Criminal Law & Procedure

An ordinance, and eight people arrested under it

Jacksonville, Florida, 1972. The city's vagrancy ordinance, copied almost word for word from English statutes hundreds of years old, made it a crime to be a rogue, a vagabond, a common night walker, a habitual loafer, a person using juggling, or a person wandering or strolling around from place to place without any lawful purpose or object. Margaret Papachristou and Betty Calloway, both white, were riding in a car with Eugene Melton and Leonard Johnson, both Black, on a main street in Jacksonville. They were on their way to a nightclub. Police stopped the car and booked all four for prowling by auto.

The Supreme Court threw the ordinance out unanimously in Papachristou v. City of Jacksonville, 405 U.S. 156 (1972). Justice Douglas did not say the city lacked power to punish loitering. He said this law was not usable as law. It failed to give an ordinary person notice of what was forbidden, and it handed police a blank warrant to arrest whoever displeased them, which is precisely what four people driving to a nightclub in a racially mixed car found out.

That single case contains the two questions this lesson answers. Where does a criminal prohibition come from, and what must it look like before a court will let the state enforce it? Get those right and the rest of the course has a floor to stand on.

One boundary before we start. This is general education about how United States criminal law works. It is not legal advice, it does not create a lawyer-client relationship, and no lesson here can substitute for a licensed attorney in your jurisdiction, because the answer to almost every real question begins with which state you are in.

Four sources, in the order they matter

American criminal law arrives from several directions at once, and knowing which source governs is often half the analysis.

Statutes come first. Today essentially every crime you can be convicted of is defined by a legislature in writing. Each state has a penal code; Congress has Title 18 of the United States Code and hundreds of criminal provisions scattered through other titles. This was not always so. English and early American courts recognised common law crimes, offences judges declared and defined themselves. Federal courts lost that authority early: in United States v. Hudson and Goodwin, 11 U.S. 32 (1812), the Supreme Court held that federal courts have no common law criminal jurisdiction, so there is no federal crime unless Congress wrote one. Most states abolished common law crimes by statute during the twentieth century, though a handful retain them in residual form.

The common law is still everywhere anyway. Abolishing common law crimes did not abolish common law meaning. When a statute says murder and does not define malice aforethought, courts reach for centuries of decisions that do. Terms like burglary, larceny, and assault come to modern codes pre-loaded with old content. You will meet this again and again: a two-line statute resting on two hundred years of case law.

The Model Penal Code is the most influential text nobody enacted. The American Law Institute, a private body of judges, lawyers, and academics, spent from 1952 to 1962 drafting a model criminal code. No jurisdiction is bound by it, because the ALI cannot legislate. Yet more than thirty states rewrote their criminal codes in the decades after 1962 and borrowed heavily from it, and courts everywhere cite it when a statute is ambiguous. Its greatest contribution, the four culpable mental states, is the subject of a lesson of its own.

Constitutions set ceilings, not floors. The federal Constitution never creates a crime. It limits what legislatures may criminalise and how the state may prove a case. State constitutions do the same and sometimes go further, protecting more than the federal minimum. The site's Constitutional Law course, LAW 310, works the constitutional doctrines themselves; here we take them as constraints on criminal law and procedure.

Two smaller sources round out the list. Administrative regulations can carry criminal penalties when a statute authorises them, which is how environmental and food safety violations become prosecutable. And municipal ordinances, like the Jacksonville one, create local offences enforceable by local police.

The point: Crimes are defined by statute, interpreted through inherited common law meanings, shaped by the Model Penal Code, and bounded by constitutions.

Federal or state, and why it usually does not matter

Students often assume criminal law is federal. It overwhelmingly is not. States hold the general police power, meaning the authority to legislate for health, safety, welfare, and morals, and roughly nine in ten criminal prosecutions in the United States are brought under state law. Congress has no general police power; every federal crime must be tied to an enumerated power, most often the commerce power, the taxing power, or federal property and personnel.

The practical consequence is that murder, robbery, burglary, and assault are prosecuted in state courts by county district attorneys, while federal prosecutors concentrate on offences with an interstate or federal hook: drug trafficking networks, immigration, firearms, fraud that crosses state lines, public corruption, and crimes on federal land. Some conduct violates both, and prosecution by both sovereigns does not violate double jeopardy under the dual sovereignty doctrine, which the Court reaffirmed in Gamble v. United States, 587 U.S. 678 (2019). The system's institutions, from police departments to corrections, are surveyed in the site's Introduction to Criminal Justice course, CRIJ 1301; this course stays on the law.

The legality principle

Here is the oldest idea in criminal law, usually given in Latin: nullum crimen sine lege, nulla poena sine lege. No crime without law, no punishment without law. Strip the Latin and it says something demanding. The state may punish you only for conduct that a written rule, in force before you acted, clearly forbade.

Four corollaries follow, and each one is enforceable in court.

1. Crimes must be defined in advance and in writing. A judge may not invent an offence to punish conduct everyone agrees is wicked.
2. Criminal statutes may not operate retroactively. This is the ex post facto ban.
3. Criminal statutes must be reasonably clear. This is the void for vagueness doctrine that killed the Jacksonville ordinance.
4. Ambiguity is resolved in favour of the defendant. This is the rule of lenity.

The principle is not squeamishness about punishing bad people. It is a rule about power. A vague or retroactive criminal law gives officials discretion to select their targets after the fact, and selection after the fact is how criminal law becomes a tool against disfavoured people rather than against dangerous conduct.

Why this matters: Legality converts punishment from an exercise of official will into the application of a rule that existed before the defendant acted.

Vagueness: the test and two modern examples

A criminal statute is void for vagueness under the Due Process Clause if it fails either of two tests: it does not give a person of ordinary intelligence fair notice of what is prohibited, or it invites arbitrary and discriminatory enforcement. Courts care more about the second prong than students expect. Notice matters, but the deeper worry is a law so open that the officer on the street, not the legislature, decides what the crime is.

In City of Chicago v. Morales, 527 U.S. 41 (1999), Chicago made it an offence to remain in a public place with no apparent purpose after an officer, believing at least one person present was a gang member, ordered the group to disperse. Police issued tens of thousands of dispersal orders and made thousands of arrests. The Court struck the ordinance down: no apparent purpose gave citizens no way to know what was forbidden and gave officers absolute discretion.

Vagueness also reaches sentencing statutes. In Johnson v. United States, 576 U.S. 591 (2015), the Court held that the residual clause of the Armed Career Criminal Act, which raised sentences for prior felonies involving conduct that presents a serious potential risk of physical injury to another, was unconstitutionally vague. The clause had produced years of irreconcilable decisions about which crimes qualified, which was itself evidence that no one could tell.

Compare a statute that survives. A law forbidding driving with a blood alcohol concentration of 0.08 percent or more is unambiguous at the edges as well as the centre. That is the target: a rule you could obey deliberately.

Remember: Vagueness is measured by notice and by the discretion the statute hands to enforcers, and the second failure is usually the fatal one.

Ex post facto and retroactive judicial expansion

Article I forbids both Congress and the states to pass any ex post facto law. In Calder v. Bull, 3 U.S. 386 (1798), Justice Chase gave the four categories still used today. A law is ex post facto if it (1) criminalises conduct that was innocent when done, (2) aggravates a crime after the fact, (3) increases the punishment for a crime after it was committed, or (4) alters the rules of evidence to make conviction easier on less testimony than the law required when the act was done.

Note what the clause does not cover. It restrains legislatures, not courts. So what stops a court from reading an old statute in a new and broader way to reach a defendant who could not have anticipated it? The Due Process Clause does. In Bouie v. City of Columbia, 378 U.S. 347 (1964), South Carolina's courts expanded a trespass statute that punished entry after notice so that it also covered remaining after being asked to leave, and applied that reading to civil rights sit-in demonstrators. The Supreme Court reversed: an unforeseeable judicial enlargement of a criminal statute, applied retroactively, does the same work as an ex post facto law and is equally forbidden. The doctrine has been narrowed since, notably in Rogers v. Tennessee, 532 U.S. 451 (2001), where a divided Court allowed abolition of the old year and a day rule in homicide to apply to the defendant before it, and the line between routine common law development and forbidden expansion remains genuinely contested.

Lenity, and why an airplane was not a motor vehicle

When a criminal statute is genuinely ambiguous after a court has exhausted the ordinary tools of interpretation, the rule of lenity resolves the ambiguity in the defendant's favour. It is a tiebreaker, not a thumb on the scale, and courts reach it last.

The classic illustration is McBoyle v. United States, 283 U.S. 25 (1931). McBoyle was convicted of transporting a stolen aircraft across state lines under the National Motor Vehicle Theft Act, which defined a motor vehicle to include an automobile, automobile truck, automobile wagon, motorcycle, or any other self-propelled vehicle not designed for running on rails. An airplane is self-propelled and does not run on rails. Justice Holmes still reversed. In everyday speech a motor vehicle runs on land, and, he wrote, a fair warning should be given to the world in language that the common world will understand. Congress could easily have said aircraft. It had not, so McBoyle went free and Congress amended the statute the next year.

Notice how tidy the division of labour is. Lenity does not say airplane theft should be lawful. It says that if the legislature wants to punish it, the legislature must say so.

Bottom line: Vagueness, ex post facto, and lenity are three enforcement mechanisms for one idea: the rule must exist, and be findable, before the conduct.

Criminal, civil, and the burden between them

The same punch can produce two cases. In the criminal case the state prosecutes for battery, the standard of proof is beyond a reasonable doubt, and the sanction is a fine, probation, or incarceration plus a conviction record. In the civil case the injured person sues for the tort of battery, the standard is a preponderance of the evidence, and the remedy is money. The two are independent, which is why an acquittal does not bar a later civil suit.

Feature	Criminal case	Civil case
Who brings it	The government, in the public's name	The injured party
Burden of proof	Beyond a reasonable doubt	Preponderance of the evidence
Outcome	Guilty or not guilty; punishment	Liable or not liable; damages
Counsel	Appointed if the defendant cannot pay	No right to appointed counsel
Jury verdict	Unanimous in all serious cases	Often less than unanimous

The heavy criminal burden is constitutional, not merely traditional. In re Winship, 397 U.S. 358 (1970), held that due process requires proof beyond a reasonable doubt of every fact necessary to constitute the crime charged. The reason is asymmetry of error: a system that convicts the innocent inflicts a harm the state cannot undo, so it accepts that some guilty people will walk.

Offences are also graded by seriousness. A felony is typically punishable by more than a year of imprisonment, a misdemeanour by up to a year in a local jail, and an infraction or violation by fine alone. That line drives everything downstream: which court hears the case, whether a grand jury is involved, whether counsel must be appointed, and what collateral consequences follow a conviction.

Common misconceptions

• Judges can create crimes when conduct is clearly harmful. Federal courts never could after 1812, and nearly all states abolished common law crimes; a legislature must act first.
• A vague law is fine as long as the defendant clearly knew it was wrong. Vagueness is judged by the statute's notice and by the discretion it gives enforcers, not by whether this defendant felt guilty.
• The rule of lenity means close cases go to the defence. Lenity applies only after the ordinary interpretive tools leave a genuine ambiguity, which is rarer than defence briefs suggest.
• The Model Penal Code is the law. It is a model drafted by a private organisation; it governs only where a legislature enacted it, though its influence is enormous.
• Being acquitted means the conduct never happened. An acquittal means the state failed to prove guilt beyond a reasonable doubt; a civil jury may still find the same conduct more likely than not.

Where this leaves us

• Criminal prohibitions come from statutes, read against inherited common law meanings, influenced by the Model Penal Code and bounded by constitutions.
• States hold the general police power and bring the large majority of prosecutions; federal crimes need an enumerated hook.
• The legality principle requires rules that are written, prospective, clear, and strictly construed.
• Papachristou, Morales, and Johnson show vagueness doctrine at work, and its sharpest concern is enforcement discretion.
• Calder sets the four ex post facto categories; Bouie extends the idea to unforeseeable judicial expansion.
• McBoyle shows lenity: if the legislature wants to punish it, the legislature must say so.
• Criminal cases are brought by the state and require proof beyond a reasonable doubt, a rule Winship made constitutional.

Sources

1. Legal Information Institute. (n.d.). Criminal law. Cornell Law School. law.cornell.edu
2. Legal Information Institute. (n.d.). Void for vagueness. Cornell Law School. law.cornell.edu
3. Britannica. (2024). Criminal law. britannica.com
4. Wikipedia. (n.d.). Papachristou v. City of Jacksonville. en.wikipedia.org
5. Congressional Research Service. (n.d.). Constitution Annotated: Article I, ex post facto clauses. constitution.congress.gov

A man carried onto a highway

Alabama, 1944. Officers came to Martin's home, arrested him there, and took him out onto a public highway. On the highway he did what drunk men do: he shouted, he swore, he staggered. He was then convicted of being drunk on a public highway and manifesting a drunken condition by loud and profane language.

Every element of the offence was literally satisfied. He was drunk. He was on a public highway. He was loud. The Alabama Court of Appeals reversed the conviction anyway, in Martin v. State (Ala. Ct. App. 1944), with a sentence that is now in every criminal law casebook: the statute presupposes a voluntary appearance. Martin did not appear on the highway. He was carried there by the state that then prosecuted him for being there.

Start from the wrong answer and you learn more than from the right one. The wrong answer is that a crime is a list of elements and the prosecution wins by ticking them off. What Martin exposes is a requirement that hides underneath the list: before any element matters, there must be conduct that was the defendant's own. That requirement is the actus reus, and this lesson takes it apart.

What actus reus actually includes

Actus reus is usually translated as the guilty act, which is a bad translation because it smuggles in blame that belongs to the mental element. A better working definition: the actus reus of an offence is the voluntary act (or a qualifying omission) plus any result the offence requires and any attendant circumstances that must be present.

Take burglary of a dwelling at night. The act is breaking and entering. The attendant circumstances are that the structure is a dwelling, that it belongs to another, and that it is night. There is no required result at all: the burglar who finds the house empty and takes nothing has still completed the offence. Now take homicide. The act might be pulling a trigger; the required result is the death of a human being; the attendant circumstance is that the victim was alive when the defendant acted. Sorting a statute into these three buckets is not busywork. In the next lesson you will see that the required mental state can differ from bucket to bucket within a single crime.

The core of it: Actus reus is conduct plus results plus circumstances, and the conduct component must be voluntary.

What makes an act voluntary

Voluntary here does not mean freely chosen in any philosophical sense. A person who hands over a wallet at gunpoint acts voluntarily in the criminal law's sense; his defence is duress, which comes later. The law's meaning is narrow and physical. Model Penal Code section 2.01 says a person is not guilty unless liability is based on conduct that includes a voluntary act or an omission he was physically capable of performing, and it then lists what does not count:

• a reflex or convulsion;
• a bodily movement during unconsciousness or sleep;
• conduct during hypnosis or resulting from hypnotic suggestion;
• any bodily movement that is not a product of the effort or determination of the actor, either conscious or habitual.

Notice the word habitual. Driving home on autopilot and shifting gears without thinking is still voluntary. The test is not whether you deliberated, but whether the movement came from you.

Two illustrations mark the boundary. A driver has a heart attack and his car kills a pedestrian: the movement of the car after the attack was not his act, and he is not guilty of a homicide based on that movement. Now change the facts. In People v. Decina (N.Y. 1956), the defendant knew he was subject to epileptic seizures and drove anyway; a seizure caused a crash that killed four children. He was properly convicted, and the reasoning shows a move you will use constantly: the court simply widened the time frame. The seizure was not voluntary, but getting behind the wheel while knowing about the seizures was, and the state charged that act instead.

The same time-framing move decides the sleepwalking cases, which are real and rare. In 1987 a Toronto man named Kenneth Parks drove roughly fourteen miles while apparently asleep, killed his mother-in-law, and then presented himself at a police station saying he thought he had killed someone. Sleep experts testified for the defence; a jury acquitted him, and the Supreme Court of Canada upheld the acquittal in 1992. Cases like that make people uneasy, and they should be scrutinised hard, because the claim is unfalsifiable in the wrong hands. But the principle behind them is not soft: the criminal law punishes people for what they do, and a body moving during sleep is not doing.

So what?: When the harmful movement was involuntary, look earlier for a voluntary act that created the risk, and charge that.

Thoughts are not crimes, and neither is who you are

You cannot be punished for wanting someone dead. The line is drawn at conduct for two reasons: proof, because a system that punished intentions would need evidence it cannot honestly obtain, and liberty, because a person who forms a plan and abandons it has harmed no one. This is why attempt, which we cover later, requires an overt act and not merely a resolved mind.

Status is the harder case. In Robinson v. California, 370 U.S. 660 (1962), California made it a misdemeanour to be addicted to the use of narcotics, carrying a minimum of ninety days in jail. The statute did not require proof that the defendant had used, possessed, or bought anything in California. The Supreme Court held it violated the Eighth Amendment: a state may punish the acts of using or selling drugs, but it may not make the illness itself the offence. Punishing a person for a condition, Justice Stewart reasoned, is like making it a crime to be mentally ill or to have a common cold.

Six years later the Court declined to extend that. Powell v. Texas, 392 U.S. 514 (1968), involved Leroy Powell, convicted of being found in a state of intoxication in a public place. He argued he was a chronic alcoholic and could not stop himself. A divided Court affirmed the conviction. The plurality drew the distinction the doctrine has run on ever since: Texas punished Powell for appearing drunk in public, an act, not for the status of being an alcoholic.

That distinction thins out under pressure, as anyone who has read the dissents can see, and the pressure is currently on homelessness. Lower courts had read Robinson to bar punishing people for sleeping outside when no shelter beds existed, on the theory that sleeping is unavoidable and the defendant's only real offence was being unhoused. In City of Grants Pass v. Johnson (2024) the Supreme Court rejected that reading, holding that generally applicable public camping ordinances regulate conduct rather than status and so do not violate the Cruel and Unusual Punishments Clause. Whether you find that persuasive depends on how firmly you think the act and status line can hold when the act is one no human being can decline to perform. The doctrine is settled for now; the argument is not.

Possession, which is a state of affairs the law calls an act

Possession offences are everywhere: drugs, firearms, stolen property, contraband. Possession looks like a status rather than an act, which is why the Model Penal Code takes the trouble to define it in. Under section 2.01(4), possession is an act if the possessor knowingly procured or received the thing, or was aware of his control of it for a period long enough to have been able to terminate the possession.

That second clause matters. Someone slips a package into your coat pocket on a bus; you discover it two stops later and get off. You did not knowingly receive it, and you have not had a realistic chance to end the possession. If instead you carry it home, look at it, and put it in a drawer, you possess it.

Two further terms travel with possession. Actual possession means the item is on your person. Constructive possession means you have the power and the intention to control it, even at a distance: drugs in your locked glove compartment, a gun in your bedroom closet. Constructive possession does real work in prosecutions and also produces real injustice, because in a car with four occupants and a pistol under a seat, the theory can reach everyone in the car. Courts respond by requiring more than mere presence or proximity, but juries are asked to draw the line on thin facts.

Omissions: when doing nothing is the actus reus

The general rule in American criminal law is blunt and often surprises students. There is no legal duty to rescue. A strong swimmer can stand on a dock, watch a stranger's child drown in three feet of water, do nothing, and commit no crime, however contemptible we find him. Most European systems disagree and impose a general duty of easy rescue; a handful of American states, including Vermont and Minnesota, have enacted narrow duty-to-assist statutes, usually with small penalties.

The common law instead recognises specific duties. The standard list, drawn from Jones v. United States (D.C. Cir. 1962) and expanded by later courts, has five sources:

1. Statute. Tax filing, accident reporting, mandated child abuse reporting.
2. Status relationship. Parent to child, spouse to spouse; sometimes an employer to an employee in peril.
3. Contract. A nurse, lifeguard, or paid caregiver has undertaken the duty by agreement. In Commonwealth v. Pestinikas (Pa. Super. Ct. 1992), a couple who contracted to feed and care for an elderly man, then withheld food until he died, were convicted of murder on exactly this theory.
4. Voluntary assumption of care that isolates the victim. If you begin a rescue and thereby stop others from helping, you may not abandon it.
5. Creation of the peril. If you push someone into a river, even accidentally, you have a duty to try to get them out.

