Somewhere in the first three weeks of September, a familiar conversation happens. A child who was fine in August is now short-tempered at five o’clock, tearful over homework, or arguing about things that were not arguments in July. The parent checks in with the school and hears that everything seems fine — settling in well, no concerns. So the difficulty gets read as a signal: something must be happening that nobody is seeing. It is a reasonable inference, and it is worth examining, because there is a less alarming explanation with a good deal of evidence behind it.

The Assumption: Difficult Behaviour Reports on a Difficult Day

The belief underneath the worry is that behaviour is informational — that if a child is struggling in the evening, the struggle is about something, and that something most likely happened between nine and three.

It is a sensible starting point and it is sometimes right. But it treats the child’s nervous system as a passive reporter, relaying the emotional content of the day. There is another possibility, which is that the system is running on less than it had in August, and that this alone changes what comes out of it — independently of whether anything went wrong.

That possibility is testable, and it has been tested.

The Test: How Little Sleep Loss It Takes to Show

In 2012 a group at McGill ran a study designed to answer a narrow question: does a modest change in sleep duration produce a change in behaviour that someone else can see?

They recruited 34 typically developing children aged 7 to 11 — importantly, children with no reported sleep problems and no behavioural, medical or academic concerns. In a randomised parallel-group design, one group had an hour added to their weeknight sleep and another had an hour removed, relative to their own baseline. Behaviour was then rated at school on the Conners’ Global Index by teachers who did not know which group any child was in.

The blinding is what makes the result worth quoting. A cumulative sleep extension of 27.36 minutes was associated with detectable improvement in emotional lability and restless-impulsive scores, along with reduced daytime sleepiness. A cumulative restriction of 54.04 minutes was associated with detectable deterioration on the same measures (Gruber et al., Pediatrics, 2012).

Read that again in terms of ordinary life. Not an hour a night for a month. A cumulative difference measured in tens of minutes was enough for a stranger, with no idea what had been changed, to notice it in the classroom.

The Mechanism: A Schedule Change Is a Physiological Event

September moves the clock. Wake times come forward, often by an hour or more, and they come forward abruptly rather than gradually — the summer schedule ends on a Sunday and the school schedule starts on a Monday.

That the two are causally linked is not an assumption. A systematic review of the experimental evidence found that when schools delayed their start times by 25 to 60 minutes, total sleep time increased correspondingly, by 25 to 77 minutes per weeknight (Minges & Redeker, Sleep Medicine Reviews, 2015). Six studies met the inclusion criteria, two of them randomised controlled trials.

One caution about how that finding is being used here. Those studies moved start times later and measured sleep going up; September moves them earlier. Running the arrow backwards is a reasonable inference, not a demonstrated result, and the review’s studies were in adolescents rather than younger children, whose circadian timing differs. The honest version is narrower than the intuitive one: school schedules and sleep duration are experimentally linked, and September is a schedule change.

Sleep is not a passive gap in the day — it is an active period with its own architecture and its own work to do, which is why subtracting from it has effects well beyond feeling tired.

Why It Surfaces at Home Rather Than at School

If the school reports that all is well, that report is probably accurate. School is a highly structured environment: the day is scaffolded, the expectations are explicit, and there is very little unallocated time in which a difficult moment can develop.

Home in the late afternoon is close to the opposite — less structure, fewer external cues, and the people a child is most secure with. A nervous system running a deficit will express that where the demand for control is lowest, which is not a statement about the quality of either environment.

This is worth holding lightly. It is a plausible reading of a real pattern, not something the studies above measured. What they measured is narrower and firmer: modest sleep loss changes observable behaviour in children who have no other difficulties.

What the Evidence Does Not Establish

Four limits, and they matter more than the headline.

The behavioural study is small and deliberately narrow. Thirty-four children, aged 7 to 11, specifically selected for having no sleep, behavioural, medical or academic problems. That selection is a strength for the question asked — it shows the effect in children with nothing else going on — and a limit on generalising to children who do have something else going on.

The school-start evidence is about adolescents, and about delaying. Applying it to a seven-year-old in early September is an extrapolation across both age and direction.

The effect may not last. A South Korean natural experiment tracking 1,133 adolescents from grade 7 through grade 12 found that delaying start times by 30 to 90 minutes was initially associated with a 19-minute increase in sleep duration, with a confidence interval running from 5.52 to 32.04 minutes (Jung & Fenelon, Preventive Medicine, 2024). The word doing the work in that sentence is initially. Effects that appear at the start of a policy do not always persist, and the same caution applies to a September adjustment.

And none of it describes one child. These are group-level findings about sleep timing. They cannot tell any parent why their own September is hard, and a difficult few weeks that does not resolve is worth raising with a professional rather than attributing to the calendar. The value of the research is not diagnostic — it is that it makes a boring explanation credible enough to check first.

What Follows From It

Mostly that the first thing to examine is the least dramatic one, and that it is examinable. Wake time and bedtime are recorded facts, not interpretations, and the September change to them is usually larger than anyone estimates from memory.

It also suggests patience with the adjustment period itself. A schedule that moves abruptly is a demand on a system built for gradual change, and the first weeks of that are not a verdict on the year.

Predictability does real work here, and it does it for reasons that are structural rather than motivational. Where the difficulty persists well past the transition, or looks less like tiredness and more like a change in how a child is processing the ordinary sensory load of a school day, that is a different question with a different answer.