I would posit that the null hypothesis should be that uncoerced human sleep patterns and historical sleep patterns are similar to A: modern human sleep patterns

I would posit that the null hypothesis should be that uncoerced human sleep patterns and historical sleep patterns are similar to A: modern human sleep patterns, and B: the sleep patterns of a majority of diurnal mammals and especially those primates most closely related to us: namely "one big stretch of sleep while it is dark a majority of the time when not disturbed by external stimuli or illness". If historical evidence does not do much to disambiguate this from the alternative hypothesis, then the null hypothesis would be the default presumption.