Eight hours a night. It’s printed on wellness app dashboards, pediatric handouts, and the side of the melatonin bottle in the medicine cabinet. It’s also, according to a sweeping new review published in the journal Brain Medicine, a number that traces back to nineteenth-century arithmetic rather than actual measurement — and one that field data from modern hunter-gatherer societies doesn’t support as any kind of ancestral baseline.

The review, led by Dr. Seithikurippu R. Pandi-Perumal of the Central Research Facility at Dr D.Y. Patil Medical College and Hospital in Pune and the Centre for Research and Development at Chandigarh University in India, draws together anthropological fieldwork, archival history, and comparative genomics — eight authors working across India, Canada, Saudi Arabia, Bahrain, the United Kingdom, and the United States. Rather than running a new experiment, the team set out to place a fractious, decades-old literature side by side and ask what it adds up to collectively.

A number nobody actually measured

The review traces the eight-hour figure to a tidy mid-nineteenth-century formula — eight hours of work, eight hours of rest, eight hours of sleep — that took hold as social convention long before anyone systematically studied how people actually slept. Even an early attempt at real measurement doesn’t hold up under scrutiny: a 1913 Stanford survey sometimes cited as evidence that people once slept nine hours a night turns out, on closer inspection, to have surveyed children between the ages of eight and seventeen, not adults.

Against that backdrop, the review turns to field data from three hunter-gatherer societies living without electric light — the Hadza of Tanzania, the San of Namibia, and the Tsimane of Bolivia — whose recorded nightly sleep runs between 6.4 and 7.1 hours. That’s not longer than the industrialized average. In several comparisons, it’s shorter.

“The literature keeps asking whether we sleep less than our ancestors did, and we think that is the wrong question,” Pandi-Perumal said. “Put the field data next to the archival record and next to the genomics, and what separates industrial sleep from pre-industrial sleep is when it happens and how reliably it repeats. Not how many hours it fills.”

What actually changed with industrialization

Sleep in these three hunter-gatherer populations tracks the local light-dark cycle closely and stays consolidated into a single block — the review finds no consistent evidence of the segmented, two-part sleep pattern often assumed to be the ancestral norm. What industrialization changed, the authors argue, comes down to four things: sleep moved later in the day, day-to-day regularity broke down, alignment with natural light weakened, and diagnosable sleep disorders became more common. Total sleep duration — the one number most people fixate on — moved the least.

That reassessment extends to a well-known historical claim. In At Day’s Close: Night in Times Past, historian Roger Ekirch argued, based on court records, letters, and literary sources, that pre-industrial Europeans slept in two segments separated by an hour or two of wakefulness. The review doesn’t dismiss that claim outright but complicates it substantially: a more recent reexamination of the same historical sources suggests the phrase “first sleep” carried multiple meanings in early modern writing, not necessarily a fixed two-part sleep architecture. A study of Roman domestic life found little archaeological or literary support for standardized segmented sleep in that culture, either. And the most frequently cited experimental support for segmented sleep — a study in which volunteers spent fourteen hours in enforced darkness with activity prohibited — describes conditions the review calls very unlike a candlelit, fire-warmed pre-industrial evening.

A gene that argues with the guideline

If history complicates the eight-hour standard from one side, genetics complicates it from the other. The review traces DEC2 (also called BHLHE41), a gene first identified in a family whose members slept roughly six hours a night with no signs of sleep deprivation. When researchers introduced the same mutation into mice, the animals slept less across both REM and non-REM sleep, with no detectable behavioral, motor, or cognitive impairment.

“Clinically, the most immediate consequence is that we should stop treating every short sleeper as a patient,” said Dr. Ahmed S. BaHammam, a co-author based at the University Sleep Disorders Center at King Saud University in Riyadh. “Where a validated variant such as DEC2 P384R is driving short sleep in a family, investigation and correction do more harm than good.”

A second thread, hypocretin (also known as orexin), runs even further through the review’s argument. Loss of hypocretin-producing neurons causes narcolepsy with cataplexy in humans, a condition also seen in dogs with related genetic mutations and in orexin-deficient mice. The DEC2 short-sleep variant boosts orexin expression in mice, and cave-dwelling populations of the Mexican tetra fish — which independently lost the need for sleep in separate caves — show elevated hypocretin signaling relative to their surface-dwelling relatives.

“Hypocretin turns out to be the thread,” said corresponding author Dr. Randy J. Nelson, of the Department of Neuroscience at West Virginia University. “It runs through narcolepsy in people and in dogs, through short sleep in mice carrying a human variant, and through fish that gave up sleeping in caves on separate occasions. A gate that conserved is telling you something about how wakefulness is built.”

Sleep as an equity issue

The review’s final turn is toward the built environment. It cites evidence that neighborhood-level socioeconomic disadvantage predicts worse sleep efficiency and more nighttime wakefulness even after individual factors are controlled for, and that nighttime light and traffic noise run consistently higher in lower-income and racially segregated U.S. neighborhoods. Roughly one-fifth of workers in industrialized countries work night shifts, and climate projections cited in the review suggest that under a high-emissions scenario, warming could add six to fourteen additional nights of insufficient sleep per hundred people annually in the U.S. by mid-to-late century, with the burden falling hardest on lower-income populations.

“Sleep is not distributed the way people assume,” said co-author Dr. Haitham Jahrami of the Psychiatry Hospital, Government Hospitals, in Manama, Bahrain. “Noise, light, heat, and housing quality fall hardest on the people least able to shield themselves, and every one of those exposures is tractable. Building codes and lighting standards are sleep interventions, even though nobody files them under that heading.”

The authors are explicit about the limits of their approach: this is a narrative review, not a systematic one, built around sources the authors judged useful for bridging large-scale and small-scale evidence, which means other methodologically sound studies were left out. One question, they note, remains entirely untested by the existing literature: whether split, two-part sleep performs as well, healthwise, as consolidated sleep of the same total duration. Nobody has run that comparison. For a species that may have slept in more than one pattern for most of its history, the authors call that a conspicuous gap.

Their practical recommendation is narrower than a new number to replace eight hours. Sleep assessment, they argue, should move past duration as the primary metric and start routinely incorporating timing, regularity, and circadian alignment — the dimensions modern life disturbs most, and the ones the current standard largely ignores.

---

Endnotes

• Genomic Press, “Hunter-gatherers sleep no longer than we do: review suggests industrialization changed sleep timing more than duration,” news release, August 25, 2026, https://www.eurekalert.org/news-releases/1140067.
• Seithikurippu R. Pandi-Perumal et al., “The new science of sleep: Insights from pre-industrial societies and phylogenomics,” Brain Medicine, 2026, https://doi.org/10.61373/bm026w.0060.