The three-part test that haunted churches for fifty years is finally gone

Every so often a legal doctrine dies, and almost nobody holds a funeral. The Lemon test is the best example in American constitutional law. For roughly half a century it was the test that courts were supposed to apply whenever somebody sued over a government action touching religion: a nativity scene, a cross, a bit of aid to a church school, a prayer. It had three prongs, it sounded sensible, and it produced results that no one could predict. The Supreme Court stopped using it in a pair of decisions, the second of which said so plainly. In 2022, for the first time, the Court's opinion told the country that it had long ago abandoned the thing. Justice Antonin Scalia, who disliked the test intensely, had a line for it that I cannot improve on: he compared it to a ghoul in a late-night horror movie that is killed and buried, and then keeps rising from the grave to frighten the children.

I am going to take you through the test, the case that made it, the confusing results, the two decisions that ended it, and what the Court uses now. I'll then give you the best objection to the replacement, because there is a real one. Think of it as a short biography of a rule, from birth to burial, with a verdict.

The case that made it

Lemon v. Kurtzman was decided in 1971. It combined challenges to two state programs. Pennsylvania had a law under which the state reimbursed nonpublic schools, most of them Catholic, for part of the cost of teachers' salaries, textbooks and instructional materials in secular subjects. Rhode Island had a law that supplemented the salaries of teachers of secular subjects in nonpublic schools by 15 percent. Alton Lemon was a Pennsylvania resident who sued.

Chief Justice Warren Burger wrote the opinion striking down both programs. The Court pulled together principles from earlier decisions into three requirements. For a law to survive an Establishment Clause challenge:

1. Purpose. The statute must have a secular legislative purpose.
2. Effect. Its principal or primary effect must be one that neither advances nor inhibits religion.
3. Entanglement. It must not foster an excessive government entanglement with religion.

The third prong did a lot of damage. In Lemon itself, the Court reasoned that to make sure the aid was spent only on secular subjects, the state would have to monitor the religious schools, and that very monitoring would entangle the government with the church. Aid that was too loose would advance religion. Aid that was policed closely would entangle the government. The Court put the schools on a seesaw and they could not win.

Why it was a mess

In theory, a three-part test gives you predictability. In practice, the test gave justices a vocabulary to reach whatever result they had in mind. Look at what happened across the years. A few examples, drawn from real cases:

• The nativity scene. In Lynch v. Donnelly (1984), the Court allowed a city's Christmas display in Pawtucket, Rhode Island, which included a crèche together with a Santa Claus house, reindeer and other seasonal figures. Five years later, in County of Allegheny v. ACLU (1989), the Court struck down a crèche on the grand staircase of a Pittsburgh courthouse but allowed a menorah displayed outside a city-county building, next to a Christmas tree. Lawyers and judges spent years debating whether a reindeer saved a manger scene.
• The Ten Commandments. On June 27, 2005, the Court ruled against a Ten Commandments display in a Kentucky courthouse and in favor of one on the grounds of the Texas Capitol, both by 5-4. I cover those two in Two Ten Commandments cases on one day, opposite results, and the reason. A single justice sat in the middle of both.
• Aid to religious schools. The Court allowed some forms of aid, such as textbook loans, and barred others, such as maps and projectors, with distinctions that few lawyers, let alone school principals, could explain in advance.
• Legislative prayer. In Marsh v. Chambers (1983), the Court upheld the Nebraska legislature's practice of opening with a prayer by a paid chaplain, and it did so without applying Lemon at all, relying instead on the fact that the First Congress had hired chaplains. So even at the height of the test, the Court had an escape hatch whenever the result looked absurd.

The result was a body of law in which a church, a school board or a town council could not be sure what was allowed until a judge told them. The justices themselves said as much. Scalia's ghoul line came in a concurring opinion in Lamb's Chapel v. Center Moriches Union Free School District in 1993, a case about a church that wanted to use school facilities after hours for a film series on family life. The school district denied the request, the Court ruled it must be allowed, and Scalia used the occasion to note that the test was invoked when it supported a result and ignored when it did not.

Death by a thousand footnotes

You can trace the test's decline through decades of opinions that politely walked around it. In the 1990s and 2000s, the Court narrowed the entanglement prong in Agostini v. Felton (1997), which reconsidered earlier rulings about public school teachers providing remedial services in parochial school buildings. In Mitchell v. Helms (2000) and Zelman v. Simmons-Harris (2002), the Court allowed more aid to flow to religious schools through neutral programs, including vouchers. In Town of Greece v. Galloway (2014) the Court upheld a town board's practice of opening meetings with prayer, and the opinion by Justice Anthony Kennedy relied on history and tradition rather than Lemon.

But the official end came in two steps.

Step one: American Legion, 2019. The American Legion v. American Humanist Association case concerned the Bladensburg Peace Cross, a large Latin cross on public land in Maryland, erected in the 1920s as a memorial to local men who had died in the First World War. Challengers argued that a cross on government property, maintained with public funds, violated the Establishment Clause. On June 20, 2019, the Court ruled 7-2 that it did not. Justice Samuel Alito's opinion explained that the Lemon test is poorly suited to old monuments, whose original purposes may be lost to history and whose meaning may have shifted over time. He wrote that the passage of time gives rise to a presumption of constitutionality for monuments, symbols and practices that have been in place a long time. Several justices went further in separate opinions. Justice Brett Kavanaugh wrote that Lemon does not apply to this kind of case and that the Court had long since moved away from it in many settings. Justice Clarence Thomas wrote that the test should be overruled.

