You have probably heard that humans used to sleep in two shifts — a first sleep, an hour awake around midnight, then a second sleep — and that artificial light destroyed it. It is a wonderful story. It is also, right now, an unsettled argument in a peer-reviewed journal, and the strongest counter-evidence is almost never mentioned.
Here is the whole thing, including the parts that do not fit.
The oldest known bedding is not 77,000 years old. That figure was superseded in 2020.
At Border Cave in South Africa, archaeologists found grass bedding laid on top of layers of ash, dated to around 200,000 years. The grass was identified from microscopic silica skeletons left behind in the sediment; one of the species still grows outside the cave mouth. The journal’s own summary noted that this pushed the record of deliberate bedding back by at least a hundred thousand years.
The 77,000-year site is real and separate — Sibudu, also South Africa, where people laid down centimetre-thick layers of sedges and rushes, at least fifteen times over forty thousand years, and capped them with leaves of a laurel that contains insecticidal compounds. Note the careful wording of the researchers: those leaves would be suitable for repelling mosquitoes. That is a reasonable inference about why. It is not proof of intent.
This finding is real, and better than its popular version.
Researchers put activity monitors on thirty-three Hadza people in Tanzania for twenty days. Across the entire period there were only eighteen minutes in total when every single person was asleep at the same time. At any given moment, roughly forty percent of the group was awake.
The interesting part is why. Nobody was taking shifts. Nobody planned anything. It falls out of the fact that young people are night owls and older people wake early — chronotype varies with age, and a mixed-age group therefore covers the night automatically.
Two honest caveats the popular retelling drops. The measurement was wrist actigraphy, not brain recording, so the paper itself hedges with “awake or in light stages of sleep.” And no guarding behaviour was actually observed — the authors propose that this provided vigilance. It is a good hypothesis, not a demonstration.
The historian Roger Ekirch has assembled a genuinely large body of references — he now claims upwards of two thousand allusions to a “first sleep” and a “second sleep” across European sources, from Roman antiquity to the twentieth century, in several languages.
The single most striking piece of evidence is medical. Francis Bacon instructed that a preparation be taken “between sleeps.” John Aubrey’s jaundice remedy involves urine collected “after the first sleep.” Ekirch calls this an early form of chronopharmacology, and it is hard to explain unless the waking interval was so ordinary that a physician could schedule around it.
But in 2023 a historian published a direct challenge in Medical History, pointing out that no single unambiguous early-modern medical discussion of segmented sleep has been identified — the remedies presuppose a gap without ever describing the pattern as doctrine — and that the existence of the phrase “second sleep” does not by itself prove a universal habit.
Ekirch replied in 2024. He conceded that he had modified his original thesis, that the change came later than he first argued, and that not everybody followed the same schedule. The critic’s response closed the exchange with a sentence worth quoting: the question of segmented sleep in early modern Europe is by no means settled but is a matter of ongoing debate.
That was two years ago. Most articles on this subject are written as though it were decided in 2005.
In the early 1990s Thomas Wehr at the US National Institute of Mental Health put volunteers on ten-hour days — fourteen hours of darkness every night — for weeks.
Their sleep expanded and, in his words, usually divided into two symmetrical bouts with a one-to-three-hour waking interval between them. In the companion paper with fifteen participants, sleep went from 7.6 hours to 10.6, melatonin secretion lengthened, and the duration of high night-time prolactin release expanded from about ten hours to nearly thirteen.
Three things need saying. “Usually” is Wehr’s word, not all subjects. The famous line about the waking interval resembling an altered state of consciousness not unlike meditation is Wehr quoted in Ekirch’s book — it is a characterisation, not a measurement in either paper.
And fourteen hours of enforced darkness every night is not a recreation of pre-industrial conditions at mid-latitudes. Ekirch himself accepts that neither he nor Wehr claimed it was. What the experiment shows is that the human sleep system can split under long nights. That is genuinely important. It is not evidence about what your ancestors did.
In 2015 a team led by researchers at UCLA put activity monitors on ninety-four people across three societies with no electricity at all: the Hadza in Tanzania, the Ju/’hoansi San in Namibia, and the Tsimane in Bolivia. Six to twenty-eight days each, summer and winter.
