The most likely scenario is that Clinton is going to trounce Trump in the general election. She is up anywhere from 6.4% to 10%. A recent Rasmussen poll that showed Trump slightly ahead is clearly an outlier. But in modern elections, the indelible images of candidates are established in the primaries, and candidates can do little to alter these in the general election. Most famously, Bob Dole’s image was defined by Bill Clinton and the Democrats in the spring of 1996 even before the Republican National Convention, and Dole could do nothing to reverse public perceptions during the general election. As high as Trump’s negatives are at this point, they’re only going to get worse when the Democratic advertising onslaught begins in earnest. Trump is counting on driving up Clinton’s negatives as well, but all he can do is remind voters of information that is already baked into her well-defined public persona. Barring indictment by the FBI, it is unlikely her negatives will change between now and the election.

Trump and Ted Cruz have argued that Mitt Romney was a flawed candidate in the 2012 election, and that too many Republicans stayed home, handing the election to Barack Obama. Political consultant Karl Rove has persuasively discredited this myth. The silent majority that Trump hopes to bring to the polls is a figment of his imagination, and his staggering negatives among women (66% unfavorable), African-Americans (79% unfavorable), and Hispanics (71% unfavorable) doomed his candidacy from the beginning. There may be some truth to his claim that he will bring the Reagan Democrats back to the Republican Party and make Republicans competitive in Rust Belt states. His anti-free trade posture will appeal to working-class white male voters, but their numbers have dwindled in the years since Reagan was elected, and the voters he has already alienated through his belligerent style and misogynist comments will dwarf the blue-collar additions.

If Trump can’t win and his nomination represents a hostile takeover of the Republican Party that jeopardizes the principles and traditions of the party, what are the alternatives for NeverTrump Republicans? Given the difficulty of getting a third-party name on the ballot in time for the general election, the only alternative is to run a fusion candidate with the Libertarian Party, which is already on the ballot in 32 states and actively petitioning to get on the ballot in the remaining 18. In 2012, 95% of the electorate had the option of voting Libertarian. While Gary Johnson, the Libertarian candidate, received barely 1% of the vote, the Libertarian share of the vote in 2016 could be dramatically higher if a Republican ran who was minimally acceptable to them. This fusion should be attractive to Libertarians because it would legitimate their party on a national scale and would provide an alternative for Republicans—particularly Republicans running for House and Senate seats—to support.

Could a Republican masquerading as a Libertarian win the election? Probably not, but consider the following long-shot possibility: If the Libertarian/Republican candidate won (was the plurality winner) in five or six states, it might prevent Trump or Clinton from getting to 270 electors in the Electoral College. If the Electoral College doesn’t produce a winner, the constitutional remedy is election by the House of Representatives. Each state gets one vote, so the representatives from each state caucus, and the winner after polling the representatives receives the vote of that state. California and Rhode Island each get one vote, so don’t expect representational equality. If there are an even number of representatives, and voting for candidates leads to a tie, then the state does not cast a ballot. This mechanism has been used to resolve two presidential elections since the ratification of the constitution: 1800 and 1824.

What would happen under these circumstances? Currently, Republicans are predominant in 33 of the 50 state delegations, so a Republican candidate would win. Few House Republicans have endorsed Trump at this point, and Paul Ryan, whose influence would be critical in this process, has distanced himself from Trump.

One more wrinkle: It would be the newly elected House that is sworn in on January 3rd, 2017, unless Congress designates another day. We can’t simply assume the partisan configuration of the old House will be duplicated by the new House. The outcome is impossible to predict.

Donald Brand is a professor of political science at the College of the Holy Cross.