What the 1994 assault weapons ban said and why it was allowed to expire

A bayonet mount. That is the feature that made a rifle a criminal menace under the 1994 federal assault weapons ban. Not how fast it fired, not how hard it hit, not how many people it could hurt, but whether the manufacturer had welded a little lug under the barrel so that you could attach a knife. Congress looked at the epidemic of drive-by bayonet attacks sweeping the American heartland and decided to act.

I'm joking, but only a little, and the joke is the whole argument. The 1994 law is a case study in what happens when legislators regulate how a gun looks instead of what it does. It was hard to defend on the merits, it was easy to get around, and when the clock ran out in 2004 Congress let it die and the republic did not notice. In this post I want to tell you exactly what the law said, because most of the people who argue about it have never read it, and then explain why I think the lapse was the right outcome even for people who share the goal behind it.

What the law was, and when

The ban was Title XI, Subtitle A, of the Violent Crime Control and Law Enforcement Act of 1994, a big omnibus crime bill that President Clinton signed on September 13, 1994. The subtitle's own name was the Public Safety and Recreational Firearms Use Protection Act, which is a title that tries to have it both ways. It made it unlawful to manufacture, transfer or possess a "semiautomatic assault weapon," and it did the same for "large capacity ammunition feeding devices."

Two things to notice up front. First, the ban applied to the future. Guns lawfully possessed on the date of enactment stayed legal, which is called grandfathering, and you could keep, sell and pass them on. Second, the law had a sunset clause: it expired ten years after enactment, on September 13, 2004, unless Congress renewed it. That clause is not a footnote. It is the reason the ban could pass at all, and it is the reason it ended.

How the law defined its target

There were three ways a firearm could be an "assault weapon" under the statute.

By name. The law listed specific models, nineteen of them, including the Colt AR-15 and its variants, several AK-pattern rifles, the Uzi, the Galil, the FN-FAL and the TEC-9. If your gun carried one of those names, it was banned, whatever it looked like.

By features. For a semiautomatic rifle that accepted a detachable magazine, the law counted five features: a folding or telescoping stock, a pistol grip that protrudes conspicuously under the action, a bayonet mount, a flash suppressor or a threaded barrel to attach one, and a grenade launcher. Two or more of those, and the rifle was a banned assault weapon. Pistols and shotguns had their own feature lists.

By copies. Rifles that were copies or duplicates of the named models were covered.

Look at that list again, because it explains everything. A rifle with a fixed wooden stock and a bayonet lug was banned. The same rifle with the lug filed off and a plain stock was legal. A pistol grip counted; a thumbhole stock that did the same job did not. None of these features changes what happens when you pull the trigger. Every one of them is a handle, a mount or an accessory. Congress had taken a long list of what a gun looks like to a person who dislikes guns and made it a statute.

The difference nobody explains: assault rifle versus semi-automatic rifle

This is where the debate usually breaks down, so let me be careful. An assault rifle, in the sense used by the military, is a rifle that can fire in a fully automatic or burst mode, meaning it keeps firing for as long as you hold the trigger. Those guns were regulated heavily under the National Firearms Act of 1934, as I cover in Why the National Firearms Act of 1934 taxed guns instead of banning them, and the supply of new ones for civilians closed in 1986, which is the story in How the 1986 Firearm Owners Protection Act passed with a machine gun ban.

A semi-automatic rifle fires one round for each pull of the trigger, then loads the next round by itself. That describes most modern hunting rifles, many handguns and a large share of the shotguns in the country. The guns banned in 1994 were semi-automatic. They were the civilian cousins of military rifles, with similar looks and a very different action.

So the phrase "assault weapon" had a problem from day one. It sounded like a machine gun and it was not one. The law's drafters borrowed a term with a frightening military ring and applied it to guns that fire at the same rate as the deer rifle in your uncle's closet. I do not say that to be cute. It is a point about honesty in lawmaking. If you want to ban a firearm, say what it does and why the danger justifies it. Calling it by a scary name and then defining the name by its handle is the legislative version of a bait and switch.

The 10-round magazine limit

The second half of the law dealt with magazines. It prohibited the manufacture and transfer of "large capacity ammunition feeding devices," defined as magazines that could hold more than ten rounds. Magazines already in existence were grandfathered, just as the guns were.

I think the magazine provision was the part with the best argument. A defender of the ban would say, correctly, that a shooter who has to stop and reload gives people a few seconds, and a few seconds matters. I do not dismiss that. But the law froze new manufacture, not the existing stock, and there were already millions of lawful magazines in circulation. A ban on making new ones changes the price of the old ones. It does not change what a person with bad intentions can pick up. And a trained shooter changes a magazine in a matter of seconds. The theory was sound in principle and thin in application.

What the research said

The best news for supporters of the ban would have been a clear drop in gun crime. That is not what the Justice Department's own evaluations found.

The National Institute of Justice, part of the Justice Department, funded the major studies, which were led by the criminologist Christopher Koper. The first came in 1997, with updates in 1999 and a final report in 2004. The reports reached careful conclusions, and I will not make them stronger than they were. The authors found that the banned guns had been used in a small fraction of gun crimes before the ban, and that this fraction fell while the ban was in effect. But the lawfully owned stock of banned guns and magazines was so large that any effect was slow to appear, and the 2004 report concluded that the ban's effect on gun violence was likely to be small at best, too small for reliable measurement. Researchers who have looked at later data continue to disagree about it, and anyone who tells you the question is settled in either direction is selling something.