The narrowness is illustrated by People v. Beardsley (Mich. 1907). Beardsley spent a weekend drinking with a woman who was not his wife; she took morphine in his presence and he had her moved to a neighbour's basement, where she died. The court found no duty, because she was not his spouse and he had assumed no care. It is a repellent result that states the rule accurately.

A cultural note worth correcting while we are here. The 1964 killing of Kitty Genovese in Queens is often taught as proof that thirty-eight witnesses watched and did nothing. Later investigation established that the original newspaper account was substantially wrong: far fewer people saw any part of the attack, several did call police, and a neighbour came to her aid. The bystander effect is a genuine finding in psychology; the Genovese story as usually told is not the evidence for it.

In short: Omission liability requires a legal duty from statute, status, contract, assumption, or creation of the peril, plus the physical ability to act.

The act and omission line, tested at a bedside

Now a case that will not resolve itself with intuition. In Barber v. Superior Court (Cal. Ct. App. 1983), physicians disconnected a comatose patient's ventilator and later his intravenous feeding, at the family's request, and prosecutors charged them with murder. Turning a dial and pulling a tube certainly look like acts. The court held otherwise: withdrawing life support is properly analysed as the cessation of treatment, an omission, and because a physician has no duty to continue treatment that has become futile, there was no unlawful failure to act and therefore no actus reus.

Sit with the reasoning rather than the result. It is not that the doctors did nothing; they moved their hands. It is that the law asked whether they breached a duty to continue, and found none. That reframing shows what the act and omission distinction really is: not a description of physical motion, but a question about which duty the state says you broke.

Common misconceptions

• Any bodily movement is a voluntary act. Reflexes, convulsions, movement during sleep or unconsciousness, and hypnotic conduct are excluded, though a court may find an earlier voluntary act that created the risk.
• Acting under threat is involuntary. Handing over money at gunpoint is a voluntary act; the answer is the duress defence, not the absence of actus reus.
• You can be prosecuted for a plan you never acted on. Criminal law requires conduct; attempt and conspiracy each demand an overt step beyond the thought.
• Anyone who watches a stranger die commits a crime. American law imposes no general duty to rescue; liability requires one of the recognised duty sources.
• Possession is not an act, so possession crimes are unconstitutional. Codes define possession as an act where the person knowingly received the item or had time to end the possession.
• Robinson means addiction is a complete defence. Robinson bars punishing the status itself; Powell allows punishing acts such as public intoxication even when they flow from addiction.

The short version

• Actus reus is a voluntary act or qualifying omission, plus any required result and attendant circumstances.
• Martin shows that literal element satisfaction is not enough when the state supplied the conduct.
• The Model Penal Code excludes reflexes, convulsions, sleep, unconsciousness, and hypnosis; habitual movements still count.
• Decina and the sleepwalking cases turn on time framing: look for an earlier voluntary act.
• Thoughts and status cannot be crimes; Robinson, Powell, and Grants Pass mark a line that is still argued over.
• Possession counts as an act by definition, and constructive possession is where the litigation happens.
• Omissions are criminal only with a duty from statute, status, contract, assumption of care, or creation of the peril.

Sources

1. Legal Information Institute. (n.d.). Actus reus. Cornell Law School. law.cornell.edu
2. Wikipedia. (n.d.). Robinson v. California. en.wikipedia.org
3. Wikipedia. (n.d.). Powell v. Texas. en.wikipedia.org
4. Britannica. (2024). Criminal law: The elements of crime. britannica.com
5. Legal Information Institute. (n.d.). Model Penal Code. Cornell Law School. law.cornell.edu

Three tons of bomb casings, sold for eighty-four dollars

Joseph Morissette was a junk dealer in Michigan and a veteran of the Second World War. In 1948 he went deer hunting on a tract of land the Air Force used as a practice bombing range. He got no deer. What he did find were heaps of spent bomb casings, rusting in the weeds where they had lain for years, some of them stacked by government crews and left. He loaded three tons of them onto his truck in broad daylight, flattened them, sold them at a market, and cleared eighty-four dollars.

He was prosecuted under a federal statute punishing whoever knowingly converts property of the United States, and he testified that he thought the casings were abandoned. The trial judge would not let the jury consider that. He instructed that the only intent required was the intent to take the casings, which Morissette plainly had, and Morissette was convicted.

A unanimous Supreme Court reversed in Morissette v. United States, 342 U.S. 246 (1952). Justice Jackson's opinion is the best short essay on mens rea in the reports. Crime, he wrote, is generally constituted only from concurrence of an evil-meaning mind with an evil-doing hand, and when Congress borrows a common law offence like conversion without saying anything about intent, silence is not an abolition of the intent requirement. It is Congress assuming the requirement is already there.

That is the subject of this lesson: not whether the defendant did it, but what he must have had in his head while doing it, and about which parts of the offence.

The old vocabulary, and why it collapsed

For centuries the common law used a scattered inventory of mental state words: maliciously, wilfully, wantonly, feloniously, corruptly, with a depraved heart, with intent to defraud. There was no agreed hierarchy and no agreed content. Courts also split crimes into general intent (intent to do the prohibited act) and specific intent (an additional purpose beyond the act, as in burglary's breaking and entering with intent to commit a felony inside). Those two labels still appear in state opinions and still matter, because some defences, notably voluntary intoxication, are available only against specific intent crimes.

The problem was that nobody could reliably say what any given word meant in any given statute. The Model Penal Code's drafters solved it by throwing the vocabulary out and replacing it with four defined terms, arranged from most to least culpable. This is the single most successful law reform project of the twentieth century in the United States, and most state codes now use some version of it.

The four states, defined

Mental state	Definition	Test question
Purposely	It is the actor's conscious object to engage in the conduct or to cause the result.	Did he want it?
Knowingly	He is aware that his conduct is of that nature or that the circumstance exists; as to results, he is aware it is practically certain to follow.	Did he know it would happen, even if he did not care?
Recklessly	He consciously disregards a substantial and unjustifiable risk, and doing so is a gross deviation from the standard a law-abiding person would observe.	Did he see the risk and run it anyway?
Negligently	He should be aware of a substantial and unjustifiable risk, and his failure to perceive it is a gross deviation from the care a reasonable person would use.	Should he have seen the risk?

The important structural line falls between recklessness and negligence. Recklessness is subjective: the defendant actually noticed the risk. Negligence is objective: he did not notice, and a reasonable person would have. Everything above the line involves awareness; everything below it involves the failure to be aware. That is why criminal negligence requires a gross deviation and not the ordinary carelessness that suffices in tort.

One worked contrast makes it stick. A man fires a rifle into a crowded room. If he wanted to kill the person he hit, that is purpose. If he aimed at someone else but knew the bullet would pass through and hit the victim, that is knowledge. If he fired at a wall to frighten people, consciously aware that a ricochet might kill and firing anyway, that is recklessness. If he genuinely never considered that a bullet might ricochet, when any reasonable adult would have, that is negligence.

Key idea: Purpose is desire, knowledge is practical certainty, recklessness is conscious risk taking, and negligence is the failure to perceive a risk you should have seen.

Element analysis, the move most students skip

Here is where the Model Penal Code is genuinely different from what came before. A crime is not one thing with one mental state attached. It is a set of material elements, and a mental state must be found for each of them. Three rules run the machinery:

• Section 2.02(3), the default. If a statute defines an element without saying which mental state applies, the state is established if the person acted purposely, knowingly, or recklessly. Recklessness is the floor.
• Section 2.02(4), the travel rule. When a statute states a single mental state without distinguishing among elements, that state applies to every material element unless a contrary purpose plainly appears.
• Section 2.02(5), the ladder. A higher mental state satisfies a requirement of a lower one. Proving purpose proves knowledge, recklessness, and negligence.

Now run a statute through it. Take: whoever knowingly receives stolen property with a value exceeding one thousand dollars is guilty of a felony. Break it into elements. The conduct element is receiving. The first attendant circumstance is that the property was stolen. The second attendant circumstance is that its value exceeded one thousand dollars. Under the travel rule, knowingly attaches to all three unless the statute says otherwise. So the prosecution must prove the defendant knew he was receiving, knew the property was stolen, and knew the value exceeded a thousand dollars.

Change one input and watch the result flip. Suppose the defendant bought a laptop for forty dollars from a stranger in a parking lot and told himself nothing. Value is easy: laptops cost more than a thousand dollars only sometimes, and if this one did not, the felony grade fails regardless of anything else. But suppose the laptop was worth two thousand and he knew it. Did he know it was stolen? He never asked. That is not automatically knowledge, which is precisely why the doctrine in the next section exists.

This is not an academic exercise. In Rehaif v. United States (2019), the Supreme Court held that to convict someone of unlawful firearm possession by a prohibited person, the government must prove the defendant knew he possessed a firearm and knew he belonged to the prohibited category. For decades, prosecutors had proved only the first. The Court applied element analysis and found the second element had been ignored. In Elonis v. United States (2015), a man was convicted for violent Facebook posts about his estranged wife under a jury instruction asking only whether a reasonable person would view them as threats. The Court reversed: a negligence standard was not enough for a federal criminal conviction when the statute was silent.

What matters here: Never ask what mental state a crime requires. Ask what mental state each element requires.

Wilful blindness

Back to the parking lot laptop. Courts will not let a defendant manufacture ignorance. Under the wilful blindness or ostrich doctrine, knowledge can be established where the defendant subjectively believed there was a high probability that the fact existed and took deliberate actions to avoid learning it. The Supreme Court stated that two-part test in Global-Tech Appliances v. SEB (2011), adopting a criminal law doctrine that had developed in cases like United States v. Jewell (9th Cir. 1976), where a driver was paid to take a car across the border and pointedly did not look inside a hidden compartment that held marijuana.

Note the limits, which juries are supposed to be instructed on. Recklessness is not enough. Suspicion is not enough. The defendant must have almost known and then chosen not to confirm it. A courier who was merely careless has not been wilfully blind.

Strict liability, and why Staples got his gun back

Strict liability offences dispense with mens rea for at least one element. They are disfavoured but real, and they cluster in two places. The first is public welfare offences: regulatory crimes involving food, drugs, hazardous materials, and traffic, usually carrying small penalties and no serious stigma, where the Court has accepted liability without fault since United States v. Balint (1922). The second is statutory rape in many states, where a reasonable mistake about the victim's age is no defence, a rule that draws consistent academic criticism and consistent legislative retention.

The best case for seeing the boundary is Staples v. United States, 511 U.S. 600 (1994). Staples possessed an AR-15 rifle that had been modified so that it fired automatically, making it a machine gun under the National Firearms Act. He testified he had never fired it automatically and did not know it could. The trial court told the jury his knowledge of that characteristic was irrelevant. The Supreme Court reversed. Guns, unlike grenades or narcotics, are lawfully owned by millions of Americans, so possession does not itself put an owner on notice that he is entering a regulated field; and the offence carried up to ten years, a penalty inconsistent with the small fines typical of public welfare offences. Silence in the statute therefore did not mean strict liability.

Two signals emerge for any statute you are asked to read: how severe is the punishment, and is the regulated item or activity one that already announces itself as dangerous and regulated? Heavy penalty plus ordinary item points strongly toward a mens rea requirement.

The upshot: Courts presume a mental state even in silent statutes, and the presumption strengthens as penalties rise and as the conduct looks more like everyday life.

Transferred intent, concurrence, and motive

Transferred intent handles bad aim. A shoots at B, misses, and kills C. A is guilty of murdering C; the intent follows the bullet. The doctrine applies within the same kind of harm. Intent to shoot a person does not transfer to the crime of destroying the window the bullet went through, which requires its own analysis.

Concurrence requires that the mental state and the act line up in time, and that the mental state actually motivate the conduct. A woman spends a month planning to kill her business partner, then accidentally runs him over on Tuesday before she has begun to carry out the plan. She is not a murderer, because the intent did not produce the act. Reverse the order and the problem returns: a man burgles a house, and only afterwards forms the intent to steal what he already carried out. Common law burglary required intent at entry, so timing decides the charge.

Motive is the reason behind the intent, and it is not an element of any offence. A man who steals bread to feed his child intends to steal; his motive may move a prosecutor's discretion, a jury's sympathy, or a judge's sentence, but it does not defeat the mens rea. The converse is true too: a good motive does not create a defence, and a bad motive does not supply a missing element. Motive is evidence, not law.

Causation, in two steps

For result crimes, the state must also connect conduct to result. Do it in two steps and you will not get lost.

Step one, actual cause. Would the result have occurred but for the defendant's conduct? If yes, causation fails here. Where two independent forces would each have sufficed, courts use a substantial factor test instead.

Step two, proximate cause. Is it fair to hold the defendant responsible, given what intervened? The Model Penal Code frames it as whether the actual result is too remote or accidental to have a just bearing on liability. Standard applications: an eggshell victim does not break the chain, so a punch that kills a man with a thin skull is still a homicide; ordinary medical negligence in treating the wound does not break the chain, but grossly abnormal treatment may; a victim's refusal of a blood transfusion on religious grounds does not break the chain; and a victim's foreseeable attempt to escape does not break it either.

Here is the pattern to memorise: an intervening cause that is dependent on the defendant's act, meaning a natural response to it, almost never relieves him. An independent intervening cause, meaning a coincidence, relieves him only if it was unforeseeable. A man leaves a stabbing victim on a beach at low tide and the tide drowns him: foreseeable, still homicide. The ambulance carrying the victim is struck by lightning: independent and unforeseeable, and the chain breaks.

Worth holding on to: Actual cause is a factual test, proximate cause is a fairness judgment, and foreseeability is what separates the intervening causes that matter from the ones that do not.

Common misconceptions

• A crime has one mental state. A crime has one mental state per material element, and the answers can differ across elements of the same offence.
• Silence in a statute means strict liability. Morissette, Staples, Rehaif, and Elonis all read a mental state into silent statutes; the presumption of fault is strong.
• Motive is an element the prosecution must prove. Motive is never an element; it is circumstantial evidence of intent and a sentencing consideration.
• Recklessness and negligence are two words for carelessness. Recklessness requires actual awareness of the risk; criminal negligence requires only that a reasonable person would have been aware, and demands a gross deviation.
• Suspecting something is enough for knowledge. Wilful blindness requires a belief in a high probability plus deliberate avoidance of confirmation, which is much more than suspicion.
• Any intervening event breaks causation. Foreseeable and dependent intervening causes, including ordinary medical error and a victim's refusal of treatment, do not.

Putting it together

• Morissette established that borrowing a common law crime without mentioning intent does not abolish the intent requirement.
• The Model Penal Code replaced a chaotic vocabulary with purposely, knowingly, recklessly, and negligently.
• The line between recklessness and negligence is awareness: subjective above, objective below.
• Element analysis assigns a mental state to each conduct, circumstance, and result element; recklessness is the default and higher states satisfy lower ones.
• Wilful blindness substitutes for knowledge only on a high probability belief plus deliberate avoidance.
• Strict liability survives for public welfare offences and statutory rape, and Staples shows how penalty severity and ordinariness of the item push against it.
• Transferred intent follows the harm; concurrence requires the intent to produce the act; motive is evidence, not an element.
• Causation runs but-for first, then proximate cause, where foreseeability decides which intervening events matter.

Sources

1. Legal Information Institute. (n.d.). Mens rea. Cornell Law School. law.cornell.edu
2. Wikipedia. (n.d.). Morissette v. United States. en.wikipedia.org
3. Wikipedia. (n.d.). Staples v. United States. en.wikipedia.org
4. Legal Information Institute. (n.d.). Strict liability. Cornell Law School. law.cornell.edu
5. Britannica. (2024). Criminal law: Mental element. britannica.com

Philadelphia, 1794: the year murder acquired degrees

Pennsylvania had a problem that was moral before it was legal. Under the common law inherited from England, murder was one offence and it carried one penalty: death. A poisoner who spent weeks planning and a tavern brawler who swung once too hard both hanged. Reformers in the new republic, many of them Quakers arguing for a narrower death penalty, wanted a way to keep capital punishment for the worst killings and remove it from the rest.

The solution the Pennsylvania legislature enacted in 1794 was to split the offence. Murder committed by poison or by lying in wait, or any other kind of wilful, deliberate and premeditated killing, or a killing committed in the perpetration of arson, rape, robbery, or burglary, became murder in the first degree, punishable by death. All other murder became murder in the second degree, punishable by imprisonment. Almost every American jurisdiction copied the structure, and two centuries later you are still reading its vocabulary in state codes. That single statute also planted the felony murder rule in American law in its modern form, which is the part still being fought over today.

This lesson takes homicide apart along those seams. Read it as a sorting problem: a person is dead, and the law must decide which box the killing goes in.

Malice aforethought, which means neither malice nor forethought

Common law murder is an unlawful killing of a human being with malice aforethought. The phrase is a fossil. It does not require ill will, and it does not require thinking ahead. It is a term of art covering four distinct mental states, any one of which is enough:

1. Intent to kill. The obvious one.
2. Intent to inflict grievous bodily harm. A man who beats another with a pipe intending to break his legs, and kills him, has murder even though he never wanted death.
3. Depraved heart, or extreme recklessness. Conduct showing extreme indifference to the value of human life: firing a rifle into an occupied house, dropping a concrete block from an overpass onto a highway.
4. Felony murder. A death caused during the commission of certain felonies, with the felony supplying the malice.

Everything else is manslaughter or no crime at all. Keep the four in mind as you read the table.

Grade	Mental state	Typical example
First degree murder	Wilful, deliberate, premeditated intent to kill; or an enumerated felony murder	Poisoning a spouse over weeks; a killing during an armed robbery
Second degree murder	Any other malice: intent to kill without premeditation, intent to do grievous harm, depraved heart	A sudden decision to shoot during an argument; firing into a crowded room
Voluntary manslaughter	Intent to kill, but formed under adequate provocation in the heat of passion	A killing moments after a violent attack by the victim
Involuntary manslaughter	Gross criminal negligence, or recklessness short of extreme indifference; sometimes an unlawful act	A death from grossly negligent handling of a loaded firearm
Negligent homicide (Model Penal Code)	Criminal negligence	A fatal collision caused by grossly careless driving

Remember: Malice aforethought is a label for four alternative mental states, not a requirement of hatred or planning.

What premeditation has to mean

The line between first and second degree turns on deliberation and premeditation, and courts have split over how much time that takes. Two cases mark the poles.

In Commonwealth v. Carroll (Pa. 1963), a man shot his wife in bed after a long marital argument, having thought about it, on his own testimony, for perhaps five minutes. The Pennsylvania court affirmed first degree murder and stated the traditional rule flatly: no time is too short for a wicked man to frame in his mind the scheme of murder. On that view, premeditation collapses into intent, since anyone who intends to kill must have formed the intent an instant before acting.

In State v. Guthrie (W. Va. 1995), the court refused that collapse. A dishwasher, needled by a coworker who snapped a towel at his nose, stabbed him to death. West Virginia held that first degree murder requires evidence that the defendant considered and weighed the decision, some appreciable time for reflection, not merely the instantaneous formation of intent. Otherwise, the court reasoned, the degrees the 1794 statute created would mean nothing.

Both rules are alive in American law. When you analyse a homicide problem, say which rule the jurisdiction uses before you answer, because the same facts can produce different degrees under each.

Provocation: the partial defence that cuts murder to manslaughter

A defendant who intended to kill can still avoid a murder conviction if he killed in the heat of passion on adequate provocation. This is not a full defence; it mitigates. Four elements are required: adequate provocation; the defendant was actually in the heat of passion; there was no reasonable cooling off period between provocation and killing; and the provocation caused the passion which caused the killing.

The historical approach was categorical. Only a closed list of events counted as adequate: a serious battery, mutual combat, an unlawful arrest, an assault on a close relative, and the discovery of one's spouse in the act of adultery. Words alone were never adequate, no matter how vicious, a rule most states still apply.

