As Britain’s Labor Party prepares to announce the result of its leadership election Saturday, attention is beginning to turn to how the party will reunite after months of infighting. Whether left-wing incumbent Jeremy Corbyn is re-elected — as is highly likely — or is defeated by challenger Owen Smith, the Labor Party looks set to remain highly divided. Indeed, the talk all summer was not of reconciliation but of a potential split in the party.

Labor’s members of parliament, mostly from the moderate wing of the party, voted against Corbyn 172 to 40 in a no-confidence motion in June. By contrast, its grass-roots membership is increasingly left-wing, with Corbyn carrying them by 62 percent to 38 percent — with 86 percent of members joining in the last year favoring Corbyn — in the most recent opinion poll.

While the electoral consequences of a party split are relatively minor under the proportional electoral systems used in much of Europe, Britain’s district plurality — or “first past the post” — electoral system punishes division heavily. In a winner-takes-all system, dividing up the votes does not mean dividing up the seats. Broad church parties that encompass various ideological strands are electorally greater than the sum of their parts.

Labor has been here before. In 1981, in outwardly similar circumstances, a group of moderates broke away to form the Social Democratic Party. The adventure ended in failure, and the 1983 election saw Margaret Thatcher lead the Conservatives to their most decisive postwar victory.

It is possible that moderate Labor MPs, feeling they are faced with dire electoral fortunes in any case, take the view that a split could not make things much worse. But just how a division of the party would play out would depend on how Labor’s voters break across the districts it holds.

It’s difficult to judge how people would vote in a hypothetical situation. With that caveat in mind, polling suggests that whichever of the parties kept the Labor name would retain about two-thirds of current Labor voters, a vote share about 10 percentage points lower than at the last election.

If the 2015 election had been contested on the electoral districting boundaries proposed for the next general election, the Conservatives would have won an estimated 321 seats in a smaller 600-seat House of Commons — they won 330 — and Labor on 200, instead of 232 seats. Under a party split situation, if the 10-point predicted drop in Labor’s vote share were uniform across Great Britain, then the party would lose an additional 38 seats under the new districting system. Taking into account movements in the polling since 2015, it’s probable that Labor would lose a further eight seats, leaving it with 154 seats and the Conservatives with a majority of 138; a similar result — in terms of the Conservative majority — to Margaret Thatcher’s 1983 landslide.

Yet the reality could be even worse. If the support lost to the new breakaway party were not 10 percentage points everywhere, but one-third of Labor’s vote in each district, the electoral arithmetic is such that Labor would be performing worse in the all-important marginal seats. An analysis of data published by the British Election Study suggests that Labor voters hostile to Corbyn are indeed spread in such a geographic pattern. Though not conclusive, it nevertheless points towards potential supporters of a moderate breakaway party being distributed that way.

Under this set of assumptions, Labor would be left with 111 seats and the Conservatives would win a majority very close to 200. While the electoral carnage for Labor could be mitigated — to a point — by electoral pacts and tactical voting, such measures would prove difficult amid the uncertainty over which party would be best placed in each district. Essentially, a split would see Labor lose the advantages it has under the current, first-past-the-post system.

Labor typically wins a given percentage of seats even on a substantially lower percentage of votes in the electoral system. And although its advantage is smaller than that enjoyed by the Conservatives, the electoral system continues to protect Labor from being supplanted as the second party. As such, staying together — at almost any cost — is likely to be the Labor Party’s least bad option. Making up is hard to do. But the alternative would be suicidal.

Singh runs Number Cruncher Politics, a nonpartisan polling and elections site that predicted the 2015 U.K. election polling failure.

Matt Singh is a columnist for Bloomberg View.