VladisChern / Getty / The Atlantic

Merely voting and counting the votes in this year’s election will be an extraordinary challenge. The country faces the worst public-health crisis in a century, a potentially severe shortage of poll workers, mail-in voting on an unprecedented scale, mounting functional problems at the U.S. Postal Service (with many alleging a plot to undermine voting by mail), and a president already dismissing mail-in voting as fraud-ridden. And all of this is taking place in one of the worst climates of partisan polarization and distrust in American history. To say that this threatens American democracy is no overstatement.

An additional factor could whip these variables into an outright constitutional crisis—one in which both Donald Trump and Joe Biden would claim to have been elected president, undertaking rival preparations for inauguration on January 20: The country’s laws on presidential elections provide a shockingly inadequate guide for resolving an Electoral College dispute. Twice in the past century and a half (in 1876 and 2000), the country narrowly averted a catastrophic deadlock over the presidential-election outcome. We may not be so fortunate in 2020.

Although multiple possible scenarios for a post-election crisis exist, the most plausible one will revolve around how long counting the votes in closely contested states might take. Remember those three states that, by a total margin of 87,000 votes, gave Trump an Electoral College victory in 2016? Michigan, Pennsylvania, and Wisconsin will again be hotly contested in 2020, in addition to a few other states such as Florida and North Carolina. The election result could ride on who wins one or two of these states, as it did in 2000.

Imagine that the presidential race in these states is still not clearly decided when, on December 8, the federal “safe harbor” period closes and states must certify who has garnered their electoral votes. Some mail-in ballots could remain to be counted, or disputes might persist about the validity of some ballots. Perhaps after trailing on Election Night, Biden has come from behind to take a lead as the mail-in ballots are counted. Following Trump’s warnings, Republicans might claim fraud. Or perhaps controversy arises over how closely the signatures on mail-in ballots match those in the voter-registration database. Then what?

Some have speculated that confusion will reign to the point that (under the Twelfth Amendment, ratified in 1804) the president will be elected by the House (with each state delegation casting one vote) and the vice president by the Senate. The much more likely scenario, however, is that partisan politics will drive decisions in each state—and quite possibly dueling decisions in some states. In Florida, Republicans hold both the legislature and the governorship; at least they could probably agree on whom to declare the winner (although the courts might step in with a divergent answer). But in Michigan, North Carolina, Pennsylvania, and Wisconsin, the legislatures are Republican and the governors are Democrats. In any of these states, a close election with an unresolved outcome could produce a scenario in which the state legislature certifies Trump’s electors and the governor certifies Biden’s electors—and both candidates claim to have been elected president.

If this happens, the situation will be so much worse than Bush v. Gore, which ended the contest over the 2000 election in December, long before it risked engulfing the nation in two conflicting claims to the presidency on January 20.

It could even be worse than the Hayes-Tilden election of 1876, which was settled two days before the inauguration. That was a close call, much too close for comfort, but a last-minute compromise (which ended Reconstruction) averted outgoing President Ulysses S. Grant’s biggest nightmare: the threat of two simultaneous inauguration ceremonies, and thus two commanders in chief demanding allegiance from the Army’s generals.

Could such a nightmare really occur this time, made all the worse by the deeply sobering necessity for America’s 21st-century military might—and its nuclear launch codes—to be under the control of one true commander in chief at all times? Additionally, the possibility of civil unrest swirling around two separate inaugurations is all the more frightening when one considers the weaponry in the hands of private militias, which could be deployed amid mass demonstrations attempting to influence the contest for power.

The Twentieth Amendment to the Constitution, ratified in 1933, was supposed to protect against such possibilities. It unambiguously ends a presidential term and begins a new one at noon on January 20. The Constitution does not allow any gap in commander-in-chief authority whatsoever.

Even more relevant, the Twentieth Amendment provides for an “acting president” if there is no president-elect when the new term begins. Save for the extremely unlikely scenario of a new vice president being elected by the Senate but not a new president by the House, the acting president (under the 1947 presidential-succession law) would be the speaker of the House (who would then be required to resign from Congress). She would exercise the full powers of the commander in chief unless and until the election controversy was resolved.

But that resolution would take effect only if there were no president-elect. What happens if there are two?

The Twelfth Amendment requires that the electors, after meeting and voting in each state, send their Electoral College votes to “the President of the Senate,” who is still the incumbent vice president of the United States. The trouble is that in some states, Biden and Trump electors might hold separate meetings on the same day and send in separate submissions, as the Hayes and Tilden electors did in 1876. After the electors send in their votes, the Twelfth Amendment then provides for a special joint session of Congress on January 6, at which “the President of the Senate” is to “open all the certificates” of electoral votes sent from the states “and the votes shall then be counted.”

