The Brandywine Review

Commentary on American law

Criminal Law

Criminal Law

Executive clemency comes in more than one shape, and the shapes do not behave the same way once a court has to define them. A pardon forgives and restores. A commutation shortens a sentence, sometimes on conditions the recipient must accept. A reprieve only postpones, and of the three it is the one courts had the least occasion to define carefully, until a governor tried to use it for something larger than a single execution date.

Ex parte Garland, 71 U.S. (4 Wall.) 333 (1867), set the outer boundary for the pardon itself, holding that a presidential pardon restores civil rights and that Congress cannot achieve indirectly, through a postwar oath requirement barring a pardoned former Confederate from practicing law, what a pardon has already forgiven directly. Schick v. Reed, 419 U.S. 256 (1974), upheld a related but narrower power, President Eisenhower's commutation of a death sentence on the express condition that the recipient never become eligible for parole, holding that a President may attach conditions no statute specifically authorizes, so long as the conditions do not themselves offend the Constitution.

The reprieve power sat mostly untested until Pennsylvania's governor used one in 2015 to do something reprieves were not traditionally used for: not to delay a single execution for a discrete reason, but to put every death sentence in the state on indefinite hold. Prosecutors argued the power did not stretch that far, that a reprieve required a fixed end date and a purpose tied to the individual prisoner's circumstances, not a standing policy dressed up in clemency language. The Pennsylvania Supreme Court disagreed, unanimously. In Commonwealth v. Williams, 129 A.3d 1199 (Pa. 2015), the court held that when the reprieve power entered the state's 1790 constitution, it was understood as covering any temporary postponement of sentence, without a requirement of a fixed term or an individualized justification. That reading makes the reprieve power, on paper, considerably stronger than either the pardon or the commutation power. A pardon in most states runs through a board or requires its recommendation; a commutation often does too. A reprieve, at least in Pennsylvania, requires neither. The court upheld only the single reprieve to Williams on that basis, and expressly declined to decide whether the same reasoning would also justify Wolf's broader announced moratorium policy. Nothing in the court's reasoning about the reprieve power itself is confined to Pennsylvania's own constitutional text.

A different kind of indeterminacy runs through how courts read each other's fractured opinions. Marks v. United States, 430 U.S. 188 (1977), held that when no rationale commands five votes, the holding of a fragmented decision is the position taken by the justices who concurred in the judgment on the narrowest grounds. The rule sounds mechanical. It is not, because plenty of splintered opinions rest on genuinely different reasoning rather than narrower and broader versions of the same reasoning, and Marks does not obviously work when that happens.

The D.C. Circuit tried to supply the missing limit in King v. Palmer, 950 F.2d 771 (D.C. Cir. 1991) (en banc), holding that the narrowest grounds approach applies only when one opinion's reasoning is a logical subset of a broader opinion's, so that the narrower view would necessarily be adopted by anyone who accepted the broader one. Where the competing opinions instead rest on different rationales entirely, the court held, no narrowest grounds opinion can be extracted, and the fragmented decision binds nobody beyond its own facts. Other circuits have kept applying Marks more loosely, treating a plurality's reasoning as controlling even where it is not obviously nested inside any other opinion in the case.

The Supreme Court had a clean chance to resolve which version is correct in Hughes v. United States, 584 U.S. 675 (2018), where the question presented asked directly whether lower courts owed obedience to a four justice plurality or to Justice Sotomayor's narrower concurrence in the earlier case the dispute traced back to. The Court answered the underlying sentencing question on the merits and left the Marks question exactly where it found it. A circuit applying King v. Palmer's logical subset test will refuse to extract a controlling rule from some fragmented capital sentencing decisions that a circuit without that limitation will treat as binding precedent, and a defendant's sentence can turn on which of those two approaches happens to govern the circuit where he was tried.