I seem to recall hearing about some study done years back where they put volunteers in a closed environment with NOTHING that would give a time signal for about a month. The idea was to observe what they would do for sleep routines.

As I recollect the results showed that the subjects seemed to settle on an approx 25 - 26 hour per day cycle, including about 9 hours of sleep.

This sort of makes sense in terms of the way timing oscillators work in electronics, it is easier to make an oscillator that runs a little slower and use a feedback circuit to make it trip a little early each cycle than it is to make one that runs an exact frequency w/o feedback. In a human, the 'feedback trigger' would presumably come in the form of day/night changes.

It was an interesting concept and it seems sort of relevant to this.

Gooserider