Step two: Kennedy v. Bremerton, 2022. Three years later, in the coach's prayer case that I discuss in five school-day situations and what the prayer rulings really decide, Justice Neil Gorsuch's opinion for the 6-3 majority stated that the Court had long ago abandoned Lemon and its endorsement test offshoot. The opinion said the Establishment Clause must be interpreted by reference to historical practices and understandings. It did not say "we hereby overrule Lemon," and that matters for a point I will return to. But it said, in plain English, that the Court was not applying the test and that lower courts should not, either. The three dissenters, led by Justice Sonia Sotomayor, objected that the majority had in effect overruled decades of precedent without saying so.

What replaced it

So what does the Court use now? The answer, in Kennedy, is the standard from Town of Greece: look to historical practices and understandings. The question is whether the government's action fits within the tradition of practices that the Framers and their immediate successors accepted as consistent with the Establishment Clause, and whether it resembles the hallmarks of an establishment that the founding generation worried about. Those hallmarks include government control over the doctrine and personnel of a church, compulsory attendance and tax support for a state church, and the punishment of dissenters.

Justice Gorsuch, in the 2022 opinion, said the line between the permissible and the impermissible has to "accord with history and faithfully reflect the understanding of the Founding Fathers." You can see the logic. If the Framers hired chaplains, issued thanksgiving proclamations and allowed religious references in public life, a test that would strike all of that down could not be a good reading of the amendment they wrote. And the test has a conservative pedigree, one I explained in Originalism Explained for People Who Skipped Law School: the meaning of a constitutional provision is fixed when it is adopted, and courts are bound by it.

For the people who suffered under the old regime, the practical change is real. A small church that wants to rent a school gym, a town that wants to keep a decades-old monument, a religious school that wants to take part in a general aid program, and a public employee who wants to pray quietly on a break all stand on firmer ground than they did under a test that rewarded the most litigious plaintiff.

The strongest objection: history can be selective

Now, a fair-minded conservative has to deal with the best objection to the new approach, and it is this: history is not a neutral referee. The founding generation did not agree among themselves. Madison, as I describe in Baptists, Madison and the Virginia fight that put religious liberty first, had views that differed from those of Patrick Henry, and states kept their establishments for decades. A judge who can pick the right slice of the record can reach nearly any result. The risk is that "history and tradition" becomes a different kind of ghoul: a vague test that lets judges find in the past whatever they like. And critics add that the founding-era population was almost uniformly Protestant, so a standard tied to its practices may fit poorly with a country that now includes many faiths and none.

That objection has teeth. Here is my answer, in three parts.

1. The old test was selective too, with less discipline. Lemon asked judges to guess at purpose, to predict effect and to estimate entanglement. Each of those is a prediction about the future or a guess about a legislature's state of mind. History at least has a record that can be cited, checked and argued against.
2. A history test has to be tested by evidence, not by assertion. The remedy for selective history is better history: briefs that show the whole record, scholars who disagree in public, and judges who must explain why a particular practice does or does not match the tradition. That is a more transparent discipline than a three-prong formula that nobody could apply the same way twice.
3. The Court has built in limits. The standard is not "whatever was done in 1791 is allowed." It asks about the kind of practice, and it still forbids coercion, a state-run church and discrimination among faiths. The coercion principle did not disappear in Kennedy, and the majority said so.

I'd add one honest caveat. The Court has not said "we overrule Lemon" in so many words, and lower courts are still working out the details of the new approach. If you are dealing with a live case, read the most recent decisions in your own circuit, or ask a lawyer who handles religious liberty. This post describes the Supreme Court's direction, not a guarantee about any particular dispute. For the earlier history of the metaphor that Lemon tried to put into practice, see where 'a wall of separation' really comes from and what it was built for, and for the school-aid line of cases see Five cases that ended America's religious penalty in school aid.

The verdict

My verdict is that the burial was overdue and the replacement is better, though not perfect. The old test failed the one job a legal test is supposed to do: tell people in advance what the rules are. It punished churches for seeking ordinary public benefits, it gave monuments and crèches to the justices' personal tastes, and it made a whole area of law a lottery. The new approach is not free from error, but it rests on something solid: the text, the history and the practice of a country that never saw faith as a danger to be quarantined.

For fifty years, the Lemon test haunted churches, school boards and town councils. It is gone, and no one needs to hold a séance.

The bottom line

• Lemon v. Kurtzman (1971) created a purpose, effect and entanglement test, and it produced contradictory results for decades.
• The Court moved away from it in American Legion (2019) and said it had abandoned it in Kennedy v. Bremerton (2022), replacing it with history and tradition.
• If your church, school board or town is dealing with a religious display, a facility rental or an aid program, ask what the history and tradition of that kind of practice show.
• Keep the objection in mind: history can be used selectively, so insist on the whole record, not a convenient slice.
• Read the current decisions in your circuit before you act, because the law is still settling.