[sooheon]

> Null hypothesis: early humans slept much like today's humans [...]

That's a terrible null hypothesis, as there is no reproducible, controllable test for it, other than a time machine.

A better one might be "people exposed to pre-industrial revolution photoperiods (<12h per day) will settle into a biphasic sleep pattern".

This was empirically tested with a small N (https://onlinelibrary.wiley.com/doi/10.1111/j.1365-2869.1992...). At least it's testable.

[notahacker]

Specifying that as something isn't directly testable, it's better that a bold new theory should actually be the null hypothesis isn't how science works! We've observed a lot of populations with little or no natural light and there is no reason to cherry pick the one or two which actually conform to Ekirch's proposed "standard" form of sleep. I mean, for a start, assuming "pre industrial revolution photoperiods" generally matched the ones in Wehr's study involves pretending that seasons don't exist in most of the places Ekirch's soured material came from!

[axiolite]

A much better study:

https://www.theatlantic.com/science/archive/2015/10/the-many...

[sooheon]

Better in what dimension? It's not a repeatable intervention, it's not controlled, it's observational. It's not a study on the same null hypothesis.

[notahacker]

If your claim concerns common habits of populations with limited or no artificial light sources, observational studies of multiple populations with no artificial light sources is a superior way of assessing it than a prior study which imposed very specific periods of light deprivation on a study sample accustomed to nightly routines governed by available light to see how their behaviour changed.