Everett / Library of Congress / Paul Spella / The Atlantic

“What did the Framers think about impeachment?” This question is everywhere these days, and the answer that follows often references James Madison’s rejection, on September 8, 1787, of the term maladministration in favor of “high crimes and misdemeanors.” The implication is, supposedly, that a president cannot be impeached for mere poor governance. It’s a good story, and one that can be found in accounts as far back as Watergate.

The source of this story is Madison’s notes, his record of the Constitutional Convention, which is today stored in a vault at the Library of Congress. But there’s just one problem: The specific sheet that is the only evidence of the famous impeachment conversation isn’t a solid source. I spent years studying Madison’s manuscript, and this sheet is the oddest one in it. It does not date from 1787, but from the early 1790s. Maybe the conversation happened in 1787 on the floor of the convention, as Madison tells it. Maybe it didn’t. But either way, the uncertainty is itself instructive, a reminder of our distance from the framing generation; historical evidence cannot absolve Americans now of their obligation to interpret the Constitution for today.

Of course, there are plenty of reasons one might not care about the details of the debate in 1787 in the first place. Although the Framers’ general concerns about corruption of power resonate, the working structure of the American constitutional system is far removed from that of 1787. An attempt to insist on the singular importance of the precise words used in 1787 seems fraught with conceptual problems, if not altogether imprudent. The practice and traditions of impeachment procedure over two centuries seem a far more constructive place to look for guidance.

But, for those tempted to return to the summer of 1787 as a touchstone, here is a caution about placing undue weight on Madison’s account of who said what on September 8, 1787.

September 8 is one of two points in Madison’s notes with significant discussion about impeachment. The first discussion occurred on July 20, as the convention debated the June report of the Committee of the Whole House. At this point, the convention had not yet agreed on how to elect a president, and indeed, was two drafts away from the final instrument. Madison heard and recorded a lengthy discussion about whether to have an impeachment power, and in the end, the power was retained. This discussion reads the way Madison’s May–July notes generally read: with debate over political principles, structural pragmatics, and historical and contemporary examples.

On September 8, the convention again discussed impeachment as it completed its review of the early-August draft of the Constitution. Postponed and difficult matters were sent to a committee consisting of one member from each state. Four days earlier, the committee had partially submitted a report with an impeachment procedure. In scattered additions and alterations, the committee proposed, among other things, conviction by the Senate “for treason or bribery.”

Not until September 8 did the convention reach review of this language. According to the official record—the journal kept by William Jackson, the convention’s secretary—at that point, the convention added additional words: “‘or other high crimes and misdemeanors against the State’ after the word ‘bribery.’” The committee vote was seven to four in favor. (For those counting, New York had left and Rhode Island refused to arrive.) The words United States were substituted for State. And the clause was agreed to 10 to one. The convention then added that the vice president and other civil officers of the United States were also subject to the impeachment procedure and inserted language that required Senate members to be “on oath.”

Madison’s notes provide a more detailed discussion of this substitution. George Mason, as Madison has it, argued for and then made a motion to add “or maladministration.” Elbridge Gerry seconded him. Madison stated: “So vague a term will be equivalent to a tenure during pleasure of the Senate.” Gouverneur Morris argued rather ambiguously, “It will not be put in force & can do no harm—An election of every four years will prevent maladministration.” Mason then withdrew “maladministration” and, according to the notes, substituted “other high crimes & misdemeanors.”

Though this story is now used to justify consequential interpretations of the Constitution, relying on Madison’s notes as a verbatim transcript is perilous. The notes were notes; more precisely, a legislative diary. They were taken by a very involved, opinionated participant—and one who repeatedly found himself on the losing side of votes that summer. Madison did not write the version in the Library of Congress on the floor of the convention. He took rough notes and, likely twice a week, sat down and turned them into the manuscript that is now called Madison’s notes. Sure, Madison tended to translate other people’s arguments into his own way of phrasing ideas and concepts, but nonetheless, the notes from May to mid-August reveal some sense of the large topics that concerned the drafters and the structural problems that bedeviled them.

But beginning with August 22, Madison’s notes present a significant problem. Madison served on multiple committees in late August, and also became sick. Whatever rough notes he took during the proceedings after that date were not written up during the summer of 1787.

Instead, the section of Madison’s notes from August 22 to the end of the convention was likely written two years later, in the winter and spring of 1789–90, when Madison knew that Thomas Jefferson finally would return to the United States from France. Madison had told Jefferson that he planned to share his notes: “As soon as I am at liberty I will endeavor to make amends for my silence, and if I ever have the pleasure of seeing you shall be able to give you pretty full gratification. I have taken lengthy notes of every thing that has yet passed, and mean to go on with the drudgery, if no indisposition obliges me to discontinue it.” The most visible difference between the original sections and these later ones is that in the original sections, Madison did not write his own name out. He wrote, instead, “M.” Later, when finishing the manuscript for Jefferson, Madison consistently wrote his name as “Madison.”