American courts moved gradually to a reasonable person standard, beginning with Maher v. People (Mich. 1862), which asked whether the provocation would cause an ordinary person to act from passion rather than judgment, and left it to the jury. That opened the category but also raised an obvious question: reasonable person means what, exactly, when we ask whether losing control was understandable?

The Model Penal Code went furthest, replacing provocation with extreme mental or emotional disturbance. Under that formulation a homicide that would otherwise be murder is manslaughter when committed under the influence of extreme mental or emotional disturbance for which there is a reasonable explanation or excuse, and the reasonableness is judged from the viewpoint of a person in the actor's situation under the circumstances as he believes them to be. That final clause imports a great deal of the defendant's own perspective, which is exactly the criticism made of it: applied loosely, it lets a jury excuse rage that the community should not accept. Several states adopted it; others rejected it for that reason.

The adultery category deserves the same scrutiny. It grew up in a legal culture that treated a wife as a form of property and a husband's violence on discovering her infidelity as understandable. A number of jurisdictions have narrowed or removed it, and a handful of states have legislated to bar or restrict provocation claims resting on the discovery of a victim's sexual orientation or gender identity, the so-called gay and trans panic defence. The doctrine is moving, and where a jurisdiction stands on this is worth checking rather than assuming.

Why this matters: Provocation asks not whether the defendant lost control but whether the law is prepared to say that losing control was understandable, which is a moral judgment wearing a doctrinal coat.

Involuntary manslaughter, and the negligence floor

Where there is no intent to kill and no extreme indifference, an unintended death may still be criminal. Two theories dominate. Criminally negligent homicide requires a gross deviation from reasonable care creating a substantial and unjustifiable risk of death, a standard well above ordinary tort negligence: leaving a loaded handgun where a five year old plays, or driving at eighty through a school zone. Unlawful act manslaughter, sometimes called misdemeanour manslaughter, holds a defendant for a death caused during a malum in se misdemeanour, though many modern codes have abolished or narrowed it because it can punish a death nobody could have foreseen.

The distinction from depraved heart murder is one of degree, and juries make it. Firing a gun into the ceiling of an occupied apartment building is closer to extreme indifference; leaving that gun unlocked in a drawer is closer to gross negligence.

Felony murder, and the long retreat from it

Under the felony murder rule, a death caused during the commission or attempted commission of a felony is murder, and the intent to commit the felony substitutes for malice. No intent to kill is needed, and in its strict form no negligence about the death is needed either.

What the rule can do is visible in People v. Stamp (Cal. Ct. App. 1969). Robbers held up a business; one victim, a sixty year old man with advanced heart disease, collapsed and died of a heart attack shortly after they fled. Nobody touched him. The court affirmed first degree murder: the robbery caused the fright, the fright caused the attack, and a felon takes his victim as he finds him.

Because the rule reaches that far, courts built limits around it.

• Inherently dangerous felony. Many states restrict the rule to enumerated felonies (arson, rape, robbery, burglary, kidnapping) or to felonies dangerous to human life.
• The merger rule. The underlying felony must have a felonious purpose independent of the killing. Otherwise every aggravated assault that caused death would automatically be murder, and manslaughter would vanish.
• Res gestae. The death must occur within the time, place, and causal ambit of the felony, including immediate flight.
• Agency versus proximate cause. When a police officer or victim kills someone, the majority agency rule holds the felons liable only for killings by a co-felon; a minority proximate cause rule holds them for any foreseeable death the felony set off, including the death of a co-felon shot by the victim.

The rule's standing has weakened. England abolished constructive malice in the Homicide Act 1957, and felony murder does not exist in most other common law countries. Michigan abolished it judicially in People v. Aaron (Mich. 1980), holding that malice must be proved. Hawaii and Kentucky have abolished it by statute. California narrowed it sharply by legislation in 2018, restricting murder liability to those who actually killed, who aided the killing with intent to kill, or who were major participants acting with reckless indifference to human life. Defenders answer that the rule deters the use of violence during felonies and expresses a real intuition that people who bring guns to robberies own what follows. Critics reply that the deterrence evidence is thin and that the rule most often falls on the least culpable participant, the lookout or the driver.

The point: Felony murder converts a felony into malice, and every limiting doctrine you learn is an attempt to keep that conversion from swallowing the law of manslaughter.

Who counts as a victim, and when

Two boundary questions still arise. At the beginning of life, the common law born alive rule required that the victim be born and breathing; most states have supplemented it with feticide statutes protecting an unborn child at some stage, and the details vary sharply by jurisdiction. At the end of life, the traditional test was cessation of heartbeat and breathing, which modern intensive care made unworkable; the Uniform Determination of Death Act of 1981 added irreversible cessation of all functions of the entire brain, and that standard is now general.

The old year and a day rule, which barred a homicide charge if the victim survived a year and a day after the injury, made sense when medicine could not trace causation over long periods. It has been abolished nearly everywhere, by statute or by decision, as you saw in the first lesson.

Common misconceptions

• Malice aforethought means the killer hated the victim and planned it. It is a technical label covering four mental states, including extreme recklessness and felony murder.
• Premeditation requires hours of planning. Some states require only an instant, others require appreciable reflection; you must name the jurisdiction's rule before answering.
• Provocation is a complete defence. It is a partial defence that reduces murder to voluntary manslaughter; the defendant is still convicted of a homicide.
• Insults can be adequate provocation. Words alone are traditionally never adequate, however severe.
• Felony murder requires that the felon cause the death directly. Under Stamp a fright-induced heart attack sufficed; the limits come from the dangerousness, merger, res gestae, and agency doctrines instead.
• Felony murder exists in every common law country. England abolished constructive malice in 1957, and several American states have abolished or sharply narrowed the rule.

What to carry forward

• Pennsylvania's 1794 statute created murder degrees to narrow the death penalty, and its structure still organises American homicide law.
• Malice aforethought covers intent to kill, intent to do grievous harm, depraved heart recklessness, and felony murder.
• First degree turns on premeditation, where Carroll and Guthrie represent opposite rules about how much reflection is required.
• Provocation reduces murder to voluntary manslaughter on four elements, with categorical, reasonable person, and Model Penal Code approaches in play.
• Involuntary manslaughter rests on gross negligence or on an unlawful act, and shades into depraved heart murder by degree.
• Felony murder substitutes felonious intent for malice; dangerousness, merger, res gestae, and agency limit it, and several jurisdictions have abolished or narrowed it.
• Born alive rules, feticide statutes, brain death standards, and the abolished year and a day rule define who counts as a homicide victim.

Sources

1. Legal Information Institute. (n.d.). Homicide. Cornell Law School. law.cornell.edu
2. Legal Information Institute. (n.d.). Felony murder doctrine. Cornell Law School. law.cornell.edu
3. Britannica. (2024). Homicide. britannica.com
4. Wikipedia. (n.d.). Felony murder rule. en.wikipedia.org
5. Bureau of Justice Statistics. (n.d.). Homicide data and reports. Office of Justice Programs. bjs.ojp.gov

Buffalo, 1984: a defendant argues the exemption is unfair to him

Mario Liberta was prosecuted in New York for raping his wife. Under the statute then in force he should not have been prosecutable at all, because New York law exempted husbands. The exemption did not reach him only on a technicality: he and Denise Liberta were living apart under a court order of protection, and the statute treated a husband living apart under such an order as not married for its purposes.

His lawyers made the argument the statute invited. If the legislature really believed a husband could not rape his wife, then punishing this husband, when a husband still living at home could not be punished for identical conduct, was irrational and denied him equal protection. It was, on its own terms, a good argument.

The New York Court of Appeals agreed that the classification could not survive, and then did the thing the defence had not wanted. In People v. Liberta (N.Y. 1984) it held the marital exemption unconstitutional and severed it from the statute rather than striking the statute down, so that the law thereafter applied to every husband in New York. It removed the statute's gender limitation for the same reason. Liberta's conviction stood, and marital rape became a crime in the state.

That case is a small window onto a two-hundred-year renovation. This lesson covers offences against the person short of homicide: first the comparatively simple law of assault and battery, then the far harder law of sexual offences, where nearly every rule you might expect to be ancient was rewritten within living memory.

Assault and battery, and the vocabulary problem

At common law these were two offences, and the modern confusion comes from statutes that merged them under one label.

Battery is the unlawful application of force to another person. It does not require injury: an offensive touching is enough, and spitting on someone is a battery. The mental state at common law was general intent, and modern codes typically require purpose, knowledge, or recklessness as to the contact.

Assault came in two forms. The first is an attempted battery, which requires the intent to commit a battery plus a substantial step. The second, added later, is the intentional creation of a reasonable apprehension of imminent harmful or offensive contact, which is why pointing an unloaded gun at someone who believes it is loaded is an assault under this branch even though a battery was impossible.

Most modern codes now call the whole family assault and grade it. Simple assault is typically a misdemeanour. Aggravated assault is a felony and is triggered by one of a familiar set of factors: use of a deadly weapon, infliction of serious bodily injury, intent to commit another felony, or the victim's protected status, such as a police officer, a child, or an elderly person. When you read a statute in this area, do not assume the word assault means what it meant at common law. Read the definition.

Two modern additions are worth naming because they changed practice rather than theory. Every state has enacted stalking statutes, beginning with California in 1990 and spreading nationwide within six years, criminalising a course of conduct that causes a reasonable person to fear for their safety. And most states have made non-fatal strangulation a distinct felony rather than a simple assault, because research on intimate partner violence found that a prior non-fatal strangulation is among the strongest predictors that a later attack will be fatal. A charge that used to be a misdemeanour with a fine is now a felony that can support pretrial detention.

In short: Battery is the touching, assault is the attempt or the threat, and modern statutes merge them and grade by weapon, injury, and victim.

What the common law of rape required, and the four hurdles it built

Common law rape was defined as carnal knowledge of a woman, not the perpetrator's wife, by force or threat of force, against her will. Four features of the surrounding doctrine made prosecution far harder than the elements suggest.

1. Force and resistance. Courts required proof of force, and many required the victim to have resisted, sometimes to the utmost. The practical result was that the complainant's conduct, not the defendant's, became the trial's subject.
2. The marital exemption. Sir Matthew Hale, writing in the seventeenth century in a treatise published in 1736, stated that a husband cannot be guilty of a rape upon his lawful wife, because by their matrimonial consent she has given herself up in this kind and cannot retract it. No case established the rule; a treatise asserted it and courts repeated it for three centuries.
3. Corroboration. Several jurisdictions refused to convict on the complainant's testimony alone, a requirement imposed for no other violent crime.
4. The cautionary instruction and prompt complaint. Juries were routinely told, again on Hale's authority, that the charge is easily made and hard to defend, and delay in reporting was treated as evidence of fabrication.

Together these did not merely make convictions difficult. They encoded a suspicion of the complainant into the structure of the trial.

The reform, decade by decade

The renovation began in earnest in the 1970s and is not finished.

Michigan, 1974. The first comprehensive statute, replacing rape with graded degrees of criminal sexual conduct. It was gender neutral, covered penetration and contact, defined the offence by the defendant's conduct and circumstances rather than by the victim's resistance, eliminated the corroboration requirement, and included a rape shield provision. Most states followed some version of this template within fifteen years.

Rape shield rules. Federal Rule of Evidence 412, enacted in 1978, and its state analogues generally bar evidence of a complainant's other sexual behaviour or sexual predisposition, with narrow exceptions: evidence that someone else was the source of semen or injury, evidence of prior sexual behaviour with the accused when offered on consent, and evidence whose exclusion would violate the defendant's constitutional rights. That last exception is not small, and it is where the litigation happens.

The marital exemption. By July 1993 marital rape was a crime in all fifty states. Full parity took longer, and in some states still has not arrived: several jurisdictions retained lower penalties, shorter reporting deadlines, or narrower definitions for spousal cases well into this century, and legislatures have been repealing those remnants one at a time.

Resistance. Most states have removed any express resistance requirement, and some define force so that the act of penetration without permission itself supplies it. New Jersey's Supreme Court took that step in In re M.T.S. (N.J. 1992), holding that the statutory element of physical force is satisfied by any act of sexual penetration accomplished without the affirmative and freely given permission of the other person.

What matters here: Nearly every doctrine that made these prosecutions distinctive, resistance, corroboration, the marital exemption, and the cautionary instruction, was removed by legislatures and courts within a single professional lifetime.

Three models of consent

Jurisdictions now differ, and the difference is not rhetorical. It changes what the prosecution must prove.

Model	What the state must show	Where the burden falls in practice
Force plus non-consent	An act of force or threat, and that it was against the victim's will	On proving force, which invites scrutiny of resistance
Non-consent alone	That the act occurred without consent, force being one way to show it	On the meaning of consent and the defendant's awareness of its absence
Affirmative consent	That there was no freely given agreement communicated by words or conduct	On evidence of communicated willingness rather than of refusal

Wisconsin, for example, defines consent as words or overt actions indicating a freely given agreement. California adopted an affirmative consent standard in 2014, but note carefully what it governs: that statute sets the standard for campus disciplinary proceedings at institutions receiving state funds, not the standard for a criminal prosecution. Students frequently conflate the two, and reporters sometimes do as well. Campus adjudication, criminal prosecution, and civil litigation apply different standards to the same events, and a finding in one does not decide the others.

The American Law Institute spent more than a decade revising the Model Penal Code's sexual assault article, finishing in 2021. The process was unusually contentious, with drafts debated and voted down at successive annual meetings. The membership declined to adopt the strongest affirmative consent proposals, settling on a definition of consent as behaviour, including words and conduct, that communicates willingness, and retaining the state's burden to prove its absence. That the country's most influential law reform body needed eleven years and could not reach consensus on the core definition tells you something honest about how unsettled this area is.

When consent cannot be given, and when it is void

Consent is ineffective where the person lacks capacity to give it. The recognised categories are unconsciousness or sleep; incapacitation by drugs or alcohol, which every state addresses though the threshold language varies from unable to consent to substantially impaired; mental incapacity; and being under the age of consent, which is the subject of statutory rape law. Note that the intoxication threshold is not simply drunk. Statutes generally require an inability to understand or to communicate, and where the defendant administered the intoxicant, the offence is usually graded higher.

Fraud divides into two categories that produce different answers. Fraud in the factum concerns the nature of the act itself, as where a patient is told a sexual act is a medical procedure; consent is void, because the person did not agree to the act that occurred. Fraud in the inducement concerns a collateral lie, such as a false promise of marriage or payment; traditionally consent remained valid, and some jurisdictions have since criminalised specific inducements, notably impersonating a spouse or partner. This is one of the areas where states diverge most sharply.

Key idea: Consent fails either because the person could not give it or because what they agreed to was not what occurred.

Statutory rape, and the argument about strict liability

Sexual conduct with a person below the age of consent is an offence regardless of the younger person's apparent willingness. Ages of consent in the United States range from sixteen to eighteen depending on the state. Most jurisdictions treat the age element as strict liability, so a reasonable and honest mistake about age is no defence, and a minority permit a reasonable mistake defence where the age gap is small. Many states also have close-in-age or Romeo and Juliet provisions reducing or eliminating liability where the participants are within a few years of each other.

The strict liability rule is defended on the ground that adults dealing with young people bear the risk of being wrong, and criticised on the ground that it convicts people who lacked any culpable mental state as to the only element that made the conduct criminal. Both arguments are serious. The legislatures have almost uniformly chosen the first.

What reform did not fix

Doctrinal change did not close the gap between offences and convictions. National Crime Victimization Survey data have consistently shown that fewer than half of rapes and sexual assaults are reported to police, and reported cases fall away at every subsequent stage. Investigations of untested sexual assault evidence kits in cities across the country found tens of thousands of kits sitting in police storage, some for decades, and federal grant programmes have since funded testing that produced both convictions and exonerations. Acquaintance cases, which are the large majority, remain the hardest to prove, because the disputed fact is rarely identity and almost always consent, and there is often no evidence beyond two accounts.

None of that is an argument against reform. It is a reminder that changing an element in a statute and changing an outcome in a courtroom are different achievements.

Common misconceptions

• Assault means hitting someone. At common law the hitting is battery; assault is the attempt or the creation of reasonable apprehension, and modern statutes redefine the words.
• The marital exemption was an ancient statute. It rested on an assertion in Hale's treatise, not on any statute or decided case, and it was repeated for three centuries before being dismantled.
• Victims must resist for a prosecution to succeed. Most states have abolished the resistance requirement, and some, following M.T.S., find the force element satisfied by penetration without permission.
• Rape shield laws exclude all evidence about the complainant. They bar general sexual history but contain exceptions for alternative sources of physical evidence, prior conduct with the accused, and constitutionally required evidence.
• Affirmative consent is the criminal standard in California. California's 2014 statute sets the standard for campus disciplinary proceedings, not for criminal prosecution.
• Being intoxicated automatically voids consent. Statutes generally require incapacity, meaning an inability to understand or communicate, rather than any level of intoxication.

Looking back

• Battery is the unlawful touching; assault is the attempt or the reasonable apprehension of one; aggravating factors turn either into a felony.
• Stalking and strangulation statutes are recent additions that changed charging practice substantially.
• Common law rape law was built around force and resistance, the marital exemption, corroboration, and a cautionary instruction, all traceable to Hale.
• Michigan's 1974 statute set the reform template: graded, gender neutral, defined by the defendant's conduct, without corroboration.
• Rape shield rules such as Federal Rule of Evidence 412 limit sexual history evidence but contain real exceptions.
• Liberta ended New York's marital exemption in 1984; by 1993 marital rape was criminal in every state, though full parity took longer.
• Three consent models are in use, and the American Law Institute's 2021 revision shows the definition is still contested.
• Incapacity and fraud in the factum void consent; statutory rape is generally strict liability as to age.

Sources

1. Legal Information Institute. (n.d.). Assault. Cornell Law School. law.cornell.edu
2. Legal Information Institute. (n.d.). Rape shield laws. Cornell Law School. law.cornell.edu
3. Wikipedia. (n.d.). Marital rape in the United States. en.wikipedia.org
4. National Institute of Justice. (n.d.). Sexual violence research and sexual assault kit testing. Office of Justice Programs. nij.ojp.gov
5. Bureau of Justice Statistics. (n.d.). National Crime Victimization Survey. Office of Justice Programs. bjs.ojp.gov

London, 1799: a bank teller and a hole in the law

Joseph Bazeley worked as a teller at a London bank. A customer handed him a hundred pound note to deposit. Bazeley put it in his own pocket. Everyone agreed what he had done was dishonest, and the prosecution charged him with larceny, which was then a capital offence.

The judges acquitted him of larceny. Larceny required a trespassory taking, meaning a taking from the possession of another. The note had gone directly from the customer's hand to Bazeley's hand. The bank never possessed it. There was no possession to trespass against, so the defining element of the crime was missing, and the most obvious thief in London walked out of court.

Parliament passed an embezzlement statute within months. That sequence, a gap discovered in litigation and a statute enacted to close it, is the whole history of theft law in one episode. What you now read in a state penal code as a single crime called theft is a collection of patches sewn over holes that defendants found first.

Larceny, and why it protects possession rather than ownership

Common law larceny is the trespassory taking and carrying away of the personal property of another with the intent to permanently deprive the owner of it. Six components hide in that sentence:

• Trespassory: without the consent of the person in possession.
• Taking (caption): acquiring dominion and control.
• Carrying away (asportation): any movement, however slight. Lifting a television an inch off a shelf satisfies it.
• Personal property: at common law, tangible movable goods, not land, not services, and originally not intangibles, which is why statutes had to be written for electricity, gas, cable, and data.
• Of another: another's possession. You can commit larceny of your own car by taking it from the mechanic who holds a valid lien on it.
• Intent to permanently deprive: borrowing without permission is not larceny, which is why joyriding needed its own statute.

The fifth point is the one students misread. Larceny is not about who owns the thing. It is about who was holding it lawfully.

The point: Larceny protects possession, and every classic theft puzzle is really a question about who possessed the property at the moment intent formed.

The two patches, and the test that separates them

Two statutory offences fill the gaps Bazeley and others exposed.

Embezzlement is the fraudulent conversion of property by a person who is already in lawful possession of it. The bank teller, the trustee, the payroll clerk, the lawyer holding client funds. The defendant got the property legitimately; the crime is what he did afterwards.

