The Rest Day
Is the Weapon

Both sides of this argument agree on the facts in this section. They disagree entirely about what the facts mean.

A published timetable requires a certain number of driver turns — one driver, one shift, one set of services. To run every advertised train every day, an operator needs enough contracted drivers to cover every turn while absorbing holiday, sickness, medical restrictions, refresher training and the long absences involved in learning a new fleet.

Almost no British operator employs that many. A driver costs salary, pension, and a training pipeline running twelve to eighteen months before they can take a train out alone. So operators run lean and buy the gap back shift by shift, as overtime, from drivers volunteering on their rostered days off. The industry calls it rest day working. Before the August settlement, on several operators, Avanti included, Sunday was not part of the standard working week at all: the Sunday service was not a rota, it was a sign-up sheet. Public reporting does not disclose whether the new agreement moved Sunday into the contracted week.

You can watch the price being set. In March 2024 Avanti agreed to pay £600 for a rest day turn, replacing an arrangement worth roughly £420 to £495. Its stated reason was that the new Hitachi fleet needed around 2,500 training days, each pulling a driver off passenger services for a fortnight, and the overtime was what would let it train drivers without cancelling trains. That agreement ran twelve months. Everything at Euston this summer is what happens when a deal like that comes up for renewal.

And now the legal part, which almost everyone gets wrong, including this publication in an earlier draft.

The comfortable assumption is that declining voluntary overtime sits outside the law entirely — not a strike, therefore not industrial action, therefore untouchable. That is false. A concerted refusal of voluntary overtime, undertaken to pressure an employer, can be industrial action: the Court of Appeal said so in Power Packing Casemakers v Faust in 1983, and the Local Government Association tells employers the same thing in plainer words.

What the refusal is not is a breach of contract — because there was never a contractual obligation to breach. That single distinction is the whole mechanism, and it has nothing to do with ballots:

• The union induces nobody to break a contract, so it commits none of the economic torts a strike commits.
• With no tort, it needs none of the statutory immunities that protect a strike.
• With no immunity in play, the employer has nothing to sue on and nothing to injunct.
• With no contractual duty breached, the employer cannot withhold basic pay.

Which is why ASLEF balloted at Avanti for strike action and action short of a strike, and never needed either. The pressure arrived before the ballot did.

Two more agreed facts, because they cut in opposite directions and both sides need them.

The state was thinning the same timetable. On 18 May the Department for Transport asked Avanti to cut costs. The operator removed 38 weekday services — one train in seven on its busiest routes — from 20 July to 28 August. The summer's disruption had two causes running concurrently, and they are not separable from published data.

And the number that would settle it does not exist. Avanti runs on a government contract, funded by the taxpayer. Neither the operator nor the Department publishes how many drivers it employs against how many its timetable requires. The size of that gap is the single most useful fact about whether your train will run, and it is not in the annual report, the contract, or any regulatory return a passenger can read.

But what the mechanism means — that is where the agreement ends.

One: leverage with no market test. In a competitive firm, a union's demands are disciplined by the possibility that the firm dies. In the NHS, a school system or a statutory monopoly railway, the employer cannot go bankrupt, the service cannot be substituted, and the people bearing the cost of a dispute are not shareholders. Pressure is applied to a third party in order to move a fourth.

Two: the third party is poorer than everyone else in the room. The public sector pay bill is roughly £240–260 billion a year, so one percentage point costs about £2.5 billion. If an award is unfunded, a department must do less — freeze hiring, let the waiting list grow — or raid the capital budget and cancel the scanner.

Three: the pension gap nobody puts in the comparison. Employer contributions run at 23.7% of pensionable pay in the NHS, against a median private employer contribution of 4% to 6% into a pot where the employee carries the investment and longevity risk. An unfunded public scheme's employer rate is an actuarial construct rather than a cash transfer, so the comparison is imperfect. The gap is still large and routinely omitted.

Four: practices that outlive their reason. Sunday outside the standard working week converts a seventh of the service into a negotiation. Route-knowledge rules can stop an operator moving a qualified driver to an adjacent route without refresher training. Each was rational when agreed; some now impose a fixed cost management cannot reorganise around without buying out — precisely what the £600, then £720, did.

Five: and the law is moving. The Employment Rights Act 2025 is being commenced in tranches through 2026: minimum service levels repealed, the 40% support threshold in important public services abolished, strike notice cut from 14 days to 10, mandates extended from six months to twelve, and most of the Trade Union Act 2016 repealed. Stronger statutory workplace-access rights are scheduled for 30 October 2026, subject to parliamentary approval.

The government's economic analysis put the direct cost of the wider package to business at up to £5 billion a year. In November 2024 the Regulatory Policy Committee — the independent body whose job is to scrutinise government impact assessments — red-rated the summary assessment and eight of the twenty-three individual ones, for failing to establish the market failure being corrected, to assess non-regulatory alternatives, or to model how compliance costs would pass through into prices and wages.

The government could not persuade its own scrutiny body that it knew what these changes would cost.

On the rest day specifically, the case is this: a workforce can disrupt a national timetable before conventional strike action is used, while the absence of an individual duty to volunteer limits the employer's ordinary contractual remedies. The pressure still lands on passengers who had no part in the dispute and no easy way to route around it.

But both of those cases have to be extracted from a public argument conducted almost entirely in numbers that do not survive being checked.

What each side gets right. The case for is right that Britain has legislated rights with limited enforcement capacity, and that a workplace representative can be the only enforcement mechanism many people encounter. The case against is right that in a non-market service, industrial pressure lands on a third party who is poorer than either negotiating party and represented by nobody — and that the pension gap is real and routinely hidden.

What this publication got wrong while reporting it. An earlier draft of this piece stated that declining voluntary overtime is not industrial action in any legal sense. That is false, and the correction is in section 02. The protection comes from contract law, not from invisibility.

Now the answer. Strip out the six claims in section 05 and here is what is left on the West Coast Main Line.

The operator does not employ enough drivers to run the timetable it advertises. The union noticed this makes the operator permanently dependent on goodwill, and priced the goodwill. The operator could not settle without the Department for Transport, because the Department holds the money — and the same Department was, that summer, instructing it to cut one train in seven to save some. The Department settled because cancellations cost more politically than the settlement cost fiscally. The passenger absorbed the difference.

Every party behaved rationally. Nobody lied. And the leverage that produced the outcome was not built by a union.

Union strength in Britain no longer comes from numbers — at 22.4% density and 4.3% among the under-25s, it cannot. It comes from position: from sitting at a point in a system optimised until there is no slack left. Under-staff a hospital and the bank shift becomes leverage. Under-staff a prison and the voluntary overtime becomes leverage. Run a railway with no spare drivers and the rest day becomes leverage. Every one of those positions was created by a decision to run a public service at the edge of its capacity, and every one of them will last exactly as long as the slack stays gone.

Which is why “are unions strangling the economy?” is the wrong question, and why it flatters both sides. It lets one blame 6.6 million people, most of them teachers and nurses, for a productivity failure that is overwhelmingly about capital, planning and energy costs. And it lets the other avoid the fact that a movement recruiting 4.5% of its members from the under-25s, in an economy whose growing sectors it cannot organise, has about one generation left to solve a problem it is not solving.