In the recent revival of debate over whether a president can pardon himself (a matter on which I weighed in at National Review last year), two authors have relied heavily on the purported significance of a moment during the Constitutional Convention of 1787, which—they argue—disposes of the question definitively in favor of a self-pardoning power. 

Here at Public Discourse, political scientist Nicholas Higgins writes that “Locke and the Founders Agree: The President Can Pardon Himself.” The part of Higgins’s article devoted to John Locke’s account of executive prerogative in his Second Treatise is interesting in its own right, but it sheds no light on the meaning of the Constitution regarding the matter at hand. Locke’s prerogative is by definition extra-legal action, whereas the presidential pardon is a power established by and answerable to the law of the Constitution.  

Higgins then turns to something that he claims is of central relevance to the question of a president’s power to pardon himself. On Saturday, September 15, 1787, on the last full day of debate on the text of the Constitution they would vote to send to the country two days later, the framers at Philadelphia had one final discussion of the scope of the president’s pardon power. Edmund Randolph of Virginia, who would on that final Monday decline to sign the Constitution, rose on Saturday to move that one more exception be made to the president’s power to “Grant Reprieves and Pardons.” It was already confined to federal crimes (“Offences against the United States”) and could not reach crimes under state law. Neither could it reach “Cases of Impeachment,” a zone of legislative power that the Constitution had already walled off from the criminal law and made wholly political, in a departure from English practice. 

But Randolph was concerned that the power was still too capacious, and moved that the pardon clause also contain the words “except cases of treason.” His motion brought to the surface lingering doubts about the unilateral executive pardon power among a handful of prominent delegates, including George Mason (also a non-signer at the end), who supported Randolph’s motion, and both Rufus King and James Madison, who thought that in treason cases at least the power should require the concurrence of the Senate. If that was meant as a friendly amendment of Randolph’s motion, it was one he refused to accept, and his motion was defeated eight states to two (one state abstaining because its delegates were evenly divided, and two other states being unrepresented at the time). 

Start your day with Public Discourse

Sign up and get our daily essays sent straight to your inbox.

Higgins’s interpretation of this moment turns on something Randolph said in making his motion, and on the response to Randolph—as Higgins understands it. As James Madison recorded it in his notes, Randolph said, “the prerogative of pardon in these [treason] cases was too great a trust. The President may himself be guilty. The Traytors may be his own instruments.” While others, as I noted above, supported or proposed to modify what Randolph had moved, James Wilson of Pennsylvania said this (quoting his remark in its entirety as Madison recorded it): “Pardon is necessary for cases of treason, and is best placed in the hands of the Executive. If he be himself a party to the guilt he can be impeached and prosecuted.” After a few more exchanges—the whole debate taking up just about a page in Max Farrand’s Records of the Federal Convention of 1787—the vote was held that defeated Randolph’s motion. 

That’s it; that’s all there is. On the question of a president’s own possible involvement in treason, we have Randolph saying, “The President may himself be guilty. The Traytors may be his own instruments,” and Wilson saying, “If he be himself a party to the guilt he can be impeached and prosecuted.” No one else said anything that could remotely be said to bear on this question, and then the vote was held.  

How does Higgins interpret this episode? He writes, “From this debate, it is clear that the founders recognized the president could legally pardon himself for federal crimes, including treason.” 

Writing in the Washington Post, law professor and former federal judge Michael McConnell relies on the same September 15 episode at Philadelphia, and is just as certain of the same conclusion Higgins draws:  

The framers of the Constitution thus specifically contemplated and debated the prospect that a president might be guilty of an offense and use the pardon power to clear himself. They concluded that the remedy of impeachment by the House and conviction by the Senate was a sufficient check on the possibility of abuse. 

But Higgins and McConnell are too sure of this reading of the event, and of its significance for our own interpretation of the Constitution. 

First, there is insufficient evidence, in these few lines, to warrant the conclusion that even the only two relevant speakers—Randolph and Wilson—were thinking about presidential self-pardon, even for a second. Randolph may only have been thinking about a treasonous president pardoning his confederates (“his own instruments”) to shield himself from discovery. Certainly this possibility was alarming enough by itself to prompt Randolph’s motion. (McConnell omits Randolph’s sentence on those other “Traytors”; Higgins includes it but makes nothing of it.) And Wilson likewise may have been thinking only of those traitors, and contenting himself that the impeachment power could reach the grave abuse of the pardon power’s use to protect them. (McConnell also overconfidently asserts that when Wilson said such a president “could be impeached and prosecuted,” he meant only the prosecution of an impeachment in the Senate. I would not be so quick to rule out Wilson’s having meant that a president could be criminally prosecuted after impeachment and removal. Without more, neither reading is definitive.) 