False pretenses is obtaining title to property by a knowingly false representation of a past or existing fact, with intent to defraud. The buyer who pays with a check he knows is worthless has induced the seller to hand over ownership, not merely custody. Note two limits: a false promise about the future is traditionally not a false representation of fact, and mere opinion or sales puffery is not either.

There is a fourth member of the family. Larceny by trick covers deception that obtains only possession rather than title: renting a car with no intention of returning it. Because possession alone passed, the original taking is still treated as trespassory, and it remains larceny.

Offence	What the defendant had	When intent formed	Example
Larceny	Nothing, or bare custody	At the taking	Shoplifting; an employee taking from the till
Larceny by trick	Possession, obtained by deception	At or before the taking	Renting equipment intending to keep it
Embezzlement	Lawful possession	After lawful possession began	A trustee spending trust funds
False pretenses	Title	At the transfer	Buying a car with a knowingly bad check

The distinction between custody and possession does real work. An employee handling the shop's cash register generally has custody only, with the employer retaining constructive possession, so taking from the till is larceny. A delivery driver entrusted with a truckload of goods for an out-of-state run generally has possession, so keeping the goods is embezzlement.

Consolidation, and the variance problem it solved

Dividing theft into three offences created an ugly practical problem. A prosecutor who charged larceny and proved embezzlement lost the case, not because the defendant was innocent but because the pleading did not match the proof. Defendants exploited it, arguing at each trial that they had actually committed the other crime.

Modern codes fixed this by consolidation. The Model Penal Code's theft article gathers larceny, embezzlement, false pretenses, extortion, receiving stolen property, and theft of services into one offence of theft, and provides expressly that an accusation of theft may be supported by evidence that it was committed in any manner that would be theft under the article, so long as the defendant is not unfairly surprised. Most states have adopted some version. Learn the historical categories anyway: they still explain what the elements mean and they still appear in appellate opinions.

Robbery, extortion, and burglary

Robbery is larceny from the person or presence of the victim, accomplished by force or by putting the victim in fear of immediate harm. Two details matter. The force must be more than the effort of taking, so a clean pickpocketing is larceny, while a purse snatch involving a struggle is robbery. And the force or fear must occur at or before the taking; jurisdictions differ on whether force used during immediate escape converts a completed larceny into robbery.

Extortion, historically blackmail, covers obtaining property by threats of future harm, including threats to accuse of a crime, to expose a secret, or to take official action. The clean line is timing: immediate harm to the person present is robbery, threatened later harm is extortion.

Burglary at common law had five elements, and modern statutes have discarded most of them. The common law required breaking and entering the dwelling house of another in the nighttime with intent to commit a felony inside. Modern codes typically require entering or remaining unlawfully in a building or occupied structure with intent to commit a crime, dropping the breaking requirement, the night requirement, and the limitation to dwellings, though many grade dwellings and night entries higher. Recall from the lesson on mens rea that the intent must exist at entry. A man who enters a store lawfully, then decides to steal, commits theft but not common law burglary, and modern remaining unlawfully language was written to reach the person who hides in the store until closing.

Receiving stolen property requires receiving property that is in fact stolen, knowing or believing it to be stolen, with intent to deprive the owner. Two recurring issues: the property must actually be stolen at the time, which is why sting operations using recovered goods produce attempt charges rather than completed offences in many states, and knowledge is often proved by circumstantial evidence such as price, place, and the seller's evasiveness. Wilful blindness, from the mens rea lesson, does a lot of work here.

Remember: Robbery is force now, extortion is harm later, and burglary is about an unlawful entry with a criminal purpose already formed.

Federal fraud, and how far a statute can stretch

Two federal statutes carry an enormous share of white collar prosecution. The mail fraud statute, 18 U.S.C. section 1341, and the wire fraud statute, section 1343, each require a scheme to defraud, plus the use of the mails or of interstate wires in furtherance of it. The mailing or the wire need not be fraudulent itself; a routine confirmation email will do.

That structure makes the statutes extraordinarily flexible, which is why federal prosecutors reach for them so often and why appellate courts have spent forty years pruning them. Three recent decisions show the trimming.

• Skilling v. United States (2010), arising from the collapse of Enron, addressed honest services fraud under section 1346. The Court held that reading the statute to cover any breach of a duty of honest services would raise vagueness problems, and confined it to bribery and kickback schemes.
• Kelly v. United States (2020), the New Jersey bridge lane closure case, reversed convictions because the object of the scheme was to exercise regulatory power for a political motive, not to obtain money or property. Wire fraud is a property crime, and corrupt purpose alone does not supply the missing element.
• Ciminelli v. United States (2023) rejected the right to control theory, under which depriving a victim of information relevant to an economic decision counted as a property deprivation. The Court held that intangible interest is not property under the statute.

A related area has moved the same way. Aggravated identity theft, 18 U.S.C. section 1028A, adds a mandatory two-year consecutive sentence when a defendant uses another person's means of identification during certain offences. Prosecutors read it broadly enough to attach to routine billing fraud. In Dubin v. United States (2023) the Court narrowed it, holding that the use of the identification must be at the crux of the criminality rather than merely ancillary to it.

Read those four cases together and you see an institutional pattern rather than four unrelated results: a statute broad enough to be useful invites expansion, and expansion invites a court to draw the line the legislature did not.

Why a dollar figure decides the grade

Nearly every theft statute grades by value, and the felony threshold is a policy choice with visible consequences. Thresholds vary widely among the states, and many were set decades ago and left to inflation, so a fixed dollar amount quietly criminalised more conduct each year. Between roughly 2001 and 2018 a large number of states raised their felony theft thresholds. Analyses comparing states that raised thresholds with states that did not found no resulting increase in property crime or larceny rates, which weakened the main argument against raising them.

The lesson generalises past theft. When a criminal statute keys liability to a number, that number is doing moral work, and it should be revisited as prices move.

The upshot: Grading rules are where legislatures make quiet decisions about who becomes a felon, and empirical evidence has begun to discipline them.

Common misconceptions

• You cannot steal your own property. Larceny protects possession, so taking your car from a mechanic with a valid lien can be larceny.
• Embezzlement means stealing large sums from a company. It means converting property you already lawfully possess, whatever the amount and whatever the setting.
• A broken promise about the future is false pretenses. The traditional rule requires a false representation of a past or existing fact; promises and opinions are excluded.
• Every taking with force is robbery. The force must exceed the effort of the taking; a clean pickpocketing remains larceny.
• Burglary requires breaking in at night to a house. Modern statutes have generally dropped breaking, night, and the dwelling limitation, though many grade those facts more severely.
• Mail and wire fraud reach any dishonest scheme. The object must be money or property, as Kelly held, and honest services fraud is limited to bribes and kickbacks after Skilling.

Pulling it together

• Bazeley's acquittal in 1799 shows why theft law is a set of statutory patches over the narrow common law of larceny.
• Larceny requires a trespassory taking and asportation of another's personal property with intent to permanently deprive, and it protects possession.
• Ask what the defendant held when intent formed: custody points to larceny, possession to embezzlement, title to false pretenses.
• Consolidated theft statutes ended the variance trap while preserving the old categories as interpretive tools.
• Robbery is force or fear of immediate harm at the taking; extortion threatens future harm; burglary needs criminal intent at entry.
• Mail and wire fraud require a scheme aimed at money or property, and Skilling, Kelly, Ciminelli, and Dubin have narrowed aggressive readings.
• Felony thresholds are policy choices, and raising them has not been shown to increase property crime.

Sources

1. Legal Information Institute. (n.d.). Larceny. Cornell Law School. law.cornell.edu
2. Legal Information Institute. (n.d.). Embezzlement. Cornell Law School. law.cornell.edu
3. Wikipedia. (n.d.). Skilling v. United States. en.wikipedia.org
4. Britannica. (2024). Theft. britannica.com
5. Bureau of Justice Statistics. (n.d.). Property crime data. Office of Justice Programs. bjs.ojp.gov

A man in a prison cell, convicted of crimes committed outside it

Walter and Daniel Pinkerton were brothers in Alabama who ran an illicit whiskey operation and were prosecuted for federal tax offences. Walter committed a series of substantive violations. During part of the relevant period, Daniel was in prison and could not have participated in any of them.

He was convicted of them anyway. In Pinkerton v. United States, 328 U.S. 640 (1946), the Supreme Court held that a member of a conspiracy is criminally liable for the substantive offences committed by co-conspirators in furtherance of the conspiracy, where those offences were reasonably foreseeable, so long as he had not withdrawn.

Start with the discomfort. It is one thing to punish a person for what he did, another to punish him for an agreement, and a third to punish him for what a partner did while he sat in a cell. This lesson is about that third territory: the offences that punish incomplete conduct and the doctrines that extend liability to people who never touched the crime.

Attempt: the intent, and the act that is enough

Every attempt requires two things: the intent to commit the target offence, and an act beyond mere preparation.

The intent element carries a trap. Attempt is a specific intent crime even when the completed offence is not. Murder can be committed with intent to inflict grievous bodily harm or with depraved heart recklessness, but attempted murder requires the intent to kill. It follows that there is no such crime as attempted felony murder or attempted involuntary manslaughter: you cannot intend to bring about an unintended death.

The act element has never had a single test. Three are worth knowing:

• Last act. The oldest and narrowest: the defendant must have done everything he believed necessary. Nearly abandoned, because it would require waiting until the trigger is pulled.
• Dangerous proximity. Associated with Justice Holmes: how close in time and space was the defendant to completion, and how grave the harm? In People v. Rizzo (N.Y. 1927), four armed men drove around New York looking for a payroll clerk they meant to rob. They never found him. The court reversed the attempt conviction: the intended victim was not present, so they were never dangerously near success.
• Substantial step. The Model Penal Code's test, now the majority. The act must be a substantial step strongly corroborative of the actor's criminal purpose. The Code lists examples that may qualify: lying in wait, following the intended victim, reconnoitring the place contemplated, unlawful entry, and possession of materials specially designed for the crime.

Notice the direction of travel. Dangerous proximity looks forward from the defendant's conduct toward the completed crime and asks how far is left. The substantial step test looks backward from the conduct and asks whether it confirms the purpose. Under the second test, the men in Rizzo are almost certainly guilty. Which test the jurisdiction uses can decide the case.

Key idea: Attempt requires purpose to complete the offence plus conduct that either brings the defendant dangerously near success or strongly corroborates his intent, depending on the jurisdiction.

Impossibility and abandonment

Factual impossibility is not a defence. A pickpocket who reaches into an empty pocket, a shooter who fires at a bed containing pillows rather than a person, a buyer who purchases what he thinks is cocaine and is in fact powdered sugar: each intended a crime and did everything he meant to do, and the failure was a matter of luck. The Model Penal Code puts the point crisply by asking whether the conduct would be a crime had the circumstances been as the defendant believed them to be.

Pure legal impossibility is a defence, and it is narrower than students expect. It applies only where the defendant intended to do something that is not a crime at all, believing wrongly that it was. A man who imports lace believing that importing lace is forbidden, when it is not, has committed no offence, because there is no crime for him to have attempted. A large middle category once called hybrid legal impossibility, covering mistakes about legal status such as receiving goods believed to be stolen that were not, has been abolished in most jurisdictions and folded into the factual side.

Abandonment divides jurisdictions. At common law the majority rule is that once the attempt is complete, abandonment is no defence, because the crime is already committed. The Model Penal Code allows a defence of renunciation where the defendant abandons the effort or prevents the crime, and the renunciation is complete and voluntary. That qualifier does the work: giving up because a police car appeared, or postponing until a better opportunity, or switching to a different victim, is not voluntary renunciation. Changing your mind because you decided it was wrong is.

Solicitation, and why refusal does not help the solicitor

Solicitation is asking, commanding, encouraging, or hiring another person to commit a crime, with the intent that the crime be committed. The offence is complete the moment the request is communicated. If the person solicited refuses, laughs, or reports the request to police, the solicitation has still occurred. If the person agrees, you now also have a conspiracy. If the crime is then attempted or completed, the solicitation merges into it and the solicitor is liable as an accomplice.

Conspiracy: an agreement, and everything that follows from it

Conspiracy is an agreement between two or more persons to commit a crime, with the intent to agree and the intent to achieve the criminal objective. Most jurisdictions add an overt act in furtherance, which can be trivial and entirely lawful, such as buying a map or making a phone call. The general federal conspiracy statute, 18 U.S.C. section 371, requires an overt act; the federal drug conspiracy statute, 21 U.S.C. section 846, does not, so the agreement alone completes that offence.

Three structural points:

• Bilateral or unilateral. The common law required at least two genuinely agreeing minds, so a defendant who agreed only with an undercover officer had no conspiracy. The Model Penal Code adopts a unilateral approach, asking whether this defendant agreed with another person, which permits conviction in the sting scenario. States are split.
• Scope. Is one large conspiracy proved, or several small ones? The distinction decides whose statements come in and who is tried with whom. In Kotteakos v. United States (1946), a single loan broker arranged fraudulent loans for many unconnected borrowers. The government charged one conspiracy; the Court found a wheel with no rim, because the borrowers had no interest in one another's transactions, and reversed convictions tainted by the mass trial.
• No merger. Unlike attempt and solicitation, conspiracy does not merge into the completed offence. A defendant may be convicted and sentenced for both the conspiracy and the crime it produced.

Why do prosecutors love conspiracy? Because of what comes with it, not because of the sentence. Venue lies wherever any overt act occurred. Co-conspirators may be tried together, so a jury hears the whole enterprise at once. Statements by one conspirator during and in furtherance of the conspiracy are admissible against every member as non-hearsay under Federal Rule of Evidence 801(d)(2)(E). The limitation period runs from the last overt act rather than from the agreement. And in federal court and many states, the Pinkerton rule attaches substantive liability for what partners did.

The traditional check is Wharton's rule, which says that where an offence by definition requires two participants, such as duelling or bribery, the agreement to commit it is not separately punishable as a conspiracy. It is a presumption about legislative intent and yields to a contrary statute.

What matters here: Conspiracy's real power is procedural, in venue, joinder, evidence, and timing, not in the sentence attached to the agreement.

Accomplice liability, and the knowledge Rosemond required

An accomplice is liable for the crime the principal commits. The federal provision, 18 U.S.C. section 2, states that whoever aids, abets, counsels, commands, induces, or procures the commission of an offence is punishable as a principal. Two elements are required:

1. Conduct: assistance or encouragement. It can be trivial and need not be necessary to the outcome. Acting as a lookout, driving, supplying a tool, or shouting encouragement all qualify. What does not qualify is mere presence: watching a crime, even approvingly, is not aiding it. Presence plus a prior agreement to be there in case help is needed does qualify.
2. Mental state: intent to assist the conduct, and intent that the underlying offence be committed. A merchant who sells an ordinary kitchen knife to a customer who then commits a stabbing is not an accomplice; a merchant who sells at an inflated price knowing and desiring the use may be.

Rosemond v. United States (2014) sharpened the mental element. Rosemond took part in a drug transaction during which someone fired a gun, and was convicted of aiding and abetting the use of a firearm during a drug crime. The Court held that liability requires advance knowledge that a confederate would carry a gun, meaning knowledge at a point when the defendant still had a realistic opportunity to withdraw. Learning of the gun only when it appears does not make him an aider and abettor of the firearm offence, because he cannot be said to have chosen to assist an armed crime.

Two further doctrines are worth naming. The natural and probable consequences rule extends accomplice liability to further crimes that were a foreseeable outgrowth of the one the accomplice intended to assist. It has been criticised for imposing murder liability on the basis of negligence, and it is in retreat: California abolished it for murder in its 2018 legislation. And an accessory after the fact is a separate, lesser offence: someone who, knowing a felony was committed, helps the offender avoid arrest or punishment. That person is not liable for the underlying felony.

Two counterintuitive consequences follow from the derivative structure. An accomplice can sometimes be convicted of a greater offence than the principal, as where the accomplice acts with premeditation while the principal kills in the heat of passion. And withdrawal is possible but demanding: the accomplice must communicate the withdrawal in time and neutralise the assistance already given, which for a supplied weapon usually means retrieving it or warning police.

So what?: Accomplice liability turns on a choice to help a particular crime, which is why advance knowledge and effective withdrawal both matter so much.

Common misconceptions

• An attempt requires getting close to completing the crime. Under the majority substantial step test, conduct strongly corroborating the purpose suffices even at a distance from success.
• It is a defence that the crime could not have worked. Factual impossibility is never a defence; only the narrow category of pure legal impossibility is.
• Backing out always cancels an attempt. The common law majority allows no abandonment defence once the attempt is complete, and the Model Penal Code requires renunciation that is both complete and voluntary.
• A solicitation fails if the other person refuses. The offence is complete when the request is communicated with the required intent.
• Conspiracy merges into the completed crime. It does not; a defendant may be punished for both, unlike attempt and solicitation.
• Being present and approving makes you an accomplice. Mere presence is insufficient; some assistance or encouragement, with intent, is required.

What you now know

• Pinkerton makes a conspirator liable for reasonably foreseeable substantive crimes committed by partners in furtherance of the agreement.
• Attempt requires intent to complete the offence plus an act beyond preparation, tested by dangerous proximity or, in most states, a substantial step.
• Factual impossibility is no defence; pure legal impossibility is, and the hybrid category has largely disappeared.
• Renunciation must be complete and voluntary, which excludes abandonment prompted by risk of detection or a better opportunity.
• Solicitation is complete on communication; conspiracy requires an agreement, intent, and usually an overt act, and does not merge.
• Conspiracy's advantages are procedural: venue, joinder, co-conspirator statements, and a limitation period running from the last overt act.
• Accomplice liability needs assistance plus intent that the offence be committed, and Rosemond requires advance knowledge of an armed confederate.
• Accessory after the fact is a distinct lesser offence, and withdrawal requires timely communication and neutralising prior help.

Sources

1. Legal Information Institute. (n.d.). Conspiracy. Cornell Law School. law.cornell.edu
2. Legal Information Institute. (n.d.). Accomplice. Cornell Law School. law.cornell.edu
3. Wikipedia. (n.d.). Pinkerton v. United States. en.wikipedia.org
4. Wikipedia. (n.d.). Rosemond v. United States. en.wikipedia.org
5. Britannica. (2024). Conspiracy. britannica.com

A subway car in December 1984, and an argument about one word

On 22 December 1984, four young men approached Bernhard Goetz on a Manhattan subway train and asked him for five dollars. Goetz, who had been mugged before and carried an unlicensed revolver, drew it and fired five shots. All four were wounded; one, Darrell Cabey, was left paralysed.

The legal fight that followed was not about what happened but about a single word in New York's justification statute. The statute allowed deadly force where the defendant reasonably believes it necessary. The prosecutor had told the grand jury to consider what a reasonable man in the defendant's situation would believe. Lower courts read that phrase as smuggling in a subjective test and dismissed the charges.

In People v. Goetz (N.Y. 1986) the Court of Appeals reinstated them. Reasonably believes, it held, means a belief that is both honestly held and objectively reasonable. A jury may take into account the defendant's own circumstances, including relevant prior experiences and what he knew about his assailants, but the ultimate question is what a reasonable person in those circumstances would have believed, not what this defendant sincerely felt. A trial jury later acquitted Goetz of the shooting charges and convicted him only of unlawful gun possession.

Hold that distinction, because it runs through every defence in this lesson. Whose perspective governs, and how far does the law bend the standard toward the individual defendant?

Justification is not excuse

The prosecution has proved actus reus, mens rea, and causation. The defendant still walks free if a defence applies, and defences come in two families that are often blurred and should not be.

A justification says the conduct was permissible, even right, in the circumstances. Society does not condemn it. Self-defence, defence of others, defence of property, necessity, and lawful law enforcement force are justifications.

An excuse says the conduct was wrongful but this actor is not blameworthy. Insanity, duress, infancy, and involuntary intoxication are excuses.

The distinction is not academic. Three consequences follow. You may lawfully assist someone acting under a justification, but assisting an excused actor may itself be criminal. You may not lawfully resist justified force, but you may resist an excused attacker, since his conduct is still unlawful. And a justification typically produces a full acquittal, while some excuses produce a special verdict with civil commitment attached.