Here is the point that should bother both sides. Even a law's defenders could not point to a drop in crime that the ban plainly caused. If you are going to ask ordinary people to give up a category of lawful property, you owe them a result. Congress could not show one.

The ban was easy to evade, and the industry noticed

Manufacturers did what manufacturers do. They read the statute and sold what it allowed. A rifle that was banned with a bayonet lug, a flash suppressor and a folding stock was sold the next week without the lug, the suppressor or the folding stock. Same action, same cartridge, same magazines for the existing supply, and it was perfectly legal. Some companies also changed model names. A law that targets features invites a game of subtraction, and the manufacturers won it every time.

That is the practical flaw in feature-based regulation. It does not touch the mechanism, so a clever engineer can restore the look by adding an unregulated accessory. You are left with a statute that burdens the customer who wants a particular grip while having almost no effect on the function of the gun. I have never heard a good defence of that design from the people who passed it.

The sunset, September 13, 2004

The ban expired on schedule. Congress had built in the sunset on purpose, as the price of the votes needed to pass the bill, and in 2004 the question was whether to renew. Supporters argued for a renewal and the issue got a lot of airtime. Congress never sent a renewal bill to the President's desk, and on September 13, 2004 the law simply lapsed.

Why did the votes not materialize? I would offer three reasons, and I think they hold up. The research had not given supporters a clean case. The experience of the law showed how easy it was to sell around. And the political price was visible: the House had changed hands in the elections of November 1994, and many people in both parties, including some Democrats, connected the gun vote to that loss. How much of that outcome was caused by the gun vote and how much by the rest of the political environment is a question for historians, and I will leave it to them. What matters is that members of Congress who watched the aftermath did not rush to repeat it.

The Constitution check: "in common use"

In 2008 the Supreme Court decided District of Columbia v. Heller and held that the Second Amendment protects an individual right to keep and bear arms, which I discuss at length in Why Heller got it right and the 'militia only' reading falls apart. The opinion, written by Justice Scalia, also said something that matters here. It tied the protection to arms "in common use at the time" for lawful purposes, and it said the amendment does not protect dangerous and unusual weapons. The Court even remarked, in a passage that has drawn endless commentary, that weapons like the M-16 rifle could be banned.

Both sides read that language as support. The supporters of bans say that anything military-style can be restricted. The opponents say that the most popular rifle platform in the country, owned by millions of Americans for target shooting, hunting and defence of the home, is in common use by any ordinary meaning of the words, and that a law banning it in 1994 would face a serious problem today. I think the opponents have the better of it. A feature-based ban hits guns that people own by the million, not the exotic, unusual weapons the Court had in mind. The test the Court wrote was about how Americans actually use a firearm, and the numbers in the American home are on the side of the owners.

The later decisions add to this. Bruen demands that a modern restriction be consistent with the country's historical tradition of regulation, and Rahimi confirmed that the test looks at principles, not exact historical twins. I walk through the new test in Bruen and Rahimi explained, the history test every gun law now faces. A ban defined by the shape of a pistol grip is going to have a hard time finding a tradition to match.

The strongest objection, answered

The best case against my position goes like this. One: mass shootings are rare but terrifying, and a few weapons and large magazines have been used in the worst of them. Two: even a small reduction in casualties is worth a modest inconvenience to gun buyers. Three: the evasion problem is a drafting problem, which Congress could fix by writing a better law, perhaps by regulating function rather than features. Four: the research was thin because the sample was limited, not because the theory was wrong.

I'll take those one at a time. On the first, I accept the premise and reject the leap. A law should be judged by what it does, and the evidence did not show it prevented much. On the second, the inconvenience is not modest when the thing banned is a lawful category of property owned by millions of people who never hurt anyone. On the third, I agree that a function-based approach would be a more honest design, and notice what that concession means: the 1994 law as written was the wrong tool, and the sunset was the correct verdict on it. A better-drafted bill would face the Heller and Bruen tests, and I think it would lose on the merits for the same reason. On the fourth, an untested theory is not a basis for taking property. If a measure's supporters cannot show that it works, the burden falls on them.

What this means for a family

If you own one of these rifles, you own a semi-automatic rifle, a design that has been sold to the American public for generations, and the 1994 experiment tells you something useful: the federal government ran the test, spent public money measuring it, and walked away. If you are asked to support the next version of a ban, ask the three questions that the 1994 law could not answer. What does the banned gun do that a legal one does not? What did the research show about the last attempt? And does the proposal define its target by function or by furniture?

The bottom line

• The 1994 law, signed on September 13, 1994, banned semi-automatic rifles by name and by features such as a bayonet mount, and banned new magazines holding more than ten rounds.
• A semi-automatic rifle fires once per trigger pull. An assault rifle in the military sense fires automatically and has been heavily regulated since 1934.
• The Justice Department-funded evaluations found at best a small effect, and the law was easy to evade by dropping cosmetic features.
• It expired on September 13, 2004 because Congress built in a sunset and did not renew it.
• When someone proposes a new ban, ask whether it describes how the gun works or how it looks, and hold lawmakers to the Constitution's "in common use" test.