Second, the debate on Randolph’s motion ranged over more ground than this question (if it touched on it at all, which there is reason to doubt) and the whole matter probably took up (at a generous estimate) no more than half an hour of the Convention’s time, with six members talking. Given the brevity and the equivocal nature of what Randolph and Wilson had to say, it seems altogether conclusory to assert a) that everyone else in the room understood a debate to have just taken place over the president’s power to pardon himself, b) that in voting on Randolph’s “except cases of treason” amendment of the pardon power, they also understood themselves to be voting whether to adopt a shared meaning of the unamended power that included a right of the president to pardon himself, and that c) therefore we have on the record a “specifically contemplated and debated” question on which the states voting came down eight to two in favor of a president’s self-pardon power. Indeed, merely to state the underlying logic of Higgins and McConnell in this fully descriptive way is to glimpse how insupportably weak the conclusion is. 

Far more persuasive is the account of law professor Brian Kalt, who wrote extensively on this question in 1996, and who says of the debate on Randolph’s motion—accurately—that “[t]he self-pardon was nowhere mentioned.” As for what we can infer from the Randolph-Wilson exchange, Kalt says there are three possibilities: that the potential of a self-pardon was just completely overlooked; that it crossed the minds of the framers present but was dismissed as clearly unthinkable and inconsistent with the constitutional principles on which they were acting; and that it came into their minds and was accepted by them as a valid potential use of the power, for which impeachment (which could result only in a “political” punishment of removal from office and disqualification for future service) was the only remedy.  

Rightly, in my view, Kalt considers the third possible reading of the debate as by far the unlikeliest. A reasonable reading of the September 15 debate would either be that a power of self-pardon was on no one’s mind at all, or that it was dismissed as an obviously invalid attempt to use the pardon power. 

But there is another problem with the argument of Higgins and McConnell. That is that this is simply not how originalism should be done at all. Let us suppose that theirs is the best, most persuasive reading of this brief and rather diffuse discussion of the pardon power on the last working day of the Constitutional Convention. As the late Justice Scalia was fond of pointing out, the views of individual lawmakers in the midst of debate are not themselves the law we must interpret. Neither are the votes taken in a deliberative body rightly viewed as votes on anyone’s interpretation of the text under discussion. The text that they passed, not what they said about what they passed, is the law.  

Surely this wisdom is doubly relevant when we are trying to infer from a debate on one thing (here the pardon’s use in treason cases) that the participants actually also debated and decided another thing that no one actually mentioned (here a presidential self-pardon). It’s triply relevant when the vote they actually cast on the thing they did debate bore no evident relationship to that other thing that . . . again . . . no one explicitly mentioned.  

And it’s quadruply relevant when we recall that the only record we have, of this debate and vote on the pardon power that was a non-debate and non-vote on the self-pardon, is from the Convention notes of James Madison, which were not publicly available for interpreters of the Constitution to consult until after Madison’s death, a half century after the event. I hasten to add that Madison’s notes are a hugely important documentary record for historians and other scholars, and I agree with political scientist Lynn Uzzell that their accuracy has not been drawn into serious question by recent scholarship that purports to do so. But whatever Madison’s notes may suggest about the question at hand—and as we’ve seen, that’s barely more than nothing at all—they could not by themselves be dispositive of the interpretive question we must answer. 

The expression “original public meaning” is often used as a description of what the present-day originalist interpreter of the Constitution is after. Ascertaining “original public meaning” is easier said than done in many cases. But it cannot be done at all by referring decision of a contested interpretive question to extrapolations from ambiguous notes on a brief discussion that occurred behind closed doors, of which the public could know nothing for the next half century of its life under the Constitution.   

In some respects, “original public meaning” is one word too many. For the public at large, in the period from 1787 to 1791 (or whatever cut-off year you please for the founding generation), may not have been in possession of the most accurate understanding of every provision of the Constitution. The Constitution has terms of art and intricacies of structure and context that might only yield their true meaning to the experienced eyes of the constitutional practitioners—not just judges and lawyers but legislators and executive officials too—who have to use the Constitution in practical ways governing concrete circumstances. As Madison said in Federalist 37, some provisions of the Constitution might perforce have to be “considered as more or less obscure and equivocal, until their meaning be liquidated and ascertained by a series of particular discussions and adjudications.” 

Just so. Thinking through constitutional principles, considering the logic and context and relations of various provisions and institutions to one another, is what we must undertake today as we search for the Constitution’s original meaning. This entails far more than cherry-picking a passing moment in a complex, multi-faceted debate that was just coming to a close, a moment that is itself (as we’ve seen) at best “obscure and equivocal,” and declaring victory for one side on a contested question we consider important today but that the founding generation never explicitly discussed at all. No, that’s just not how our work is done at all.