The core of it: A justification denies that the act was wrong; an excuse concedes it was wrong and denies that the actor is to blame.

Self-defence: four requirements

A person may use force against another when he reasonably believes such force is necessary to defend against an imminent unlawful use of force. Break that into four checks and run them in order.

1. Unlawful threat. The force threatened must be unlawful. You may not use self-defence against a lawful arrest, even a mistaken one, in most states.
2. Imminence. The threat must be immediate, not past and not merely expected. Retaliation for yesterday is revenge; preemption against next week is not self-defence.
3. Necessity. Force must be needed, which is why the retreat question below matters.
4. Proportionality. The force used must not exceed what is needed. Deadly force, meaning force likely to cause death or serious bodily injury, is permitted only against death, serious bodily injury, or, in most states, an enumerated forcible felony such as rape, kidnapping, or robbery.

Each is filtered through the reasonable belief standard that Goetz defined. A defendant who is genuinely but unreasonably afraid fails the standard. Many states soften that result through imperfect self-defence, which reduces murder to voluntary manslaughter where the belief was honest but unreasonable, or where the defendant was the initial aggressor.

The aggressor rule and the retreat question

An initial aggressor generally forfeits the right to self-defence. He can regain it in two ways: by withdrawing from the encounter and communicating the withdrawal, or, in most jurisdictions, where he began with non-deadly force and the other party escalated to deadly force, leaving him facing death without a chance to retreat.

On retreat, American jurisdictions split, and the split is old. The majority no-retreat rule, sometimes called the true man rule, holds that a person who is not the aggressor need not retreat before using deadly force. The minority retreat rule requires retreat before deadly force if it can be done in complete safety. Every retreat jurisdiction recognises the castle doctrine exception: there is no duty to retreat from your own home.

Since 2005, when Florida enacted the first modern statute, roughly thirty states have passed stand your ground laws. These do more than codify the no-retreat rule. They typically add a presumption that a person who forcibly and unlawfully enters a dwelling or occupied vehicle intends to do serious harm, and often provide civil and criminal immunity determined at a pretrial hearing rather than by a jury at trial. That procedural feature is easy to overlook and does a great deal of work.

The empirical question has been studied. A 2017 analysis published in JAMA Internal Medicine examined Florida's homicide rates before and after the 2005 law and found a substantial increase in homicide and firearm homicide that was not matched in comparison states without such laws. Other studies have reported racial disparities in which claims succeed. Supporters answer that people should not be legally obliged to flee from an attacker and that the earlier rule effectively punished victims for judgment calls made in seconds. Both the statistical findings and the moral claim are serious, and a lesson that pretended otherwise would be lying to you.

Why this matters: Whether the law requires retreat, and who decides the immunity question, changes outcomes more than any refinement of the reasonable belief test.

The imminence problem, and battering evidence

The hardest self-defence cases involve defendants who killed an abuser during a lull rather than during an attack. State v. Norman (N.C. 1989) is the standard example. Judy Norman had endured decades of extreme violence from her husband, including being forced into prostitution and beaten with objects; she shot him while he slept. The trial court refused a self-defence instruction, and the North Carolina Supreme Court agreed: however brutal the history, the threat at the moment of the killing was not imminent.

Courts and legislatures have responded not by abolishing imminence but by changing what the jury may hear. Most states now admit expert testimony on battering and its effects, which explains why a defendant may perceive an attack as imminent that an outsider would not, why she may not have left, and why her assessment of danger may be more accurate than a stranger's. This is evidence bearing on reasonableness; it is not a separate defence, and calling it battered woman syndrome as though it were a diagnosis that excuses conduct misstates the law. Some states have also created clemency review processes for defendants convicted before such evidence was admissible.

Defending other people, and defending things

For defence of others, jurisdictions split. The majority rule lets the defender stand in the shoes of the person he protects but judges his conduct by reasonable appearances, so a person who reasonably but mistakenly believes he is stopping an assault is protected. The minority alter ego rule gives the defender exactly the rights the defended person actually had, so if the apparent victim was in fact the aggressor, the intervener is liable. The scenario that separates them is real: a plainclothes officer struggling with a suspect, and a passerby who intervenes on the wrong side.

For defence of property, the rule is short. Reasonable non-deadly force may be used to prevent an imminent unlawful taking or damage. Deadly force is never permitted to protect property alone. That is why spring guns and other mechanical devices are unlawful: in Katko v. Briney (Iowa 1971), a civil case that every criminal law course borrows, a shotgun rigged to fire at anyone opening a door in an unoccupied farmhouse maimed a trespasser, and the owners were held liable. The Model Penal Code states the principle directly, forbidding devices designed to cause death or serious harm. A person defending an occupied home may still have a self-defence claim, because there the interest at stake is the occupants, not the furniture.

Necessity, and the boat that ate the cabin boy

Necessity, the choice of evils defence, applies where breaking the law avoids a greater harm. Five requirements are standard:

• the harm to be avoided was imminent;
• there was no reasonable legal alternative;
• the harm avoided was greater than the harm caused, judged objectively;
• the defendant was not at fault in creating the situation;
• the legislature has not already resolved the balance the other way.

Easy applications: breaking into a cabin to escape a blizzard, speeding to a hospital, destroying part of a structure to create a firebreak. The Supreme Court's decision in United States v. Bailey (1980) added a practical condition in the escape context: a prisoner who flees intolerable conditions must offer evidence of a bona fide effort to surrender once the emergency has passed.

The hardest case is the oldest one. In 1884 the yacht Mignonette sank in the South Atlantic and four men escaped in an open boat. After weeks adrift, with no food or water, Dudley and Stephens killed the cabin boy, Richard Parker, who was already gravely ill, and the survivors ate his body. Rescued four days later, they were tried for murder. In R v. Dudley and Stephens (1884) the court held that necessity is no defence to murder, and it declined to make a rule that would require judges to decide whose life is worth less. The men were sentenced to death and the sentence was commuted to six months. American courts have almost uniformly followed the holding, and the Model Penal Code is the notable dissenter, permitting the defence in principle where a genuinely lesser evil is chosen.

In short: Necessity asks a court to endorse a choice between harms, which is why it is available for property and liberty and almost never for killing.

A note on police use of force

Officers act under a justification of their own, and two decisions define its outer limits. Tennessee v. Garner (1985) held that the old common law rule permitting deadly force against any fleeing felon is unconstitutional; deadly force to prevent escape requires probable cause that the suspect poses a significant threat of death or serious injury. Graham v. Connor (1989) set the standard for non-deadly force as objective reasonableness judged from the perspective of a reasonable officer on the scene, without the benefit of hindsight. Notice how closely that tracks Goetz: the same tension between an objective rule and a perspective taken from inside the moment. Use of force policy, training, and accountability are the subject of the site's Introduction to Criminal Justice course.

Common misconceptions

• Self-defence requires that you actually be in danger. A reasonable belief in the need for force suffices, and honest but unreasonable belief may reduce murder to manslaughter.
• Justification and excuse are interchangeable words. They differ in what they claim and in what follows for accomplices, resistance, and the form of the verdict.
• You may use deadly force to stop a thief. Deadly force is never justified to protect property alone; spring guns are unlawful for that reason.
• Battered woman syndrome is a defence. Expert evidence about battering informs the reasonableness of the defendant's perception; it is evidence, not a separate excuse.
• Stand your ground laws merely abolish the duty to retreat. They also commonly add presumptions of reasonable fear and pretrial immunity hearings, which change results independently.
• Necessity can justify killing an innocent person. Dudley and Stephens holds otherwise, and almost all American courts agree.

The takeaway

• Goetz settled that reasonably believes means honestly and objectively reasonable, with the defendant's circumstances admissible.
• Justifications say the act was permissible; excuses say the actor was not blameworthy, and the difference affects accomplices, resistance, and verdicts.
• Self-defence requires an unlawful, imminent threat, necessity, and proportionality; deadly force needs a deadly or forcible felony threat.
• Aggressors forfeit the defence unless they withdraw and communicate, or face escalation to deadly force.
• Retreat rules split; the castle doctrine is universal, and stand your ground statutes add presumptions and immunity hearings.
• Norman shows the imminence barrier in battering cases, which evidence rules rather than doctrine have addressed.
• Defence of others follows either a reasonable appearance or an alter ego rule; deadly force for property alone is never justified.
• Necessity requires imminence, no legal alternative, a genuinely lesser evil, clean hands, and no contrary legislative choice, and is barred for murder.

Sources

1. Legal Information Institute. (n.d.). Self-defense. Cornell Law School. law.cornell.edu
2. Wikipedia. (n.d.). People v. Goetz. en.wikipedia.org
3. Wikipedia. (n.d.). R v Dudley and Stephens. en.wikipedia.org
4. Legal Information Institute. (n.d.). Necessity. Cornell Law School. law.cornell.edu
5. Wikipedia. (n.d.). Tennessee v. Garner. en.wikipedia.org

Whitehall, January 1843

Daniel M'Naghten, a Glasgow woodturner who believed the Tory government was persecuting him, walked up behind a man in Whitehall and shot him in the back. He had aimed at Prime Minister Robert Peel. The man he shot was Edward Drummond, Peel's private secretary, who died five days later.

Nine medical witnesses testified for the defence. The jury acquitted M'Naghten on the ground of insanity, and the public reaction was ferocious enough that the House of Lords summoned the judges and put a set of questions to them. The judges' answers, delivered later that year, became the M'Naghten rules: to establish a defence on the ground of insanity, it must be proved that at the time of the act the accused was labouring under such a defect of reason, from disease of the mind, as not to know the nature and quality of the act, or, if he did know it, not to know that what he was doing was wrong.

Nearly two centuries later, roughly half the American states still use a version of that test. This lesson covers the excuses: the defences that concede the conduct was wrong and deny that this defendant should be blamed for it.

Three questions that are constantly confused

Before any test, separate three distinct legal questions that all involve mental illness.

• Insanity asks about the defendant's mental state at the time of the offence. It is a defence to the charge.
• Competence to stand trial asks about his mental state now. Under Dusky v. United States (1960), a defendant must have sufficient present ability to consult with his lawyer with a reasonable degree of rational understanding, and a rational as well as factual understanding of the proceedings. An incompetent defendant is not acquitted; the case is suspended while he is treated.
• A psychiatric diagnosis is neither. Insanity is a legal conclusion. A defendant may have a serious diagnosed illness and still fail every insanity test, and diagnosis alone establishes nothing.

Remember: Insanity is about then, competence is about now, and a diagnosis by itself decides neither.

Four tests, and what each one asks

Test	Question asked	Status
M'Naghten (1843)	Did a mental disease leave him unable to know the nature of the act, or that it was wrong?	Used in about half the states; purely cognitive
Irresistible impulse	Even if he knew, was he unable to control his conduct?	Added by some states as a supplement to M'Naghten
Durham product test (1954)	Was the act the product of mental disease or defect?	Adopted in the District of Columbia, abandoned there in 1972; survives only in New Hampshire
Model Penal Code (1962)	Did he lack substantial capacity either to appreciate the criminality of his conduct or to conform it to the law?	Adopted widely, then narrowed after 1984

Two design choices distinguish them. The first is cognitive versus volitional: M'Naghten asks only what the defendant understood, while irresistible impulse and the Model Penal Code's second clause also ask what he could control. The second is absolute versus substantial: M'Naghten's know is all or nothing, while the Code's lack of substantial capacity to appreciate admits degrees, and appreciate means more than intellectual awareness. A defendant who can recite that killing is illegal but cannot grasp what it means may satisfy the Code and fail M'Naghten.

Durham is the interesting failure. Decided by the District of Columbia Circuit in 1954, it asked simply whether the act was the product of mental disease, on the theory that psychiatry rather than law should define the categories. In practice it handed the verdict to expert witnesses, since the whole question turned on their testimony, and the same court abandoned it in 1972.

Hinckley, and the reform that followed within two years

On 30 March 1981 John Hinckley Jr. shot President Reagan and three other men outside a Washington hotel, having concluded that the act would win the attention of an actress he had never met. He was tried in federal court under the Model Penal Code test, where the prosecution bore the burden of proving sanity beyond a reasonable doubt. In June 1982 the jury acquitted him by reason of insanity.

Congress responded with the Insanity Defense Reform Act of 1984, which did three things at once. It narrowed the federal test to the cognitive prong alone, requiring that a severe mental disease or defect left the defendant unable to appreciate the nature and quality or the wrongfulness of his acts. It shifted the burden to the defendant, who must now prove insanity by clear and convincing evidence. And it barred expert witnesses from stating an opinion on the ultimate question of whether the defendant had the mental state constituting an element of the offence.

States moved too. Many narrowed their tests or shifted burdens; several adopted a guilty but mentally ill verdict, which produces a conviction and a sentence with treatment attached rather than an acquittal. Four states, Kansas, Montana, Idaho, and Utah, abolished the affirmative insanity defence altogether, permitting evidence of mental illness only to negate the mental element the prosecution must prove. In Kahler v. Kansas (2020) the Supreme Court upheld that approach, holding that due process does not require a state to adopt a moral incapacity test, and in Clark v. Arizona (2006) it had already allowed states to channel and restrict how mental illness evidence is used.

Two facts belong beside the outrage. The insanity defence is raised in roughly one percent of felony prosecutions and succeeds in a minority of those, so it accounts for a very small share of case outcomes. And defendants found not guilty by reason of insanity are typically committed to secure psychiatric facilities, frequently for longer than the prison term the conviction would have carried. Hinckley himself remained in institutional care for decades, moving to conditional release in 2016 and receiving unconditional release only in 2022.

The point: The insanity defence is rare, usually unsuccessful, and rarely a shortcut to freedom, which is the opposite of the impression the famous cases leave.

Duress: someone else's threat

Duress excuses a defendant who commits a crime because another person threatened him. The traditional elements are a threat of imminent death or serious bodily injury to the defendant or another, a well-grounded fear that the threat will be carried out, no reasonable opportunity to escape, and no fault of the defendant in exposing himself to the threat, as by joining a violent gang.

The Model Penal Code reframes it around a standard rather than a list: the defence applies where a person of reasonable firmness in the defendant's situation would have been unable to resist. That formulation drops the strict imminence requirement and drops the limitation to threats of physical harm.

The traditional rule is that duress is no defence to murder. A person may not save his own life by taking an innocent one, and the Model Penal Code is again the outlier in principle. In federal court, Dixon v. United States (2006) holds that the defendant bears the burden of proving duress by a preponderance of the evidence.

Duress and necessity are cousins, and the classic distinction is the source of the pressure: duress is a human threat and works as an excuse, while necessity is pressure from circumstances and works as a justification. The distinction is rough, and modern codes often blur it, but it explains why duress is treated as excusing a wrongful act rather than approving it.

Intoxication, which almost never helps

Voluntary intoxication is not a defence. In many states it may still negate a specific intent, so a defendant too drunk to form the intent to permanently deprive may defeat a larceny charge while remaining liable for a general intent offence. The Model Penal Code makes the point sharply for recklessness: if the defendant would have been aware of the risk had he been sober, his unawareness because of voluntary intoxication is immaterial. And states are free to go further. In Montana v. Egelhoff (1996) the Supreme Court upheld a statute barring juries from considering voluntary intoxication at all.

Involuntary intoxication is different and can be a complete defence where it produces a condition equivalent to insanity. It covers intoxication that was coerced, taken by mistake as to the substance, produced by an unexpected reaction to properly prescribed medication, or pathological, meaning grossly excessive relative to the amount consumed and unknown to the defendant.

Mistake, and the narrow exceptions to ignorance of the law

Mistake of fact is not really a defence at all. It is an argument that the prosecution cannot prove the mental element. The analysis is mechanical once you have done element analysis: if the mistake means the defendant lacked the required mental state as to an element, he is not guilty. Traditional courts asked whether the mistake was reasonable for general intent crimes and accepted any honest mistake for specific intent crimes. For a strict liability element, no mistake helps at all, which is what makes statutory rape's age element so severe.

Mistake of law is generally no defence. If it were, the more ignorant a person, the more freedom he would enjoy. Three narrow exceptions exist:

1. Official reliance. Reasonable reliance on an official statement of the law, in a statute, judicial decision, or an official interpretation by the person charged with administering the law, later determined to be invalid. Advice from your own lawyer does not count.
2. Lack of fair notice. In Lambert v. California (1957), a Los Angeles ordinance required convicted felons to register with police within five days of arriving in the city. Lambert had lived there for years and had no idea. The Court reversed her conviction: where the conduct is wholly passive, involves no circumstances that should alert a person to a duty, and the violation is unaccompanied by any activity giving notice, punishment without knowledge violates due process. The exception is real but has been read very narrowly since.
3. Knowledge of the law as an element. Some statutes, especially tax and regulatory offences, require that the defendant acted wilfully, meaning with knowledge that his conduct was unlawful. In Cheek v. United States (1991), an airline pilot stopped filing returns after being persuaded that wages are not income. The Court held that a genuine good faith misunderstanding of the tax laws negates wilfulness even if the belief is objectively unreasonable, because the statute made knowledge an element. It then drew a firm line: a belief that the tax laws are unconstitutional is not a misunderstanding of what the law requires but a disagreement with it, and it is no defence.

Key idea: Mistake of fact defeats a mental element; mistake of law defeats nothing unless the defendant relied on an official statement, had no fair notice, or faced a statute that made knowledge of the law an element.

Infancy and entrapment

Infancy at common law was tiered: a child under seven was conclusively incapable of crime, a child between seven and fourteen was presumed incapable subject to rebuttal, and a person of fourteen or over was tried as an adult. Modern practice replaced the presumptions with a separate juvenile system, with transfer or waiver provisions permitting prosecution of older juveniles as adults for serious offences. The institutional design of juvenile justice belongs to the Introduction to Criminal Justice course; what matters here is that age operates through jurisdiction rather than through an element.

Entrapment is a defence when the government originates a crime and induces someone not otherwise disposed to commit it. Two tests compete. The majority and federal subjective test focuses on the defendant's predisposition: government inducement plus a defendant who was not already ready and willing. The minority objective test, favoured by the Model Penal Code, focuses on the government's conduct and asks whether its methods created a substantial risk that an ordinary law-abiding person would commit the offence. Jacobson v. United States (1992) shows the subjective test biting: federal agents spent about two and a half years, through fictitious organisations and repeated mailings, working on a man who had once lawfully ordered magazines, until he ordered illegal material. The Court reversed, holding the government had failed to prove predisposition independent of its own campaign.

Common misconceptions

• Insanity and incompetence are the same. Insanity concerns the time of the offence and is a defence; incompetence concerns the present and suspends the case.
• An insanity acquittal means going home. Defendants found not guilty by reason of insanity are usually committed, often for longer than the corresponding prison term.
• The insanity defence is used constantly. It appears in about one percent of felony prosecutions and succeeds in a minority of those.
• Every state must offer an insanity defence. Kahler v. Kansas held that due process does not require a moral incapacity test, and four states allow mental illness evidence only to negate mens rea.
• Being drunk reduces every charge. Voluntary intoxication may negate specific intent in some states, is immaterial to recklessness under the Model Penal Code, and can be excluded entirely.
• Not knowing the law is a defence if you tried to check. Only reliance on an official statement of the law counts; advice from your own lawyer does not.

Summing up

• The M'Naghten rules came from the 1843 acquittal of Daniel M'Naghten and still govern in roughly half the states.
• Insanity, competence under Dusky, and psychiatric diagnosis are three different questions.
• The four tests differ on cognitive versus volitional capacity and on absolute versus substantial incapacity; Durham failed by handing the verdict to experts.
• The 1984 federal reform narrowed the test, shifted the burden to the defendant, and limited ultimate issue testimony after the Hinckley acquittal.
• Duress requires a human threat, imminence, no escape, and clean hands, and traditionally does not excuse murder.
• Voluntary intoxication rarely helps and may be excluded entirely; involuntary intoxication can be a full defence.
• Mistake of fact negates a mental element; mistake of law is excused only by official reliance, absent fair notice, or a wilfulness element as in Cheek.
• Infancy now operates through juvenile jurisdiction, and entrapment turns on predisposition in most courts and on government methods under the Model Penal Code.

