This theory comes from the work of Roger Ekrich, a professor of history at Virginia Polytechnic. In 2001, he published a paper in the American Historical Review titled Sleep we have lost: pre-industrial slumber in the British Isles, in which he provided a lot of references from historical sources about segmented sleep. In 2005, he expanded this work into a book called At Day’s Close: Night in Times Past, published by W.W. Norton & Co. This book received a lot of publicity, with Ekrich himself penning articles and Op-Eds in various media sources, including the NYT.

His idea was that people used to sleep in two segments, which in his sources are referred to as “first sleep” and “second sleep”. First sleep apparently began soon after dusk, after people got home from the day’s work, ate dinner, did chores and then went to bed. They would then wake up around midnight for a brief stretch, from a few minutes to an hour or more, and then resume their “second sleep” from which they woke up in the morning. The period of wakefulness in the night might be relatively inactive (going to the bathroom, getting a snack, lying awake a while) or might be work-related (going to the barn to check on the animals), or it might be social (chatting with one’s spouse or children).

Ekrich offers a lot of evidence for the idea of segmented sleep from historical sources. It also appeals to “common sense” from the perspective that the hours of darkness were long, most people couldn’t afford candles, and there’s not much that humans can do after dark other than to sleep. Sleep phases are a well-known phenomenon, and most of us have experience of waking up in the middle of the night, only to return to sleep because it’s not morning yet. So it’s not a particularly outlandish or hard to believe suggestion that segmented sleep was a frequent occurrence in the past.

The problems are in the details. Historical records are very sparse in relation to the long expanse of history. It’s difficult to generalize from “here’s someone talking about first and second sleep in this document from medieval times” to “this is how people slept back in those days”. Almost all of his references are from the British Isles, and by the nature of historical records, over-represent the wealthy and educated populace rather than the average person, who wasn’t keeping any diaries.

And while he provides enough references to show that it was not the rare isolated case, he doesn’t provide enough to do any meaningful statistics. Were people doing this seasonally? After all, the length of the day and night varies significantly over the course of the year. At the latitude of London, the period from twilight to daybreak is as short as 7.5 hours at summer solstice and as long as 16.5 hours on winter solstice. And London is in southern England, the situation is even more extreme the farther north you go in the British Isles.

Should we assume that sleep patterns remained unchanged over the course of the year while darkness varied from 7.5 to 16.5 hours? And what about other locations in the world? Ekrich’s research is confined to the British Isles, but what about continental Europe? What about the tropics or near the equator? Seasonal changes are much less pronounced the closer you go to the equator, would this have an effect on sleep cycles? What about the amount of sunlight? We know that sleep cycles vary based on the levels of several hormones, which are affected by sunlight exposure. The dim light or clouded skies of northern climates are not equivalent to the bright sun of the tropics, so we can’t draw any equivalence between an hour of sunlight at different latitudes from physiological considerations alone.

I am not aware of any serious research on this subject by other people looking at other locations. Ekrich did an excellent job looking at British records, but we need other people to spend a lot of time digging up German or Spanish or Italian records. Or for that matter, China or India. Until we have representative samples from across the world, we can’t really say anything about human physiology as such, only about particular conditions on the British Isles.

And even the case for the British Isles isn’t entirely convincing. I think he provides enough evidence to show that segmented sleep was clearly not an unknown thing. But how common was it exactly, and how did it vary by social class, lifestyle or season? I don’t have much of a problem believing that 12 hours of darkness is too much for a single stretch of sleep. Probably people did wake up in the middle of the night and go back to sleep again. But would the same people do that in summer when there were only 7.5 hours of darkness? I’d say he offers an interesting idea for further research, but I’m not ready to buy it as a pre-industrial norm just yet.