: segmented sleep happened, but it was not necessarily the predominant model.
Ekirch replied in the same journal in 2024, and did not concede the ground. Boyce, he wrote, had argued against a paper from 2001 while largely ignoring two decades of subsequent work, including a body of allusions now numbering in the thousands and a public archive of more than two hundred previously unpublished excerpts. He pointed to at least eighteen seventeenth- and eighteenth-century medical texts that describe the interval after a first sleep as a known, schedulable moment.
So: unresolved, in public, between two serious people. I am not going to pretend otherwise, because the honest version of this story is better than the tidy one. What survives the argument intact is narrower and sturdier than the internet's version: a divided night was a familiar, unremarkable thing for centuries. Whether it was the norm is still being fought over.
The people who sleep without electric light don't sleep the way the story needs them to
There is a second receipt, and it cuts against the romance.
In 2015, Gandhi Yetish, Jerome Siegel and colleagues put actigraphs on adults in three preindustrial societies — the Hadza of Tanzania, the San of Namibia, the Tsimane of Bolivia — people whose nights are not organised by electricity. If long, segmented, luxuriant sleep were the human default that modernity stole, this is where it should have shown up.
It did not. Sleep periods ran 6.9 to 8.5 hours; actual sleep, 5.7 to 7.1 hours — at the low end of what industrialised populations report. Sleep did not begin at sunset but about 3.3 hours after it. It was not interrupted by extended periods of waking. Napping happened on fewer than 7% of winter days. And the variable that tracked sleep best was not light at all: people fell asleep as the ambient temperature fell and woke near its low point.
Which tells you something more useful than either myth. There is no single ancestral night waiting to be recovered. What the dark gives you depends on how long it is — and, apparently, on how cold. (The same measurement that governs the room you sleep in has its own tangled history: the story of 37 °C is a lesson in how a range becomes a digit.)
The number came from a factory, not a laboratory
So where did eight come from?
Not from a sleep laboratory — there were none. It came from a mill.
Robert Owen ran the cotton works at New Lanark in Scotland, in an era when fourteen- and sixteen-hour days were ordinary and children worked them. By 1817 he was campaigning under a formula built for a banner: eight hours labour, eight hours recreation, eight hours rest.
Read it again as arithmetic. Twenty-four divided by three. The eight that belongs to sleep is not an observation about human physiology; it is the remainder after a demand about work and a demand about leisure have taken their equal shares. Owen was right about almost everything that mattered, and the number was never the point. It was the shape of fairness, expressed in the only unit a factory owner and a factory hand could argue about.
The slogan lost for most of a century, then won. It became the eight-hour movement, then law. And somewhere in the winning, its third clause quietly changed jobs.
How a demand about work became a rule about health
A slogan that wins stops being read as a slogan. It becomes the shape of the ordinary day, and then the shape of the ordinary day becomes the standard against which people measure themselves.
The consolidation of the night ran on the same track. Ekirch's later work places the shift from divided to continuous sleep in the nineteenth century, later and messier than he first supposed: better lighting, later evenings, shift work, and the growing conviction that a night interrupted was a night that had failed.
That last conviction is the expensive one. Once eight consolidated hours is the definition of a good night, every deviation becomes evidence of a problem — and there are now several screens willing to agree with you before breakfast. The number outlived the argument that produced it, which is the usual fate of numbers. Ours is a culture that will happily buy a device to be told it is failing at something a nineteenth-century trade unionist invented on a picket line.
What the sleep societies actually recommend
Here is the part most people get wrong in the other direction.
In 2015 the American Academy of Sleep Medicine and the Sleep Research Society convened a fifteen-member expert panel, reviewed 5,314 papers, and voted through a formal consensus process on the sleep duration that promotes optimal health in adults aged 18 to 60. Their recommendation: seven or more hours per night on a regular basis.
Not eight. Seven or more — a floor, not a target. Regularly sleeping less than seven hours is associated with a long list of adverse outcomes. And above nine hours the panel did something rarer than a recommendation: it declined to make one, stating plainly that for most people it is uncertain whether sleeping more than nine hours regularly carries risk.
Sit with that. The most authoritative statement on how much adults should sleep sets a minimum, brackets an upper region as unknown, and never names eight at all.
The largest study of sleep and survival did not find eight either
In 2002, Daniel Kripke and colleagues published an analysis of more than 1.1 million American adults followed in the American Cancer Society's second Cancer Prevention Study. The paper opens by naming the belief it is testing: that people complain of insufficient sleep because they think they need eight hours.
