Rail delays are not spread evenly across a month. They arrive in clusters, with several bad days accounting for a large share of the disruption passengers experience across a whole season.
A timetable is a set of promises about track
Every train on a shared line occupies a slot, and those slots are planned to fit together with only small gaps. The plan assumes each train behaves roughly as expected.
When one train misses its slot, it cannot simply be inserted later, because the next slot belongs to something else. The dispatcher must decide who waits and by how much.
That decision cascades. A freight train held for a passenger service blocks a junction later, and a regional service held for an intercity misses its connection at the far end.
Slack was removed to add capacity
Recovery time built into a schedule allows a late train to catch up. It also means the track sits empty for part of the day, which is capacity nobody is using.
Pressure to run more services steadily consumes that margin. The timetable becomes achievable on a good day and fragile on any other, which is a rational trade until conditions turn.
A network with little slack has no mechanism for absorbing a small shock, so small shocks become large ones. This is why delays appear to have a threshold rather than a gradient.
Infrastructure work concentrates risk
Maintenance is planned into windows, often at night or at weekends, and the work is scheduled to finish before service resumes. Overruns are not rare and their effects last all day.
Single-track working around a site halves capacity at that point, and everything upstream queues for it. A stretch that normally passes unnoticed becomes the limiting factor for the whole corridor.
Because major works cluster in specific periods, the bad days cluster with them. A calendar of engineering activity predicts much of the pattern before any weather is considered.
Weather acts on the same weak points
Heat causes rails to expand and forces speed restrictions, which lengthens every journey through the affected section. Storms bring debris onto the line and take down overhead wires.
Leaf fall reduces adhesion in autumn, extending braking distances and requiring wider spacing between trains. That directly reduces the number of services a line can carry.
None of these events is unusual. What makes them decisive is that they land on a timetable that has already spent its margin, so the effect appears immediately.
Crew and stock have to be in the right place
A train needs a driver who is qualified for that route and that vehicle type, and duty hours are strictly limited. A driver stranded by an earlier delay cannot simply be replaced.
Rolling stock circulates on fixed diagrams, so a set that arrives late is late for its next working. Cancellations then appear in places with no obvious connection to the original incident.
Reserve crews and spare units exist to break these chains, but holding them idle is expensive. The size of that reserve largely determines how quickly a bad morning becomes a normal afternoon.