Sources

1. Legal Information Institute. (n.d.). Insanity defense. Cornell Law School. law.cornell.edu
2. Wikipedia. (n.d.). M'Naghten rules. en.wikipedia.org
3. Wikipedia. (n.d.). Kahler v. Kansas. en.wikipedia.org
4. Wikipedia. (n.d.). Cheek v. United States. en.wikipedia.org
5. Legal Information Institute. (n.d.). Duress. Cornell Law School. law.cornell.edu

A glass booth on Sunset Boulevard

In 1965 Charles Katz was making a living transmitting wagering information on college basketball games. He used a bank of public telephone booths on Sunset Boulevard in Los Angeles, and he shut the door behind him. FBI agents had taped a microphone and recorder to the roof of two of the booths, outside the glass. They never entered, never drilled, never touched the interior.

Under the law as it then stood, that was not a search. Olmstead v. United States (1928) had held that wiretapping without a physical trespass into a constitutionally protected area fell outside the Fourth Amendment entirely, since conversations are not persons, houses, papers, or effects.

Katz v. United States, 389 U.S. 347 (1967), overruled that. The Fourth Amendment protects people, not places, the Court held. What a person knowingly exposes to the public is unprotected, but what he seeks to preserve as private, even in an area accessible to the public, may be constitutionally protected. Katz had shut the door and paid the toll; he was entitled to assume the words he spoke would not be broadcast to the world.

The rule everyone actually applies is not in the majority opinion. It is in Justice Harlan's concurrence: a search occurs where a person has exhibited an actual, subjective expectation of privacy, and that expectation is one society is prepared to recognise as reasonable. Six decades of doctrine have been built on that sentence.

The order of the questions

The Fourth Amendment protects the right of the people to be secure in their persons, houses, papers, and effects against unreasonable searches and seizures, and provides that no warrants shall issue but upon probable cause, supported by oath or affirmation, and particularly describing the place to be searched and the persons or things to be seized.

Analyse any problem in three steps, in this order:

1. Was there a search or a seizure? If not, the Fourth Amendment has nothing to say, and the analysis stops. Most of this lesson lives here.
2. Was it reasonable? Which normally means: was there a warrant based on probable cause, or does a recognised exception apply? That is the next lesson.
3. What is the remedy? Which is usually the exclusionary rule, with its own set of limits. That is the lesson after.

Two threshold limits apply throughout. The amendment restrains government action, so a purely private search, even a burglary that hands stolen documents to prosecutors, does not implicate it. And it forbids only unreasonable searches; it does not require a warrant in terms.

Key idea: Ask whether there was a search before asking whether it was reasonable, because most Fourth Amendment litigation is won and lost at the threshold.

The third-party doctrine

The most consequential consequence of Katz was not privacy protecting. If the test is what you expose, then anything you hand to somebody else is exposed.

In United States v. Miller (1976) the Court held that a bank customer has no reasonable expectation of privacy in his own cancelled checks and deposit slips, because he voluntarily conveyed them to the bank. In Smith v. Maryland (1979) it held that installing a pen register to record the numbers dialled from a home telephone was not a search, because a caller knowingly conveys those numbers to the phone company to complete the call. That is the third-party doctrine: information voluntarily disclosed to a third party carries no reasonable expectation of privacy against the government.

The doctrine was written for cancelled checks and dialled digits. Then everyone began carrying a device that continuously reports its location to a company. That collision produced the most important Fourth Amendment case of this century, which we reach below.

Where you are not protected: fields, air, garbage, and dogs

A cluster of decisions marks the outer edges of protection, and they are easier to remember as a group.

• Open fields. Land outside the curtilage is unprotected, even if fenced and posted with no trespassing signs. Hester v. United States (1924) established it and Oliver v. United States (1984) reaffirmed it: fields are not persons, houses, papers, or effects.
• Curtilage, the area immediately surrounding the home and intimately tied to its use, is protected. United States v. Dunn (1987) lists four factors: proximity to the home, whether it is enclosed with the home, how it is used, and what the resident did to shield it from observation.
• Aerial observation. Police in a fixed-wing aircraft at 1,000 feet may look into a fenced yard, California v. Ciraolo (1986), and a helicopter at 400 feet may do the same, Florida v. Riley (1989), because the public navigable airspace is a place from which anyone might look.
• Garbage. Trash left at the kerb for collection is abandoned to the public and to anyone who wants it, California v. Greenwood (1988).
• Dog sniffs. A trained dog's sniff of luggage in an airport or the exterior of a lawfully stopped car reveals only the presence of contraband, in which no one has a legitimate privacy interest, so it is not a search: United States v. Place (1983) and Illinois v. Caballes (2005). But the stop may not be extended even briefly to run the dog, Rodriguez v. United States (2015), and bringing the dog onto a front porch is a search, Florida v. Jardines (2013).

Notice the through line. The unprotected items are things you have exposed, abandoned, or placed where any member of the public might observe them. The protected ones are inside the home or its immediate surroundings.

The trespass theory comes back

For forty-five years, Katz looked like the whole test. Then the government attached a GPS tracker to the underside of Antoine Jones's Jeep and monitored it for twenty-eight days.

In United States v. Jones (2012) the Court held unanimously that this was a search, but it split badly on why. Justice Scalia's majority did not use Katz at all. It held that the government had physically occupied private property, an effect, for the purpose of obtaining information, and that this was a search when the amendment was adopted. Katz, on this view, added to the older trespass rule rather than replacing it. Concurring, Justice Alito would have used Katz and argued that long-term monitoring of every movement defeats a reasonable expectation of privacy even without a trespass. Justice Sotomayor wrote separately to question whether the third-party doctrine could survive the digital age at all.

The practical result is two tests running in parallel. A search occurs if the government physically intrudes on a constitutionally protected area to gather information, or if it violates a reasonable expectation of privacy. Jardines, the front porch dog case, was decided on the trespass ground for exactly this reason.

The upshot: Since 2012 the threshold question has two independent answers, and a defendant needs only one of them.

Technology, and the line the Court has drawn three times

Three cases show a consistent instinct: when a technology makes the government's reach categorically different rather than merely more convenient, the Court requires a warrant.

Kyllo v. United States (2001). Agents aimed a thermal imager at a home from a public street to detect the heat signature of indoor marijuana lamps. Held: where the government uses sense-enhancing technology not in general public use to obtain information about the interior of a home that could not otherwise be obtained without entry, it is a search. The home line is drawn firmly, and the general public use qualifier is doing a lot of work as technology becomes ordinary.

Riley v. California (2014). Police searched an arrestee's smartphone under the search incident to arrest exception. A unanimous Court held that phones are different in both a quantitative and qualitative sense from a wallet or a cigarette pack, holding the privacies of life; a warrant is required. The opinion is unusually direct that mechanically applying old categories to new objects produces absurd results.

Carpenter v. United States (2018). Investigating a series of robberies, the government obtained from wireless carriers 127 days of historical cell site location information for Timothy Carpenter, producing nearly thirteen thousand location points, roughly a hundred a day. Under Smith and Miller that data belonged to the carriers and was voluntarily conveyed to them. The Court held it was a search requiring a warrant, declining to extend the third-party doctrine to this category. Two features mattered: the records are a comprehensive chronicle of a person's movements, and their production is not truly voluntary, since carrying a phone is a condition of participating in modern life. The Court called the decision narrow, and it left Smith and Miller standing, so the lower courts have been sorting out its reach ever since.

Seizures, and whose rights are at stake

A seizure of property is a meaningful interference with a possessory interest. A seizure of a person occurs when, in view of all the circumstances, a reasonable person would not feel free to leave or to terminate the encounter. Two refinements matter. Under California v. Hodari D. (1991), a show of authority alone is not a seizure until the person submits, so a suspect who runs is not seized while running. Under Torres v. Madrid (2021), the application of physical force with intent to restrain is a seizure even if the person breaks free and escapes.

Finally, whose rights can be asserted? Fourth Amendment rights are personal. In Rakas v. Illinois (1978) the Court held that passengers with no possessory interest in a car or the items seized could not challenge a search of it, even though the evidence was used against them. The rule has an important counterpart: under Brendlin v. California (2007) a passenger is seized when the car is stopped, so he may challenge the lawfulness of the stop itself. Learn both, because they combine constantly in real cases: a passenger may attack the stop and lose the ability to attack the search.

In short: A defendant must show that his own expectation of privacy or possessory interest was invaded, not merely that the police behaved unlawfully toward somebody.

Common misconceptions

• The Fourth Amendment requires a warrant for every search. Its text forbids unreasonable searches; the warrant requirement is judicially developed and riddled with exceptions.
• Anything on private property is protected. Open fields are unprotected even when fenced and posted; only curtilage receives the home's protection.
• Katz means privacy is protected wherever you feel private. The expectation must be one society recognises as reasonable, which is a legal judgment, not a personal one.
• Carpenter abolished the third-party doctrine. It declined to extend it to comprehensive location records while expressly leaving Smith and Miller in place.
• Fleeing from an officer's command is a seizure. Under Hodari D. a show of authority becomes a seizure only on submission, though physical force with intent to restrain suffices under Torres.
• Any defendant can move to suppress illegally obtained evidence. Rakas requires a personal invasion of the movant's own protected interest.

What to remember

• Katz replaced the trespass-only rule with Harlan's two-part reasonable expectation of privacy test.
• Ask search or seizure first, reasonableness second, remedy third; the amendment binds government actors only.
• The third-party doctrine from Miller and Smith removes protection from information voluntarily conveyed to others.
• Open fields, aerial observation, kerbside garbage, and most dog sniffs are not searches; curtilage is protected, and Jardines shows why.
• Jones revived the trespass theory, so a search may be shown by physical intrusion or by violated expectations.
• Kyllo, Riley, and Carpenter show the Court requiring warrants when a technology changes the government's reach in kind.
• Seizure of a person turns on free to leave, with Hodari D. and Torres refining it, and Rakas limits who may complain.

Sources

1. Legal Information Institute. (n.d.). Fourth Amendment. Cornell Law School. law.cornell.edu
2. Wikipedia. (n.d.). Katz v. United States. en.wikipedia.org
3. Wikipedia. (n.d.). Carpenter v. United States. en.wikipedia.org
4. Wikipedia. (n.d.). Kyllo v. United States. en.wikipedia.org
5. Congressional Research Service. (n.d.). Constitution Annotated: Fourth Amendment. constitution.congress.gov

An anonymous letter, and the test it broke

In May 1978 the police department in Bloomingdale, Illinois, opened a handwritten letter that nobody had signed. It said that Lance and Susan Gates made their living selling drugs, that Susan would drive their car to Florida to be loaded, that Lance would fly down and drive it back, and that at that moment there was over a hundred thousand dollars worth of drugs in their basement.

An anonymous letter is close to worthless as evidence. The writer might be a rival, a jilted partner, or a fantasist, and there is no way to test any of it. So a detective went and checked. Lance Gates did have a reservation on a flight to Florida. Agents watched him take it, saw him spend the night in a room registered to his wife, and saw the couple start driving north the next morning. A magistrate issued a warrant, and the search turned up marijuana in the car and in the house.

The Illinois courts suppressed all of it. Under the two-part test built in Aguilar v. Texas (1964) and Spinelli v. United States (1969), an informant's tip supported probable cause only if the affidavit separately established the informant's veracity and his basis of knowledge. An unsigned letter established neither, and failing one prong put the tip out whatever else was in the file.

In Illinois v. Gates, 462 U.S. 213 (1983), the Court threw the two-pronged test away. Veracity and basis of knowledge remain relevant, Justice Rehnquist wrote, but as considerations in a single practical judgment rather than as independent hurdles. A magistrate is to ask whether, given everything in the affidavit, there is a fair probability that evidence of a crime will be found in the place to be searched. A weak showing on one factor can be made up by a strong showing on another, or by exactly what the Bloomingdale detective did: going out and confirming the tip's predictions.

That is the test you now apply. It is also why several state high courts, New York and Massachusetts among them, kept Aguilar and Spinelli alive under their own constitutions: a state may always protect more than the federal floor.

A boundary, repeated because this lesson touches conduct people meet in real life: this is general education about how the law works, not legal advice, and nothing here tells you what to do at a traffic stop.

What probable cause is, and what it is not

Probable cause is not a percentage, and the Court has refused to make it one. It is a fair probability, judged by what a reasonable officer would conclude from the facts, including inferences the officer's training supports. It is less than more likely than not in the sense a statistician would use, and vastly less than the trial standard.

It helps to see the ladder at once, because criminal procedure runs on the differences between these rungs.

Standard	Roughly what it takes	What it authorises
Hunch	An officer's inarticulate feeling	Nothing beyond a consensual encounter the person may walk away from
Reasonable suspicion	Specific, articulable facts plus rational inferences	A brief investigative stop; a frisk if the officer reasonably fears a weapon
Probable cause	A fair probability of criminal activity or of finding evidence in a particular place	Arrest, and the issue of a search warrant
Preponderance of the evidence	More likely than not	Contested factual questions at a suppression hearing
Beyond a reasonable doubt	No reasonable doubt left standing	Conviction, on every element

One feature of the ladder trips students constantly. Probable cause to arrest and probable cause to search are different questions: the first asks whether this person committed an offence, the second whether evidence is in this place now. You can have one without the other.

Key idea: Probable cause is a practical, common sense judgment about probabilities, not a mathematical threshold, and the question it answers depends on whether you are arresting or searching.

Four things a warrant must have

A warrant that is missing any of these is not a warrant.

1. A neutral and detached magistrate. In Coolidge v. New Hampshire (1971) the state attorney general, who was personally leading the investigation, signed the warrant as a justice of the peace. That was not a neutral magistrate. Johnson v. United States (1948) put the rationale plainly: the inferences should be drawn by a judicial officer rather than by the officer engaged in the competitive enterprise of ferreting out crime.
2. Oath or affirmation. Somebody swears to the facts and can be prosecuted for perjury if they lie.
3. Probable cause, on the face of the affidavit. The magistrate decides on what is written down, not on what the officer knew but did not say.
4. Particularity. The warrant must describe the place to be searched and the things to be seized. In Groh v. Ramirez (2004) an officer left the description of the items blank on the warrant form itself, though the application described them; the Court held the warrant invalid on its face. Particularity is what prevents a general warrant, the eighteenth century abuse the amendment was written against.

Two doctrines police the edges. Under Franks v. Delaware (1978) a defendant who makes a substantial preliminary showing that the affiant knowingly, or with reckless disregard for the truth, included a false statement necessary to the finding of probable cause is entitled to a hearing; if he proves it, the false material is set aside and the warrant is tested on what remains. And under Wilson v. Arkansas (1995) the common law requirement that officers knock and announce is part of the reasonableness inquiry, with exceptions for danger and destruction of evidence. Whether violating it gets evidence suppressed is a separate question, answered no in Hudson v. Michigan (2006).

The exceptions, laid out together

Courts recite that warrantless searches are per se unreasonable subject only to a few specifically established and well delineated exceptions. Count them and few looks generous.

Exception	What triggers it	How far it reaches	Anchor case
Search incident to arrest	A lawful custodial arrest	The person and the area within immediate control; not phone data	Chimel (1969); Riley (2014)
Automobile	Probable cause that a readily mobile vehicle holds evidence	Anywhere in it the object could be, containers included	Carroll (1925); Ross (1982)
Exigent circumstances	Hot pursuit, imminent destruction of evidence, or a need to render aid	Only as far as the emergency requires	Warden v. Hayden (1967)
Consent	Voluntary consent by someone with actual or apparent authority	What a reasonable person would understand it to cover	Schneckloth (1973); Matlock (1974)
Plain view	Lawful vantage, lawful access, incriminating character immediately apparent	Seizure of that item only, with no further searching	Horton (1990); Hicks (1987)
Terry stop and frisk	Reasonable suspicion, plus reasonable fear of a weapon to frisk	Brief detention; pat down of outer clothing for weapons	Terry (1968)
Inventory	Lawful impoundment or booking, under a standardised policy	What the policy authorises, not an investigatory rummage	Opperman (1976)
Special needs and checkpoints	A programmatic purpose beyond ordinary crime control	Limited, suspicionless intrusion matched to that purpose	Sitz (1990); Edmond (2000)

Three repay a closer look.

Gant, Collins, and Lange: three trims

Arizona v. Gant (2009). Rodney Gant was arrested for driving with a suspended licence, handcuffed, and locked in a patrol car. Officers then searched his car and found cocaine in a jacket pocket. For twenty-eight years New York v. Belton (1981) had been read to allow exactly that: arrest an occupant, search the passenger compartment. The Court cut it back. Police may search a vehicle incident to arrest only if the arrestee is unsecured and within reaching distance, or if it is reasonable to believe the vehicle contains evidence of the offence of arrest. Gant was handcuffed in a squad car, and no evidence of driving on a suspended licence lives in a glovebox. The search failed both branches.

Collins v. Virginia (2018). An officer walked up a driveway, lifted a tarpaulin off a motorcycle parked beside the house, and read its identification number. The automobile exception covers vehicles, the state argued. The Court answered that it does not license entry onto the curtilage of a home to reach one. An exception does not dissolve another protection on contact.

Lange v. California (2021). An officer activated his lights just as a driver reached his own driveway, then followed him into the attached garage. Pursuit of a fleeing suspect, the state argued, is always an exigency. The Court refused to make it categorical for misdemeanours: the officer needs an actual, case-specific reason such as violence, escape, or destruction of evidence. Compare Kentucky v. King (2011), upholding an entry to prevent the destruction of drugs even though the police knocking caused the scrambling inside, and Missouri v. McNeely (2013), holding that the natural dissipation of alcohol in the blood is not a per se exigency justifying a warrantless blood draw.

What matters here: An exception is defined by the justification that created it. When the justification is absent, as with a handcuffed arrestee or a licence offence that leaves no physical trace, the exception is absent too.

Consent, which does most of the work in practice

Far more warrantless searches rest on consent than on any other doctrine in the table, and the law of consent is more permissive than most people expect. Schneckloth v. Bustamonte (1973) held that consent must be voluntary on the totality of the circumstances, and that police need not tell the person he is free to refuse. Knowledge of the right to refuse is one factor, not a requirement. That single holding explains a great deal about roadside encounters.

Whose consent counts? Under United States v. Matlock (1974), a third party with common authority over the premises may consent, because co-inhabitants assume the risk that one of them will let someone in. Under Illinois v. Rodriguez (1990) it is enough that the officers reasonably believed the consenter had authority, even if she did not. But under Georgia v. Randolph (2006), when one occupant is physically present and expressly refuses, another occupant's consent does not override the refusal. And Fernandez v. California (2014) narrowed that: once the objecting occupant has been lawfully removed, the remaining occupant may consent.

Read those four together and the shape of the rule appears. Consent is easy for the state to obtain, hard for a defendant to undo, and the objection of an absent person carries no weight.

Terry, and the stop that is not an arrest

In October 1963 a Cleveland detective named Martin McFadden watched two men take turns walking past a shop window, peering in, and returning to confer. Suspecting a robbery was being cased, he approached them, patted down the outside of their coats, and found guns. There was no probable cause to arrest anyone for anything.

Terry v. Ohio, 392 U.S. 1 (1968) held that the encounter was a seizure and the pat down a search, but that both could be reasonable on less than probable cause. The officer needs specific and articulable facts, plus rational inferences, that criminal activity is afoot. The frisk is separately justified only by a reasonable belief that the person is armed and dangerous, and is limited to outer clothing.

The refinements that matter:

• Plain feel. Minnesota v. Dickerson (1993): if the incriminating character of an object is immediately apparent from the lawful pat down, it may be seized; but the officer in that case had to squeeze and manipulate a lump to work out what it was, which exceeded the frisk.
• Flight. Illinois v. Wardlow (2000): unprovoked headlong flight in a high crime area can supply reasonable suspicion, though neither factor alone would.
• Anonymous tips. Florida v. J.L. (2000): a bare anonymous tip that a young man at a bus stop has a gun does not justify a stop, since it shows no knowledge of concealed criminal activity. Navarette v. California (2014) came out the other way where a 911 caller reported being run off the road, giving a vehicle, a location, and an eyewitness basis.
• Identification. Hiibel v. Sixth Judicial District Court (2004) upheld a state law requiring a person lawfully stopped to give his name.

