The argument shows up regularly in sleep circles: humans evolved to sleep in two distinct segments—first sleep and second sleep—separated by 1–2 hours of quiet wakefulness. Then industrialization and electric lighting forced us into a single consolidated 8-hour block, and now we're all sleep-deprived. If you're waking at 3 AM, you're not broken; you're remembering your ancestral sleep pattern.
This narrative is appealing. It reframes a common sleep problem as a historical inevitability rather than a personal failure. But the historical evidence is weaker than the story's popularity suggests.
What the historical record actually shows
The primary evidence for segmented sleep comes from preindustrial diary entries and household manuals that mention "first sleep" and "second sleep." Roger Ekirch, a historian, compiled these references in his 2001 paper and subsequent book. The mentions exist—in 18th-century England, in pre-industrial France, in some sleep medicine texts from the era.
But here's the problem: these references are scattered, geographically inconsistent, and often ambiguous about whether they describe a universal practice or an occasional occurrence in specific populations or seasons. A mention of "rising between sleeps to tend the fire" doesn't tell you whether this was normal, rare, or exceptional.
Historians have since pointed out that Ekirch's interpretation may have overstated the prevalence. A 2016 analysis by sleep researcher Until Frohlich and others noted that segmented sleep descriptions cluster in certain literary traditions and may reflect scribal convention rather than actual widespread practice.
The anthropological angle is unclear too
The assumption that pre-industrial humans always slept segmented rests partly on anthropological studies of modern societies without electricity. But these studies show huge variation. Some societies nap heavily during the day and sleep shorter at night. Others consolidate sleep at night. Sleep patterns are shaped by work schedules, climate, social structure, and individual preference—not by some fixed ancestral template.
Why people wake at 3 AM anyway
If segmented sleep isn't ancestral, why do so many people report middle-of-the-night wakings? Several boring explanations fit the data better:
- Circadian biphasic tendency: Some people's circadian systems naturally have a dip around 3–4 hours into sleep. This isn't a second sleep window; it's a normal transition point.
- Sleep fragmentation from stress or hormones: Cortisol, anxiety, sleep apnea, and menopause can all fragment sleep.
- Light and temperature changes: A cooler bedroom or a partner moving can nudge you awake during lighter sleep stages.
- Normal variation: Even consolidated sleepers don't sleep perfectly continuously. Micro-arousals are part of normal sleep.
The real usefulness of the segmented sleep idea
Even if it's not ancestral, the concept has one genuine use: if you wake for 20–30 minutes in the middle of the night, getting up and doing a quiet, dimly lit activity (reading, journaling, stretching) is often better than lying in bed fighting wakefulness. This is sound behavioral advice, and you don't need to believe it was normal in 1650 for it to help you now.
What we actually know about sleep timing
Modern sleep medicine has solid evidence that sleep timing and architecture vary widely by individual and by culture. Some people naturally sleep more fragmented; others consolidate easily. Exposure to artificial light and irregular schedules can disrupt sleep architecture. But the claim that everyone was segmented 300 years ago and should be now isn't well-supported.
The more useful question isn't "Did we used to do this?" but "What sleep pattern works for me right now?"
The bottom line
Segmented sleep may have occurred in some preindustrial populations under specific conditions. But it wasn't a universal human default, and waking at 3 AM doesn't mean you're recovering ancestral wiring. It usually means your current sleep conditions or physiology need adjustment. That's less romantic, but it's also more actionable.
