The extensive literature on circadian variation in human performance
contains relatively few industry-based studies of human error throughout
the 24 h in which the opportunity for error was constant at all times. The
work of Bjerner et al. (1955) remains the classic study of circadian error
patterns. They examined tens of thousands of errors made by workers at
two gasworks and a paper mill, where in each case standard tasks were
being performed around the clock. Errors exhibited a strong circadian
pattern with a marked peak at 03:00 h and a smaller peak at 15:00 h.