Best survival was found among those who reported sleeping seven hours a night. Those reporting eight or more had a significantly higher mortality hazard, as did those reporting six or less, with the excess exceeding 15% at the extremes.
Now the caveats, because they matter more than the headline. Sleep was self-reported, once, in 1982. This is an observational association, not a demonstration that long sleep causes anything — people who sleep a great deal are frequently sleeping a great deal because something else is going on, and no amount of statistical adjustment fully separates cause from consequence. The honest reading is not "sleep less." It is that the largest dataset anyone has assembled on this question puts its optimum at seven, and the eight-hour rule has never had a study behind it — not then, and not since.
What happens when people actually sleep more
The best modern test of sleeping longer is not about mortality at all. In 2022, Esra Tasali's group at Chicago published a randomised trial in JAMA Internal Medicine: 80 adults aged 21 to 40, carrying some extra weight, habitually sleeping under 6.5 hours a night. Half received a single individualised counselling session aimed at extending time in bed; the rest carried on as normal. Everyone stayed in their own homes, with no prescribed diet and no prescribed exercise.
The extension worked: about 1.2 additional hours of sleep per night. And energy intake — measured properly, by doubly labelled water rather than a food diary — fell by 270 kilocalories a day in the extended group relative to controls, with no change in total energy expenditure.
Two things are worth taking from that. The first is that a chronically short night is genuinely worth lengthening, and that one conversation was enough to move it. The second is quieter: the trial recruited people below 6.5 hours and moved them toward the floor. Nobody has randomised well-slept adults to seven versus eight hours for years on end, because that trial is close to impossible to run. The eight-hour rule has never been tested. It has only been repeated.
One number, four different jobs
What “eight hours” has actually meant
A demand about work
Robert Owen's formula divided the day into three equal parts. The rest-third was arithmetic, not evidence.
A population floor
AASM and the Sleep Research Society: seven or more hours for adults 18–60. A minimum, with above-nine left explicitly uncertain.
An observed optimum
Best survival at seven self-reported hours across 1.1 million adults. Observational, single questionnaire, and the direction of cause never settled.
A range you find
The only version of the number that was ever about you — and the only one you can measure.
If you wake in the middle of the night
This is the part of the history that pays rent.
A waking in the small hours is one of the most reliably catastrophised events in modern life. The archive and the dark-room experiment both suggest it is, at minimum, an ordinary thing a human night can do — particularly a long one — and that for centuries nobody thought it needed a name.
What the history cannot tell you is that your own waking is harmless, and I am not going to imply it. Persistent trouble sleeping, or waking unrefreshed night after night, is a conversation to have with a clinician rather than a historian. But there is a difference between a symptom and a verdict, and the eight-hour rule has been converting the first into the second for a hundred years.
The practical move is small. If you wake and feel calm, the evidence for lying there quietly is at least as good as the evidence for reacting to it as an emergency. If you wake and immediately start doing arithmetic about how many hours are left, the arithmetic is the problem, and the number it is measuring against was invented at a cotton mill.
Finding your own number
You can do better than inheriting a slogan. Give yourself two weeks where the wake time is fixed and bedtime is allowed to drift earlier, then look at what your sleep settles on when nothing is forcing it. That settled figure — a range, not a digit, and probably a wider one than you expect — is more informative than any population recommendation, because the population recommendation was never about you.
While you are at it, notice what else the night is made of. Regularity outperforms duration in more analyses than most people realise, and a single nightly score hides most of what actually varies. If you track your nights with a wearable, read the stage breakdown as a weekly trend rather than a nightly grade — what these devices can and cannot see is a subject worth an hour of your attention. And the machinery underneath all of it is worth knowing: the circadian system responds to light and timing far more than to effort.
The bottom line
Eight hours is a fine amount of sleep. It is simply not a finding. It entered the world as one-third of a banner about factory hours, acquired medical authority it was never given by evidence, and now sits on a screen every morning telling several hundred million people whether they passed. The sleep societies say seven or more. The largest survival dataset says seven. The dark-room experiment says the night has more than one shape. The people who sleep without electric light sleep less than the story requires. Nobody, anywhere, has produced the trial that would justify the number you are grading yourself against.
So this week, do one thing: stop scoring the night. Hold your wake time steady, let bedtime find its own level for a fortnight, and learn the range your body actually settles into. Then use that — not a nineteenth-century slogan — as the thing you are trying to protect.