Completing the notes in 1789–90 allowed Madison to integrate verbatim sections of the official convention journal with his rough notes. At the end of the convention, George Washington was put in charge of the journal, the only official record of the proceedings. At some point, likely the fall of 1789, Madison made a personal copy. Why and how Washington lent him the original manuscript to copy is unknown, because Washington’s diary for that time period is missing. In the May-to-mid-August sections of his notes, Madison occasionally added information from the journal to correct and clarify his own account of the procedures. These changes are obvious. But in the post–August 22 section, the journal material is wholly integrated, and Madison’s discussion of procedures tracks the journal’s precisely.

How much distance was there between the discussions of 1787 and Madison’s efforts to recall it two years later? By the winter of 1789, Madison had heard endless debates over the meaning of the Constitution, including within the First Congress itself, and had led the effort to have amendments added. These debates—and his views on them—could not have been erased from his mind as he went back to finalize his notes from August 22 on. In fact, when he asked his friend Edmund Randolph to rewrite Randolph’s opening speech so Madison could include it, Randolph declined, explaining that he would “mingle inadvertently much of what I have heard since, without being able to separate it from what occurred then.”

But even in this post–August 22 section, one sheet stands out as highly peculiar—the sheet with the September 8 colloquy. When Madison wrote his notes, he folded large sheets in half and wrote on the four resulting pages. On this sheet, Madison’s writing does not extend across the four sides. The final words trail off in the middle of page four: “Mr Madison, objected to a trial of the President by the Senate …” The partial sentence was completed on the next sheet: “Senate especially as he was to be impeached by the other branch of the Legislature, and for any act which might be called a misdemeanor.” Clearly, Madison wrote the September 8 sheet to fit between two other sheets. One can see how he wrote “Senate” at the far left of the page, only to realize that the next sheet began with “Senate,” and so he crossed it out and rewrote “Senate” as a carry-word on the right side.

Additionally, an unusual watermark can be seen in the center of the sheet: Taylor. There is no other sheet marked “Taylor” in the notes manuscript.

And perhaps the most revealing clue: This sheet was missing in the early 1790s, when Jefferson asked his nephew John Eppes to make a copy of the notes. Eppes reached this section of the manuscript and noted: “There appears to be wanting in this place part of a days debate …”

I have a theory as to why: As I wrote in my book Madison’s Hand, I suggest that Madison had replaced an earlier version of the September 7–8 material for mysterious reasons that involved the treaty power. The treaty-power discussion on September 7 was a matter of considerable controversy. In 1793 and 1796, Madison, Jefferson, and Alexander Hamilton disputed constitutional treaty power. Something came to bother Madison in this section. But in the process of reworking that material, Madison messed up. He accidentally created a gap between sometime on September 7 and sometime on the eighth, though the precise times of day are unknown. When John Eppes noticed, Madison wrote the Taylor sheet to fill the gap.

In filling the gap, did Madison copy verbatim his reconstructed 1789-1790 version of September 8’s impeachment proceedings or, by the mid-1790s, did he remember the proceedings in a way that led him to emphasize his rejection of maladministration? Did his reconstructed 1789-1790 version itself accurately represent the discussion leading to “high crimes and misdemeanors”? And even if Madison implied some distinction between the terms, did the Convention care or was “high crimes and misdemeanors” chosen because it was so expansively ambiguous? We do not know.  

Does it matter? Does it matter whether Madison—never mind anyone else—worried about maladministration? In a world filled with anxiety about official abuse of power, the convention had agreed to “conviction of malpractice or neglect of duty” (July 20, eight to two), then “treason, bribery, or corruption” (August 6, Committee of Detail report), then “treason or bribery” (September 4, Committee of Eleven report), and finally “treason or bribery or other high crimes and misdemeanors” (September 12, Committee of Style and Arrangement report), all as reasons for impeachment. The takeaway from this shouldn’t be the specifics of what impeachment was for—the delegates could have landed on any combination of these—but all of these phrases intended to convey that impeachment was intended as a safeguard, to prevent the undermining of normal political processes. The concept ensured a fail-safe mechanism in which senators would serve on oath or affirmation in a different role from ordinary political representatives.

The historical record seduces us with the illusion of answers—answers that might allow us to evade the responsibility for deciding in each generation what it means to “secure the blessings of liberty to ourselves and our posterity.” The history of the framing helps us understand the reverberating concerns of American constitutional government. But we do not know precisely what James Madison said on September 8, 1787, and we shouldn’t imply to the American people that we do—nor that we need to.

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.