Their findings, which contradict nearly every part of the popular story:
They slept 5.7 to 7.1 hours a night — average 6.4 — which is at or below the industrial average. Sleep efficiency was 81–86%, similar to industrial populations. They did not go to sleep at dusk; they stayed awake about three and a half hours after dark, with measured night-time light below five lux. They woke about an hour before sunrise. They barely napped.
And the line that matters most: most subjects did not awaken during the nightly sleep period — night-time awakenings occurred on ten nights out of 210.
Their conclusion was that the two-sleep pattern that may have existed in Western Europe is not present in equatorial groups today, and was probably a consequence of longer winter nights at higher latitudes rather than a human default. One of the authors put it bluntly: the idea that modern life cut our sleep below what our ancestors got is a myth.
Ekirch published a rebuttal in Sleep in 2016, calling the latitude explanation highly questionable and citing historical and ethnographic accounts of segmented sleep in equatorial cultures.
Then a separate team studied a non-electric agricultural village in Madagascar — equatorial, which is the crucial control — with twenty-one participants over 292 days. They found activity peaks after midnight on 49% of nights, and concluded that this was evidence contrary to the latitude hypothesis.
That same team also handed out mobile field lights as an experiment, to test whether artificial light collapses segmented sleep. It did not. Sleep segmentation was not influenced by access to the lights, and there were no significant shifts in sleep timing.
So in the one place where somebody directly tested the central causal claim of the whole story, the answer was no.
Even the Hadza data disagree with itself: a different research group measured Hadza sleep efficiency at 69% and found napping on 54% of days, against the first study’s near-zero napping.
The honest summary is this. Extended night waking looks rare in the three groups one team measured and common in the village another team measured. The latitude explanation is contested. The historical thesis is under live peer-reviewed dispute. Two sleeps is a hypothesis with real evidence behind it, not a settled finding.
A popular claim is that the mattress and sleep industry invented the eight-hour norm to sell you things. We looked for a source for that. There is not one.
The real history is better. “Eight hours” comes from the labour movement, not from medicine. Robert Owen had the slogan by 1817: eight hours’ labour, eight hours’ recreation, eight hours’ rest. A political division of the day was later read back as a biological requirement.
The clinical recommendation is not eight either. The consensus figure from the American Academy of Sleep Medicine and the Sleep Research Society is seven or more hours — a floor, with no upper number attached.
And there is a lovely piece of research on the anxiety itself. Reviewing thirty-two sets of sleep recommendations published between 1897 and 2009, researchers found that recommended sleep fell by about 0.71 minutes per year while children’s actual sleep fell by 0.73 minutes per year — so the recommendation sat roughly thirty-seven minutes above reality for the entire century. Every era blamed the technology of its time. Their conclusion: no matter how much sleep children are getting, it has always been assumed they need more.
The reassurance in the popular story is correct. The reasoning is not.
You do not need your ancestors to justify it. A large clinical analysis found that in the absence of other insomnia symptoms, night-time awakenings on their own are unlikely to be associated with daytime impairment. Waking in the night is not by itself a disorder. That is a finding about you, not about the Paleolithic.
Two practical corrections while we are here.
Darkness is not “the strongest sleep drug.” It is the strongest timing signal — the cue that tells your clock it is night, which is what every one of these studies actually measured. Its downstream effect on sleep itself is modest, and melatonin’s own hypnotic effect is small: across nineteen trials, about seven minutes faster to sleep and eight minutes more of it.
Cold is not the mechanism. Sleep begins when your body dumps heat through the skin of your hands and feet, which drops your core temperature. The best predictor of how fast someone falls asleep turned out to be the temperature gradient between their extremities and their trunk — better than core temperature, heart rate or melatonin. That is why a warm bath an hour or two before bed shortens sleep onset, and why one of the key papers is titled “Warm feet promote the rapid onset of sleep.” A cool room and warm feet. Around 20–25 °C is where sleep efficiency peaks; too cold fragments sleep as surely as too hot.
One last detail from the equatorial data, and it is the one that has changed how I think about this. What tracked sleep onset in all three societies was not low temperature. It was falling temperature. They fell asleep as it dropped and woke at the coldest point of the night.
Short detours through the parts of being human that nobody explains.
Subscribe on YouTube