In short: A Terry stop trades a lower threshold for a narrower intrusion, and the frisk is a weapons search, not an evidence search.

Whren, and the question the Court refused to ask

Plainclothes vice officers in an unmarked car in Washington, D.C. watched a truck wait at a stop sign an unusually long time and then turn without signalling. They stopped it for the traffic violations and saw crack cocaine in the driver's hands. The defendants argued that no reasonable officer patrolling vice would have made that stop for a turn signal, so the traffic offences were a pretext.

Whren v. United States, 517 U.S. 806 (1996), rejected the argument unanimously. If there is probable cause to believe a traffic violation occurred, the stop is reasonable, and the officer's actual motivation is irrelevant. Selective enforcement based on race is unconstitutional, the Court said, but under the Equal Protection Clause. That relocation matters: an equal protection claim requires proof of discriminatory purpose and effect, which is far harder than a suppression motion.

Atwater v. City of Lago Vista (2001) went further: a full custodial arrest for a fine-only seatbelt offence does not violate the Fourth Amendment.

There is one place where purpose does control: checkpoints. Michigan Department of State Police v. Sitz (1990) upheld sobriety checkpoints. City of Indianapolis v. Edmond (2000) struck down a checkpoint whose primary purpose was ordinary crime control, namely drug interdiction. Illinois v. Lidster (2004) upheld a checkpoint seeking information about a hit and run committed a week earlier. So an individual officer's motive is irrelevant while a programme's articulated purpose is decisive. Students find that inconsistent. The Court's answer is that a programme has a stated institutional purpose a court can evaluate, and an officer's mind does not.

Common misconceptions

• Probable cause means more than fifty per cent. The Court has declined to quantify it; a fair probability is deliberately less demanding than a preponderance.
• Anonymous tips cannot support a search. Gates says they can, when corroboration or other indicia make the totality sufficient.
• If police do not tell you that you may refuse, your consent is invalid. Schneckloth holds the opposite; the warning is a factor, not a prerequisite.
• An arrest lets police search the whole car. Only under the two branches Gant left standing, or under a separate exception such as the automobile exception or inventory.
• A frisk lets an officer look for drugs. The frisk is justified by officer safety and is limited to weapons, subject to the narrow plain feel rule.
• A pretextual traffic stop is unconstitutional. After Whren the challenge belongs under equal protection, and it is much harder to win there.

Recap

• Gates replaced the rigid Aguilar and Spinelli prongs with a totality of the circumstances test asking for a fair probability.
• A valid warrant needs a neutral magistrate, an oath, probable cause on the face of the affidavit, and particularity; Franks polices lies in the affidavit.
• The warrant requirement is surrounded by exceptions, each defined by the justification that created it, and Gant, Collins, and Lange show the Court refusing to let one outrun its rationale.
• Consent searches are governed by voluntariness without any warning requirement, and by common or apparent authority.
• Terry allows a brief stop on reasonable suspicion and a weapons frisk on a separate reasonable fear.
• After Whren an officer's motive is irrelevant if probable cause exists, though a checkpoint programme's purpose is decisive.

Sources

1. Legal Information Institute. (n.d.). Probable cause. Cornell Law School. law.cornell.edu
2. Legal Information Institute. (n.d.). Terry stop. Cornell Law School. law.cornell.edu
3. Wikipedia. (n.d.). Illinois v. Gates. en.wikipedia.org
4. Wikipedia. (n.d.). Terry v. Ohio. en.wikipedia.org
5. Congressional Research Service. (n.d.). Constitution Annotated: Fourth Amendment. constitution.congress.gov

Three officers, a piece of paper, and a trunk in the basement

In May 1957 three Cleveland police officers came to Dollree Mapp's house looking for a man wanted for questioning about a bombing. She telephoned her lawyer and refused to let them in without a warrant. They waited. Hours later more officers arrived, forced a door, and when Mapp demanded to see the warrant one of them held up a piece of paper. She grabbed it and put it inside her blouse. An officer retrieved it by reaching in. No warrant was ever produced at trial, and no witness ever explained what the paper was.

The officers did not find the bombing suspect. In a trunk in the basement they found books and photographs, and Ohio convicted Mapp of knowingly possessing obscene material. Her sentence was one to seven years.

Everyone agreed the search was unlawful. The question in Mapp v. Ohio, 367 U.S. 643 (1961), was what follows from that. And the answer the Court gave, that unlawfully obtained evidence may not be used in a state criminal trial, is the most consequential and the most contested remedy in American criminal procedure. It is worth being clear at the outset that reasonable and well informed people disagree about it, and that the Supreme Court itself has spent sixty years steadily narrowing what it announced.

Weeks, Wolf, and the forty-seven years in between

The Fourth Amendment says what the government may not do. It says nothing about what happens if the government does it anyway. For most of American history the answer was: nothing much. The evidence came in, and the person whose door was broken could sue in trespass, which meant suing an officer with no money before a jury unlikely to care.

Weeks v. United States, 232 U.S. 383 (1914), created the remedy for federal cases. Federal marshals had searched Fremont Weeks's house twice without a warrant. A unanimous Court held that the papers had to be returned and could not be used against him, reasoning that if letters seized in violation of the amendment could be held and used in evidence, the protection was of no value.

But Weeks bound only the federal government, and until 1949 the Fourth Amendment itself did not apply to the states. Wolf v. Colorado (1949) took the first half of the step and refused the second: the core of the Fourth Amendment is enforceable against the states through the Fourteenth Amendment's due process clause, but the exclusionary rule is not required, because states might devise other remedies. For twelve years that left an odd arrangement in which state police could seize evidence unlawfully and hand it to federal prosecutors, a practice known as the silver platter doctrine, ended in Elkins v. United States (1960).

Mapp overruled the remedial half of Wolf. Justice Clark's opinion gave two reasons. The first was deterrence: without exclusion, the constitutional right is a form of words. The second was judicial integrity, the idea that courts should not be receivers of stolen goods, or, in Justice Brandeis's older phrase from his Olmstead dissent, that the government should not become a lawbreaker to catch lawbreakers. Hold on to the fact that Mapp gave two rationales, because the Court has since discarded one of them.

Remember: Exclusion is a judicially created remedy, not a right the amendment names. Everything the Court has done since 1961 follows from that classification.

Fruit of the poisonous tree

Suppressing the thing seized is only the start. Suppose an unlawful search of an office turns up a ledger, and the ledger names a supplier, and the supplier testifies. Is the testimony in?

Silverthorne Lumber Co. v. United States (1920) supplied the principle. Federal agents unlawfully seized company books, photographed them, and after being ordered to return them tried to subpoena the same documents. Justice Holmes wrote that the essence of forbidding the acquisition of evidence in a certain way is that it shall not be used at all, and that knowledge gained by the government's own wrong may not be used in this way. The phrase fruit of the poisonous tree came later, from Justice Frankfurter in Nardone v. United States (1939).

Wong Sun v. United States (1963) gave the test its modern shape and, more usefully, gave it a limit. The question is not simply whether the evidence would have come to light but for the illegality. It is whether the evidence was come at by exploitation of the illegality, or instead by means sufficiently distinguishable to be purged of the primary taint. That sentence is doing all the work in the three doctrines below.

Three ways the fruit gets in anyway

Independent source. If the evidence was in fact obtained through a lawful route that did not depend on the illegal one, it is admissible. In Murray v. United States (1988) agents unlawfully entered a warehouse, saw bales of marijuana, left, and obtained a warrant without mentioning what they had seen and without relying on it. The Court held that if the later warrant was genuinely independent, meaning the decision to seek it was not prompted by what they saw and the affidavit did not include it, the evidence comes in.

Inevitable discovery. If the evidence would inevitably have been found by lawful means, the prosecution may show that by a preponderance of the evidence. Nix v. Williams (1984) is the case: a defendant's unlawfully obtained statements led police to a child's body, but a search party of two hundred volunteers was already working a grid that would have reached the location within hours. Note the difference from independent source. Independent source asks what the police actually did; inevitable discovery asks what they would have done.

Attenuation. Time, intervening events, and the flagrancy of the misconduct can break the chain. Brown v. Illinois (1975) lists the factors: how much time passed, whether there were intervening circumstances, and, importantly, the purpose and flagrancy of the official misconduct.

Utah v. Strieff (2016) shows how much that doctrine can carry. An officer conducting surveillance on a suspected drug house stopped Edward Strieff after he left it. The stop was conceded to be unlawful. The officer ran his name, found an outstanding traffic warrant, arrested him on it, searched him incident to the arrest, and found methamphetamine. The Court held the evidence admissible: the valid pre-existing warrant was an intervening circumstance, and the officer's error was at most negligent rather than flagrant. Justice Sotomayor's dissent argued that in jurisdictions with very large numbers of outstanding minor warrants the holding gives police a reason to make unlawful stops and then check for a warrant.

Leon, and the turn to deterrence

In United States v. Calandra (1974) the Court described the rule as a judicially created remedy designed to safeguard Fourth Amendment rights generally through its deterrent effect, rather than a personal constitutional right of the party aggrieved. Once that is the premise, the analysis becomes a cost-benefit question: exclude only where exclusion will actually deter police.

United States v. Leon, 468 U.S. 897 (1984), drew the conclusion. Officers obtained a warrant that a magistrate issued and a reviewing court later found unsupported by probable cause. The Court held the evidence admissible, because suppressing it would punish the officers for the magistrate's error and would deter nothing. That is the good faith exception: evidence obtained in objectively reasonable reliance on a warrant later held invalid is not suppressed.

Leon named four situations where reliance is not reasonable, and they are worth memorising because they are where a defendant still wins: the affiant lied or was reckless with the truth, the magistrate wholly abandoned his judicial role, the affidavit was so lacking in indicia of probable cause that belief in it was unreasonable, and the warrant was so facially deficient that no officer could reasonably presume it valid.

The core of it: After Leon, the question stopped being whether the search was lawful and became whether suppressing this evidence would change future police behaviour.

Herring and Davis: how far negligence goes

The good faith idea then travelled beyond warrants. Illinois v. Krull (1987) applied it to reliance on a statute later held unconstitutional. Arizona v. Evans (1995) applied it to a court clerk's database error. Herring v. United States (2009) applied it to an error by police employees themselves: a neighbouring county's warrant database wrongly showed an active warrant that had been recalled months earlier. Chief Justice Roberts wrote that to trigger the rule, police conduct must be sufficiently deliberate that exclusion can meaningfully deter it and sufficiently culpable that deterrence is worth the price. Isolated negligence attenuated from the search does not qualify. Davis v. United States (2011) added reliance on binding appellate precedent that is later overruled, which is how the officers in the post-Gant cases kept their evidence.

Read the sequence and you can see the trajectory. In 1961 the rule applied to unlawful searches. By 2009 it applied to unlawful searches carried out deliberately, recklessly, or with gross or systemic negligence.

Where the rule does not apply at all

• Grand jury proceedings. A witness may not refuse to answer questions based on unlawfully obtained evidence: Calandra.
• Impeachment. A defendant who testifies may be confronted with suppressed evidence to contradict him: Walder v. United States (1954) and United States v. Havens (1980). The rule is a shield, not a licence to lie.
• Parole and probation revocation. Pennsylvania Board of Probation and Parole v. Scott (1998).
• Knock and announce violations. Hudson v. Michigan (2006): the interests protected by the knock rule, such as avoiding property damage and preserving dignity, have nothing to do with the seizure of the evidence, so suppression is not the remedy.
• Civil, immigration, and most administrative proceedings.
• Anyone but the person whose rights were violated. This is the standing point from Rakas in the previous lesson.

The argument, and what would settle it

Here are the two positions in their strongest form, which is the only way to judge them.

The case against. Exclusion punishes the public rather than the officer. The officer keeps his job and his salary; the person who suffers is the victim of the crime whose case collapses. It is also badly targeted: it does nothing for the innocent person whose home is wrongly searched, because there is no prosecution in which to suppress anything, and it rewards only the guilty, in rough proportion to how much evidence the police found. Chief Justice Burger, dissenting in Bivens in 1971, urged replacing it with a statutory damages scheme. Judge Henry Friendly made the same argument in the academic literature. Other common law countries manage with discretionary exclusion: in Canada, section 24(2) of the Charter directs a court to exclude only where admission would bring the administration of justice into disrepute, and English law under the Police and Criminal Evidence Act 1984 gives the trial judge a discretion rather than a rule.

The case for. The alternatives do not work, and this is an empirical claim rather than a slogan. Damages actions under 42 U.S.C. section 1983 and, against federal officers, under Bivens v. Six Unknown Named Agents (1971), run into qualified immunity, which protects officials unless they violated clearly established law, and the Court has spent two decades declining to extend Bivens to new contexts, most recently in Egbert v. Boule (2022). Internal discipline is administered by the same institution being deterred. Meanwhile the measured costs of exclusion are small: reviewing the available studies for the American Bar Foundation in 1983, Thomas Davies found that the cumulative loss of felony arrests attributable to illegal searches was on the order of one to two per cent, concentrated almost entirely in drug and weapons cases. And there is direct evidence of the rule changing behaviour, since police departments across the country wrote warrant procedures, training programmes, and legal advice into their operations after 1961 in a way they had not before.

What would settle it. The honest answer is that the central question is very hard to measure. To know whether exclusion deters, you would need to observe searches that were never made because an officer expected suppression, and unmade searches leave no record. Suppression motion rates measure the visible tail of police conduct, not the invisible part the rule is supposed to shape. A convincing study would need something close to a natural experiment: a jurisdiction where the remedy changed sharply while everything else held constant, with measurement of search quality rather than case outcomes. Mapp itself was such a moment for the twenty-odd states that had no exclusionary rule in 1961, which is why the studies of that period remain the most cited evidence on both sides.

Bottom line: Whether you think the rule is worth its costs depends less on the Fourth Amendment than on how much you believe the alternatives could be made to work.

Common misconceptions

• Illegally obtained evidence is always thrown out. Good faith, independent source, inevitable discovery, and attenuation each let it in, and several proceedings are outside the rule entirely.
• The exclusionary rule is in the Fourth Amendment. The amendment provides no remedy; exclusion is judge-made, which is why the Court can and does adjust it.
• If the police broke the rules, the case gets dismissed. Suppression removes evidence. The prosecution may proceed on whatever is left, and often does.
• Good faith means the officer meant well. The test is objective reasonableness, not sincerity, and Leon lists four situations where reliance is unreasonable however honest.
• Mapp settled the question. The Court has narrowed the rule repeatedly since 1974 by treating deterrence as its only justification.
• Suppressed evidence can never be mentioned at trial. Under Havens it may be used to impeach a defendant who takes the stand and contradicts it.

Where this leaves us

• Weeks created the federal rule in 1914; Wolf applied the amendment to the states without the remedy; Mapp supplied the remedy in 1961.
• Fruit of the poisonous tree extends suppression to derivative evidence, subject to Wong Sun's exploitation and purged taint test.
• Independent source asks what police actually did; inevitable discovery asks what they would have done; attenuation weighs time, intervening events, and flagrancy.
• Leon created the good faith exception and listed four circumstances in which reliance on a warrant is still unreasonable.
• Herring and Davis extended good faith to police negligence and to reliance on later overruled precedent.
• The rule does not reach grand juries, impeachment, revocation hearings, knock and announce violations, or defendants without standing.
• The dispute is empirical: whether exclusion deters, and whether damages actions could be made to work instead.

Sources

1. Legal Information Institute. (n.d.). Exclusionary rule. Cornell Law School. law.cornell.edu
2. Wikipedia. (n.d.). Mapp v. Ohio. en.wikipedia.org
3. Wikipedia. (n.d.). United States v. Leon. en.wikipedia.org
4. Congressional Research Service. (n.d.). Constitution Annotated: Fourth Amendment. constitution.congress.gov
5. Davies, T. Y. (1983). A hard look at what we know (and still need to learn) about the costs of the exclusionary rule. American Bar Foundation Research Journal, 1983(3).

Kemper County, Mississippi, 1934

A white planter named Raymond Stewart was killed in Kemper County, Mississippi, in March 1934. Within days three Black tenant farmers, Ed Brown, Henry Shields, and Yank Ellington, had confessed. Ellington confessed after being hanged by a rope from a tree twice, let down, and whipped. Brown and Shields confessed after being stripped and beaten across the back with a leather strap fitted with buckles, and were told the whipping would continue until they talked. It did, and they talked.

None of this was in dispute. A deputy testified to the whipping at trial and, asked how severe it had been, gave an answer so casually brutal that it has been quoted ever since. The trial lasted a day. All three were convicted and sentenced to death on the confessions alone.

In Brown v. Mississippi, 297 U.S. 278 (1936), the Supreme Court reversed unanimously. Chief Justice Hughes wrote that the rack and the torture chamber may not be substituted for the witness stand, and that a conviction resting solely on confessions extracted by torture is a denial of due process. This is where the modern law of interrogation starts: not with warnings, but with the proposition that there are things a state may not do to obtain the truth.

Everything after 1936 is the story of the Court trying to find a rule that would work in the ordinary case, where nobody is whipped and the pressure is quieter.

The due process test, and why it stopped being enough

For thirty years the standard was voluntariness, assessed on the totality of the circumstances. Was the defendant's will overborne? Courts weighed the length of questioning, whether he was fed and allowed to sleep, his age, education, and mental condition, whether he was held incommunicado, and whether threats or promises were made.

The cases show the range. Ashcraft v. Tennessee (1944): thirty-six hours of continuous questioning by relays of officers under an electric light, without sleep or rest, was inherently coercive. Spano v. New York (1959): a suspect's childhood friend, by then a rookie police officer, falsely told him that his job and his pregnant wife's welfare depended on a confession, and the Court found the will overborne.

The test had two problems, and both were fatal. First, it was almost impossible to apply in advance. An officer planning an interrogation could not know which combination of hours, tactics, and personal characteristics a court would later find over the line, and neither could a trial judge. Second, and worse, the facts were contested in exactly the cases where it mattered. Interrogations happened in a closed room with no recording, and the resulting hearing was the defendant's account against several officers'. Judges resolved those swearing contests in the predictable direction.

Escobedo v. Illinois (1964) tried one fix. Danny Escobedo asked repeatedly to see his lawyer, who was in the police station trying to reach him; both were refused. The Court reversed, but on reasoning nobody could generalise. Two years later it tried again, and this time produced a rule.

Miranda, and what it actually requires

Ernesto Miranda was arrested at his home in Phoenix in March 1963, taken to an interrogation room, and questioned for about two hours by two detectives. He emerged with a written confession, signed above a typed paragraph reciting that the statement was voluntary and made with full knowledge of his legal rights. Nobody had told him what those rights were.

Miranda v. Arizona, 384 U.S. 436 (1966), decided with three companion cases, held five to four that custodial interrogation is inherently compelling, and that the Fifth Amendment privilege against compelled self-incrimination therefore requires procedural safeguards. Before any custodial questioning, a suspect must be told:

1. that he has the right to remain silent;
2. that anything he says can be used against him in court;
3. that he has the right to the presence of an attorney; and
4. that if he cannot afford an attorney, one will be appointed for him before questioning if he wishes.

No magic words are required, so long as the substance is conveyed. If the warnings are not given, statements obtained are inadmissible in the prosecution's case in chief, whatever their truth.

One detail corrects a common assumption about what suppression does. Miranda's confession was thrown out. He was retried without it, convicted on other evidence, and served his sentence.

The point: Miranda did not replace the voluntariness test. A confession must still be voluntary under due process; the warnings are an additional requirement layered on top of that one.

Custody and interrogation, both of which are terms of art

Warnings are owed only when both boxes are ticked. Miss either and there is no violation.

