There is now certainty. Or is there? On October 2, Theresa May set out a timetable for Brexit—the UK’s departure from the EU. The British prime minister has said she will invoke Article 50 of the EU treaty before the end of March 2017. This will start the formal process of negotiation, which can last up to two years. It follows that the UK will have left the EU by March 2019.

That is probably what will happen; but there is a small chance that it won’t. One reason is that the prime minister is being deliberately vague about her strategy, which means she has said nothing about what she will do if the negotiations go badly.

In some ways, May’s caution makes sense. She does not want to show her hand in advance to Brussels, Berlin, Paris, or any other European capital. However, in her speeches to her Conservative Party’s conference earlier in October, she did stress two imperatives: full UK control over who is allowed to settle in Britain; and free access for the UK’s companies, in both manufacturing and services, to the EU’s markets.

What she has not said is what she will do if the other 27 members of the EU tell her she must make a choice: immigration controls or free access to the single market. The early signs are that the choice will be a tough one. On October 6, Angela Merkel, Germany’s chancellor, turned the screw by saying that if the UK abandons the free movement of people, the EU will impose tough conditions on the terms by which the UK trades with the EU. No wonder sterling has fallen sharply in the past few days.

Even that problem, big though it is, is only one element of what could become a major political headache, and even a crisis for May’s government.

The evidence from the UK’s June 23 referendum is clear. The 52 percent majority for Brexit was based on three propositions: outside the EU, Britain would have far fewer immigrants; the UK’s contribution to the EU budget would be spent instead on the National Health Service (NHS); and more well-paid jobs would be available to British workers. It was the prospect of these things coming to pass that converted the long-standing anti-EU passion of a minority into a referendum-winning majority.

May’s big problem is that there is a real chance that none of these things will happen.

First, immigration might not fall as much as voters expect. The prime minister may have to allow some degree of free movement of people if British companies—and multinational companies operating in Britain—are to be able to trade on decent terms with the single market.

Second, the NHS is unlikely to receive anything like the extra £350 million ($430 million) a week that Brexit campaigners promised. This, they said, is the current weekly contribution that the UK pays toward the EU budget. The figure was, misleadingly, the gross contribution, not the much smaller net contribution. What is more, independent analysts reckon that Brexit will damage public finances and prevent anything more than a slight increase in NHS spending.

Finally, it may become clear that Brexit will not generate the extra jobs that British workers are expecting if, as many independent analysts predict, the economy falters in the medium term.

In twelve or eighteen months’ time, May could well have to dig herself out of a giant hole if it becomes clear that Brexit is not going to deliver much lower immigration, a big injection of extra cash for the NHS, and well-paid work for people who are today without work or stuck in low-paid jobs. She may be caught between the committed Euroskeptics in her own party, for whom national sovereignty matters more than the practical consequences of leaving the EU, and an electorate that is suffering buyers’ remorse and telling pollsters that a new referendum would produce a clear majority for staying in the EU after all. (This prospect would be enhanced if the rest of the EU responded to public pressure across the continent and decided to amend the freedom of movement rules for all 28 countries.)

What then? Could the prime minister tell the EU that the UK does not want to leave the EU after all? In the referendum campaign itself, she argued—albeit more softly than some of her cabinet colleagues—for the Remain cause.

Unfortunately, Article 50 does not say what should happen if a country changes its mind. What it does say is that withdrawal can be deferred beyond the two-year deadline—but this needs the unanimous agreement of all member states. The probability is that the unanimity principle would be applied to any attempt to abandon Brexit.

For the sake of argument, let’s assume that May does end up wanting the UK to stay in the EU and convinces the other 27 member states to accept her change of heart. She would then have to deal with the domestic political consequences. Given the fury she would provoke among a minority of Conservative members of parliament (MPs), she would probably need to call either a fresh referendum or an early general election. The second is more likely, given the current chaotic state of the Labour Party, with most local party members backing its left-wing leader, Jeremy Corbyn, while three-quarters of the party’s MPs at Westminster have no confidence in him. May could look forward to five more years in government with a substantially increased majority.

Enough of this speculation. The overwhelming probability is that May will not change course, that the UK will leave the EU by 2019, and that Britain’s next general election will not be until 2020 as planned. But as long as there is even a 5 or 10 percent chance that Brexit will not happen, the possibility of a U-turn should not be dismissed completely.

 

Peter Kellner is a journalist, political commentator, and former president of YouGov.