A jail cell on death row at the state penitentiary in Huntsville, TexasReuters

The only people who truly know whether Oklahoma's three-drug cocktail causes an agonizing death are the ones it kills. Charles Warner asked the Supreme Court to stay his execution last night so they could hear his doubts about the sedative midazolam, one of the three lethal injection drugs used in Oklahoma and other states. The justices refused his appeal in a 5-4 vote, and the state of Oklahoma executed him for the rape and murder of an 11-month-old girl in 1997 shortly thereafter.

Did he die in excruciating pain? We don't know. An AP reporter who witnessed Warner's execution quoted him as saying "It feels like acid" and "My body is burning" during the process. Other witnesses told reporters, however, that Warner didn't appear to visibly suffer during the process. Was Warner lying? Thanks to one of the other drugs in Oklahoma's cocktail, we don't know that, either. "[B]ecause Oklahoma injected Mr. Warner with a paralytic tonight, acting as a chemical veil," federal public defender Madeline Cohen said in a statement after her clients' death, "we will never know whether he experienced the intense pain of suffocation and burning that would result from injecting a conscious person with rocuronium bromide and potassium chloride."

What else don't we know? To light our way through this ontological darkness, we turn to former Secretary of Defense Donald Rumsfeld during a press conference in 2002:

Reports that say that something hasn't happened are always interesting to me, because as we know, there are known knowns; there are things we know we know. We also know there are known unknowns; that is to say we know there are some things we do not know. But there are also unknown unknowns—the ones we don't know we don't know. And if one looks throughout the history of our country and other free countries, it is the latter category that tend to be the difficult ones.

There are some known knowns. We know the drugs used—the sedative midazolam, the paralytic rocuronium bromide, and the toxin potassium chloride—and their dosages. We know it took about 20 minutes for Warner to die, a slightly above-average span but still shorter than Clayton Lockett's 43-minute death last April or Joseph Wood's one hour and fifty-seven minute execution last July in Arizona.

But there are many known unknowns. We don't know where Oklahoma obtained the drugs it used thanks to a state-secrecy law that conceals their manufacturer. Did they come from unregulated compounding pharmacies, some of which have been criticized for using substandard ingredients when manufacturing drugs? We don't know. We also don't know who administered the drugs or their qualifications.

Some unknowns surround the court's decision last night, too. Why didn't Anthony Kennedy, the court's current swing justice, join the dissenters? Since 2000, he's been the key vote in many major capital cases, often by siding with the court's liberal wing. The death penalty's flaws are a known known for Kennedy, who has limited its scope on multiple occasions. He even authored the court's opinion in three of the most important post-Furman decisions: Atkins v. Virginia, which forbade executions of the mentally disabled, Roper v. Simmons, which banned the death penalty for juvenile offenders, and Kennedy v. Louisiana, which abolished capital punishment for offenses other than murder or crimes against the state. As the court refuses to address major constitutional questions about state-secrecy laws and new drug cocktails, his silence is striking.

We don't know why Kennedy and the four conservative justices denied Warner's request for a stay, but we do know why their liberal colleagues disagreed. In her eight-page dissent, Justice Sotomayor, writing for herself and the other three justices, grasped the ontological problem at hand:

Although the State emphasizes that Florida continues to employ a lethal injection protocol that utilizes the same drug types and amounts as will now be employed in Oklahoma, its apparent success with that method is subject to question because the injection of the paralytic vecuronium bromide may mask the ineffectiveness of midazolam as ananesthetic: The inmate may be fully conscious but unable to move. The deficiency of midazolam may generally be revealed only in an execution, such as Lockett’s, where the IV fails to sufficiently deliver the paralyzing agent.

Indeed, there's a haunting possibility that Clayton Lockett's botched execution last April may have given us a glimpse into what inmates experience during a typical execution with midazolam. Toxicology reports showed he had enough of the drug in his system to induce unconsciousness, and yet observers could tell he suffered because the IV failed to deliver enough of the standard paralytic. They heard him moan and gasp. They saw him twitch and writhe. An unknown unknown became known.

The idea that Lockett's experience could be the silent norm in executions with midazolam, not an exception, is deeply disturbing. Even more troubling is the inmates' inability to untangle these knowns and unknowns. The Supreme Court has created a constitutional catch-22 that only it can resolve. This paradox strikes at the foundation of a condemned inmate's constitutional rights. Perhaps many Americans support inflicting pain on child murderers in their final moments, but the Eighth Amendment does not.

The only people who truly know if the midazolam cocktail causes excruciating pain can't tell us. But this could change as more states experiment with new and unpredictable lethal injection drugs. If the justices want firsthand testimony, America's rapidly deteriorating capital punishment system may soon provide it. The states shouldn't fear another inmate dying in a botched execution; what they should fear is an inmate surviving it.

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