The problem with this formulation, as is likely obvious to many readers, is its use of the passive voice. The amendment doesn’t say exactly who counts the electoral votes in this joint session of Congress—whether it’s Congress as a whole, the Senate president, or some other actor. As was noticed in the 1830s by Justice Joseph Story in his Commentaries on the Constitution, this phrasing leaves a dangerous ambiguity if the counting of electoral votes were to be seriously disputed when Congress convened one of these joint sessions.

This is precisely what happened when Congress deadlocked over the Hayes-Tilden dispute. The Senate wanted Hayes; the House, Tilden—and the Constitution provided no tiebreaker. The fight went down to the wire. Yes, the last-minute bargain averted the calamity of two simultaneous inauguration ceremonies, but it also consigned African Americans in the South to another century of subjugation.

A decade later, Congress passed the Electoral Count Act of 1887. With another potentially close election fast approaching, Congress wanted additional procedures in place. They knew that the compromise they cobbled together in the act was flawed. A leading political scientist of the day called it “almost unintelligible.” We defy anyone to read its operative section, a morass of contorted syntax, and profess to truly understand its meaning with respect to the circumstance when it would matter most: If Congress were to receive more than one submission of electoral votes from the same state, and the Senate and the House, on January 6, disagreed on which submission to accept as valid.

In that event, under one interpretation of the 1887 statute, the electoral votes would go to the slate of electors backed by the state’s governor. However, given the impenetrability of the statute’s wording, other interpretations, which have some support in the scholarly literature on the topic, would give authority to the Senate president to dictate a different outcome in certain situations. In the current circumstance, this would entail the sitting vice president, Mike Pence, deciding whether he and his running mate had been reelected—which, whatever the legal reasoning for this process, would be a disaster for the outcome’s legitimacy.

Conversely, Speaker Nancy Pelosi might assert that the electoral votes backed by a given Democratic governor must be counted, based on the interpretation of the statute that makes the governor’s certification the tiebreaker when the two houses of Congress disagree on which set of the state’s electoral votes to accept.

How to break this deadlock? The 1887 statute says that the joint session cannot consider the electoral votes of the next state (alphabetically) until the dispute over the one in question is resolved. Speaker Pelosi could therefore declare that the counting of electoral votes, and thus the presidential election itself, would remain incomplete unless and until the Senate president—who, after Pence’s term ends without a new vice president, would be the Senate’s “president pro tempore”—accepts the governor’s submission from the disputed state (which would elect Biden president). While still in office, however, Pence could point to separate sections of the same statute that give him the power to preserve order and move the proceedings along so that they can be completed in time for the inauguration to take place as scheduled.

In this way, January 20 could arrive with Vice President Pence insisting that President Trump has been reelected to a second term—while at the same time, Speaker Pelosi insists that there is no president-elect, because the process remains deadlocked, and hence she will assume the role of acting president until the counting of electoral votes from the states resumes with the disputed state resolved in favor of the governor’s submission.

The Supreme Court might ultimately decide the issue—days before the inauguration, with riotous waves of rival mass demonstrations demanding “justice” for their respective parties. The public might not, unlike in 2000, accept the Court’s view, and any Court decision could result in civic unrest on either side, or worse. But there is no guarantee that the Court would intervene at all. Unlike Bush v. Gore, which involved claims at a much earlier stage of the electoral process—before the electors met that year and before the election reached Congress—a dispute over who is commander in chief that arises because of congressional deadlock might be viewed by the Court as an inherently “political question” beyond judicial power.

In that event, who ought to be obeyed as commander in chief starting at noon on January 20: a reelected President Trump, or an Acting President Pelosi?

The best time to act to avert this disaster scenario is now—well before the November election. The sooner preventive steps are taken, the easier it is to separate them from immediate partisan calculations. In particular, two important steps can be taken immediately.

One good idea, advanced by Senator Marco Rubio, is to give states three and a half more weeks to count their popular vote and resolve any disputes. He has introduced a bill that would extend the federal “safe harbor” deadline for certifying electors, from December 8 to January 1. This would reduce the chances of a state submitting two conflicting slates of electoral votes.           

Second, we support the idea of a bipartisan commission to help guide the resolution of a dispute if the joint session of Congress becomes deadlocked during the two weeks between January 6 and 20. The bipartisan commission could emerge from civil society or be a creation of Congress. But its membership would need to be persuasive to a divided Congress in need of guidance, which would be much more likely if it were appointed by Congress. Moreover, if congressional Republican and Democratic leaders could agree on the structure and composition of such a commission—even if its writ were purely advisory—that agreement could ease the intense atmosphere of anxiety and animosity that suffuses the election.

The scenarios we have sketched here may seem improbable, but they are too risky to ignore. Better to create the commission now and have it be unnecessary than to have January arrive with it needed but nonexistent. And better, in this uniquely challenging election year, to give the states more time to count and certify their votes than to rush the electoral process, at risk of gravely damaging its legitimacy—and that of our entire democratic system.

This story is part of the project “The Battle for the Constitution,” in partnership with the National Constitution Center.

We want to hear what you think about this article. Submit a letter to the editor or write to letters@theatlantic.com.