Custody means formal arrest or a restraint on freedom of movement of the degree associated with formal arrest, judged objectively from the perspective of a reasonable person in the suspect's position. An ordinary roadside traffic stop is not custody, because it is brief, public, and involves few officers: Berkemer v. McCarty (1984). Under J.D.B. v. North Carolina (2011), a child's age is part of the objective analysis when it was known to the officer or objectively apparent, because a reasonable thirteen year old questioned by police in a school conference room does not feel free to leave when an adult would.

Interrogation means express questioning or its functional equivalent. Rhode Island v. Innis (1980) defined the equivalent as words or actions police should know are reasonably likely to elicit an incriminating response. Two officers driving a suspect past a school for children with disabilities remarked to each other that it would be terrible if a child found the missing shotgun. He told them where it was. The Court held this was not the functional equivalent of questioning, which tells you how much room the standard leaves.

There is one open exception. New York v. Quarles (1984): officers chased an armed rape suspect into a supermarket, found him wearing an empty shoulder holster, handcuffed him, and asked where the gun was before giving any warnings. The Court recognised a public safety exception, where the questions are reasonably prompted by a concern for public safety rather than by a desire to gather evidence.

Waiver, and the strange asymmetry of invocation

A suspect may waive his rights, and most do. The waiver must be knowing, intelligent, and voluntary, but under North Carolina v. Butler (1979) it need not be express: a court may infer it from the suspect's conduct after warnings.

Berghuis v. Thompkins (2010) shows how far that goes. Van Chester Thompkins sat through nearly three hours of questioning saying almost nothing. Asked whether he prayed to God to forgive him for shooting the boy, he said yes. The Court held that his long silence was not an invocation of the right to remain silent, and that by answering he impliedly waived it. To stop the questioning, he had to say so.

So the asymmetry is this. Silence does not invoke the right to silence; you must invoke unambiguously. And under Davis v. United States (1994) a statement like a suggestion that maybe he should talk to a lawyer is not an unambiguous request, so officers may keep going and need not ask a clarifying question.

Invoke clearly, though, and the protection is strong. Under Edwards v. Arizona (1981), once a suspect requests counsel, police may not reinitiate interrogation until counsel is present, even about a different offence: Arizona v. Roberson (1988). Maryland v. Shatzer (2010) put a limit on it, holding that a break in Miranda custody of fourteen days ends the Edwards protection.

One case marks the outer boundary of the waiver rules. In Moran v. Burbine (1986) a lawyer retained by the suspect's sister telephoned the station and was falsely told he would not be questioned that night. The suspect, who did not know a lawyer was trying to reach him, waived and confessed. The Court upheld the waiver: events unknown to a suspect cannot affect whether his own decision was knowing and voluntary.

Why this matters: The rules reward a suspect who says one clear sentence and punish one who is merely uncooperative, which is close to the opposite of what most people assume.

What a Miranda violation costs the state, and what it does not

Miranda is a prophylactic rule, and the remedy has been trimmed accordingly.

• Impeachment. Statements taken in violation of Miranda, if voluntary, may be used to contradict a defendant who testifies: Harris v. New York (1971).
• Physical fruits. United States v. Patane (2004): a gun found because of an unwarned but voluntary statement is admissible. Contrast the Fourth Amendment fruit doctrine from the last lesson; the two remedies do not work the same way.
• A later warned statement. Oregon v. Elstad (1985) admitted a warned confession that followed an earlier unwarned one, where the first was voluntary and the failure to warn was inadvertent. But Missouri v. Seibert (2004) suppressed the second statement where officers had deliberately used a question first, warn later protocol designed to make the warnings meaningless.
• Damages. Vega v. Tekoh (2022) held that a Miranda violation is not itself a violation of the Fifth Amendment right for which a police officer can be sued under 42 U.S.C. section 1983. The remedy is suppression, and that is all.

Against that narrowing sits Dickerson v. United States (2000). Two years after Miranda, Congress had enacted a statute purporting to make voluntariness the sole test for admitting confessions in federal court. When the question finally reached the Court, Chief Justice Rehnquist, who had spent his career criticising Miranda, wrote the opinion holding that it is a constitutional decision that Congress may not supersede by statute, and declined to overrule it.

The other right to counsel

The Miranda right to counsel is a Fifth Amendment device for protecting against compelled self-incrimination. The Sixth Amendment right to counsel is a different animal with a different trigger, and confusing them is the most common error students make here.

Question	Fifth Amendment (Miranda)	Sixth Amendment (Massiah)
When does it apply?	Custodial interrogation, at any stage	Once adversary judicial proceedings begin, meaning the first appearance under Rothgery (2008)
What triggers a violation?	Questioning without warnings, or after invocation	Deliberate elicitation by a government agent without counsel or waiver
Offence-specific?	No; invocation covers any offence (Roberson)	Yes; it attaches only to the charged offence (Texas v. Cobb, 2001)
Undercover agents?	No violation; there is no police-dominated atmosphere (Illinois v. Perkins, 1990)	Violation if the agent deliberately elicits (Massiah, 1964; United States v. Henry, 1980), but not mere passive listening (Kuhlmann v. Wilson, 1986)
Must the suspect ask?	Yes, unambiguously	No; it attaches automatically at the critical stage

The Sixth Amendment right also carries the promise most people associate with the phrase. Powell v. Alabama (1932), from the Scottsboro prosecutions, required counsel in a capital case as a matter of due process; Betts v. Brady (1942) refused to generalise it.

Then Clarence Earl Gideon, charged with breaking into a poolroom in Panama City, Florida, asked the trial court to appoint him a lawyer, was told Florida did so only in capital cases, defended himself, and lost. From prison he wrote a petition in pencil. Gideon v. Wainwright, 372 U.S. 335 (1963), overruled Betts unanimously: the right to appointed counsel in felony cases is fundamental and applies to the states. Gideon was retried with a lawyer and acquitted.

The line moved twice more. Argersinger v. Hamlin (1972) and Scott v. Illinois (1979) settled that it attaches wherever actual imprisonment is imposed, whatever the offence is called, and Alabama v. Shelton (2002) extended it to suspended sentences.

Having a lawyer is not the same as having an effective one. Strickland v. Washington (1984) set the two-part test: the defendant must show that counsel's performance fell below an objective standard of reasonableness, and that there is a reasonable probability that, but for the errors, the result would have been different. Courts apply a strong presumption that the challenged action was sound strategy, which is why Strickland claims very rarely succeed.

Why innocent people confess

Students find this the hardest thing in the lesson to accept, so start with a case. In 1989 five teenagers from Harlem confessed on videotape to the rape and beating of a jogger in Central Park. Their accounts contradicted each other and the physical evidence, and all five were convicted. In 2002 a man already serving a life sentence confessed, DNA matched him alone, and the convictions were vacated.

Research on interrogation explains how that happens. The dominant American method, associated with the Reid technique, is guilt-presumptive: the interrogator has already decided the suspect did it, confronts him with unwavering certainty, cuts off denials, and then offers a face-saving explanation that minimises the moral seriousness of the act while maximising the perceived cost of continued denial. Officers may lawfully lie about evidence, and often do, since Frazier v. Cupp (1969) approved a false claim that an accomplice had confessed. Under enough hours of that, some people conclude that confessing is the only way the session ends.

Three groups are far more vulnerable than the average adult: juveniles, people with intellectual disabilities, and people with mental illness. And a confession, once in the file, is unusually corrosive, because it tends to change how everyone downstream reads the rest of the evidence, from witnesses to forensic analysts.

The reform with the broadest support is also the simplest: record the whole interrogation. Alaska and Minnesota required it under their own constitutions in the 1980s and 1990s, many states have followed by statute, and the federal Department of Justice adopted a presumption of recording custodial interviews in 2014. Recording does not stop coercion, but it ends the swearing contest that made the voluntariness test unworkable in 1966.

Worth holding on to: Miranda regulates the warning, not the interrogation. Almost everything that produces false confessions remains lawful after the warnings are read and waived.

Common misconceptions

• Police must read you your rights when they arrest you. Warnings are owed before custodial interrogation. An arrest with no questioning requires none.
• If the warnings were not given, the case is dismissed. Only the statement is suppressed, and even then it may be used to impeach, and physical fruits survive under Patane.
• Staying silent invokes the right to silence. Berghuis holds the opposite: invocation must be unambiguous.
• Asking whether you should get a lawyer stops the interview. Under Davis, an ambiguous reference to counsel does not, and officers need not seek clarification.
• Police may not lie to a suspect. Deception about evidence is generally lawful; it is one of the reasons false confessions occur.
• Innocent people do not confess. Documented exonerations and decades of research say otherwise, especially for juveniles and people with intellectual disabilities.

What to carry forward

• Brown v. Mississippi established that coerced confessions violate due process; voluntariness remains a separate requirement to this day.
• Miranda requires four warnings before custodial interrogation, and both custody and interrogation are defined terms.
• Berkemer, J.D.B., and Innis set the boundaries of those two terms; Quarles supplies the public safety exception.
• Waiver may be implied; invocation must be unambiguous, and Edwards then bars reinitiation subject to Shatzer's fourteen-day rule.
• A Miranda violation costs the state its case in chief only: impeachment, physical fruits, and damages actions all survive it after Harris, Patane, and Vega.
• The Sixth Amendment right attaches at the first appearance, is offence-specific, and is measured for adequacy by Strickland.
• False confessions are real, concentrated among juveniles and vulnerable suspects, and recording is the reform with the broadest support.

Sources

1. Legal Information Institute. (n.d.). Miranda warning. Cornell Law School. law.cornell.edu
2. Legal Information Institute. (n.d.). Right to counsel. Cornell Law School. law.cornell.edu
3. Wikipedia. (n.d.). Miranda v. Arizona. en.wikipedia.org
4. Wikipedia. (n.d.). Gideon v. Wainwright. en.wikipedia.org
5. Kassin, S. M., Drizin, S. A., Grisso, T., Gudjonsson, G. H., Leo, R. A., and Redlich, A. D. (2010). Police-induced confessions: Risk factors and recommendations. Law and Human Behavior, 34(1).
6. Lewis, A. (1964). Gideon's Trumpet. Random House.

One withheld page

William Brooks was killed in Maryland in 1958. John Brady and Donald Boblit were both convicted of first-degree murder and sentenced to death. Brady admitted taking part in the robbery but told the jury that Boblit had done the actual killing, and his lawyer asked the prosecution to hand over Boblit's statements. The prosecution produced several of them. It did not produce the one in which Boblit admitted strangling Brooks himself. Brady's lawyer learned of that page only after the conviction had been affirmed on appeal.

In Brady v. Maryland, 373 U.S. 83 (1963), the Supreme Court held that suppression by the prosecution of evidence favourable to an accused, on request, violates due process where the evidence is material to guilt or to punishment, and that this is so irrespective of the good faith or bad faith of the prosecutor. It is one of the shortest and most cited rules in American criminal procedure.

Keep that page in mind, because it comes at the end of this lesson. What comes first is everything that happened to Brady before a jury ever heard his name: a charging decision, a first appearance, a bail determination, an indictment, and an arraignment. This lesson walks that road one step at a time, naming the standard that applies at each. As throughout the course, this is general education about the process, not legal advice about any particular case.

Step one: the charging decision, and how little law governs it

An arrest is not a charge. The decision to charge, what to charge, and how many counts to charge belongs to the prosecutor, and it is the least regulated decision in the system. A prosecutor who is satisfied that probable cause exists may still decline. A prosecutor who charges may stack counts that all arise from one act, which as the next lesson shows is what gives plea bargaining its leverage.

Courts have refused to supervise this in any meaningful way. Two doctrines nominally constrain it, and both are close to unwinnable.

• Selective prosecution. A defendant must show both discriminatory effect, meaning that similarly situated people of another class were not prosecuted, and discriminatory purpose. In United States v. Armstrong (1996) the Court held that even to obtain discovery on such a claim, a defendant must first produce credible evidence of the effect prong, which is normally in the government's exclusive possession. Note the shape of that: you need the evidence to get the evidence.
• Vindictive prosecution. Where a prosecutor increases charges to retaliate for the exercise of a legal right, a presumption of vindictiveness may apply, as in Blackledge v. Perry (1974) after a defendant sought a trial de novo. The presumption does not apply during ordinary plea negotiation, which is the subject of Bordenkircher in the next lesson.

Key idea: Almost every filter in the system, from bail to sentencing exposure, operates on charges the prosecutor selected, and that selection is barely reviewable.

Step two: first appearance, and the forty-eight hour clock

A person arrested without a warrant has been jailed on nothing but an officer's judgment. Gerstein v. Pugh (1975) held that the Fourth Amendment requires a prompt judicial determination of probable cause as a condition of any extended pretrial detention. It need not be an adversarial hearing; a judge reviewing an affidavit is enough.

County of Riverside v. McLaughlin (1991) put a number on prompt. A probable cause determination within forty-eight hours of arrest is generally presumed reasonable, and beyond that the burden shifts to the government to show a bona fide emergency or extraordinary circumstance. Weekends and the convenience of consolidating hearings do not count.

At the first appearance the defendant is told what he is charged with, told his rights, and, under Rothgery v. Gillespie County (2008), his Sixth Amendment right to counsel attaches, because this is the initiation of adversary judicial proceedings. The court also decides whether he goes home.

Step three: bail, and the shift from flight to dangerousness

The Eighth Amendment says that excessive bail shall not be required. It does not say bail must be granted. In Stack v. Boyle (1951) the Court held that bail set higher than an amount reasonably calculated to ensure the defendant's presence at trial is excessive, which framed bail as a device with one purpose: appearance.

That framing did not survive. The Bail Reform Act of 1984 authorised federal courts to detain a defendant before trial on a finding, by clear and convincing evidence, that no conditions would reasonably assure the safety of the community. In United States v. Salerno, 481 U.S. 739 (1987), the Court upheld it six to three, reasoning that the detention was regulatory rather than punitive, that the government's interest in preventing crime by arrestees is legitimate and compelling, and that the Act's procedural protections were adequate. Justice Marshall's dissent objected that the majority had approved imprisonment on a prediction of future conduct, before any adjudication of guilt.

Then there is money. In most state courts the practical question is not whether the judge orders detention but whether the defendant can post a bond, which turns on his resources rather than on his risk. That has driven the most active reform movement in criminal procedure. New Jersey largely replaced money bail with a risk-based system in 2017. Illinois eliminated cash bail outright under the Pretrial Fairness Act, which took effect in September 2023. Detailed statistics on jail populations and pretrial status are published by the Bureau of Justice Statistics, and the arguments on both sides now turn on those numbers rather than on doctrine.

Step four: grand jury, or preliminary hearing

The Fifth Amendment requires that no person be held to answer for a capital or otherwise infamous crime except on a grand jury indictment. That clause is one of the few in the Bill of Rights never incorporated against the states: Hurtado v. California (1884) held that due process does not require it. So roughly half the states proceed by information, a charging document filed by the prosecutor, tested at a preliminary hearing.

The two routes differ more than students expect.

	Federal grand jury	Preliminary hearing
Who decides	16 to 23 citizens; 12 must concur to indict	A judge or magistrate
Who is present	Prosecutor, witness, court reporter, jurors; no judge and no defence counsel in the room	Both sides, in open court
Standard	Probable cause	Probable cause
Rules of evidence	Largely inapplicable; hearsay alone may support an indictment (Costello, 1956)	Apply, though many jurisdictions relax them
Exculpatory evidence	No duty to present it (United States v. Williams, 1992)	Defence may introduce it and cross-examine
Secrecy	Proceedings are secret	Public

Put those cells together and you can see why the grand jury's screening function is largely nominal. The prosecutor picks the witnesses, states the law, faces no opponent, and needs only twelve of twenty-three votes on a probable cause standard. Sol Wachtler, then chief judge of New York, is widely quoted as remarking that a prosecutor could get a grand jury to indict a ham sandwich, and the line stuck because practitioners recognised it.

The grand jury has a second function that remains formidable: investigation. It can subpoena documents and compel testimony under oath, and a witness who refuses without a valid privilege may be jailed for contempt. Under United States v. Calandra (1974) a witness cannot even refuse on the ground that a question is based on unlawfully seized evidence.

The upshot: As a shield the grand jury is weak; as an investigative instrument it is one of the strongest tools the government has.

Step five: arraignment, motions, and the clock

On the indictment or information the defendant is arraigned: the charge is read, and he pleads guilty, not guilty, or in some jurisdictions no contest. A not guilty plea starts the pretrial phase, where suppression motions of the kind the last three lessons described are filed and heard.

Two speedy trial rules run in parallel. The constitutional right is measured by the four factors of Barker v. Wingo (1972): the length of the delay, the reason for it, whether and when the defendant asserted the right, and the prejudice he suffered. No factor is dispositive, and the only remedy for a violation is dismissal, which is why courts are reluctant to find one. Separately, the federal Speedy Trial Act of 1974 imposes concrete deadlines, generally thirty days from arrest to indictment and seventy days from indictment or first appearance to trial, subject to a long list of excludable periods.

Step six: disclosure, which is not what television suggests

Criminal discovery is far narrower than civil discovery. There are no depositions of witnesses as of right in most systems. Federal Rule of Criminal Procedure 16 gives the defence the defendant's own statements, his criminal record, documents and objects material to the defence, examination and test reports, and a summary of expert testimony, with reciprocal obligations running back to the government. The Jencks Act generally delays production of a witness's prior statements until after the witness testifies.

Against that thin background, Brady does most of the work. Four cases build it out:

• Giglio v. United States (1972): the duty covers impeachment evidence, including a promise of leniency to a cooperating witness, and the knowledge of one prosecutor in the office is attributed to the office.
• United States v. Bagley (1985): evidence is material when there is a reasonable probability that, had it been disclosed, the result would have been different. That is the same phrase Strickland uses, and it is a demanding one.
• Kyles v. Whitley (1995): materiality is assessed on the suppressed evidence considered cumulatively, not item by item, and the prosecutor has a duty to learn of favourable evidence known to others acting on the government's behalf, including police.
• United States v. Ruiz (2002): there is no obligation to disclose impeachment information before a guilty plea, which matters enormously given how most cases end.

The rule's weakness is remedial. A violation is judged after the fact by whether it probably changed the outcome, so a prosecutor who withholds evidence in a strong case risks little. Connick v. Thompson (2011) held that a district attorney's office could not be held liable under section 1983 for a single Brady violation on a failure-to-train theory, even where the defendant had spent years on death row. Several states have responded by legislating open-file discovery instead of relying on prosecutorial judgment about materiality; North Carolina adopted such a statute in 2004 after a series of exonerations exposed withheld statements.

Bottom line: Brady asks the party with the least incentive to disclose to decide what is material, and reviews the decision only afterwards, by a standard that forgives most of what it finds.

Common misconceptions

• Being arrested means being charged. The charging decision is separate, belongs to the prosecutor, and is barely reviewable.
• The Constitution guarantees bail. The Eighth Amendment forbids excessive bail; Salerno upheld outright pretrial detention on a dangerousness finding.
• Every serious case goes to a grand jury. The indictment clause was never incorporated, so many states proceed by information and preliminary hearing.
• A grand jury weighs both sides. There is no defence counsel in the room and no duty to present exculpatory evidence under Williams.
• Criminal defendants get full discovery. Criminal discovery is much narrower than civil, and there is generally no right to depose witnesses.
• Any withheld evidence overturns a conviction. Only material evidence does, judged cumulatively under Kyles by whether it probably changed the result.

The short version

• The charging decision is the system's least regulated step; selective and vindictive prosecution claims almost never succeed.
• Gerstein requires a prompt judicial probable cause determination, and McLaughlin generally sets the outer limit at forty-eight hours.
• The Sixth Amendment right to counsel attaches at the first appearance under Rothgery.
• Stack v. Boyle tied bail to appearance; Salerno approved detention for dangerousness, and money bail is now the main reform battleground.
• The grand jury requirement binds only the federal government after Hurtado; as a screen it is weak, as an investigative tool it is powerful.
• Speedy trial protection comes from the Barker factors and, federally, from the deadlines of the Speedy Trial Act.
• Brady, extended by Giglio, Bagley, and Kyles, requires disclosure of material favourable evidence, but Ruiz exempts pre-plea impeachment material and the materiality standard is forgiving.

Sources

1. Legal Information Institute. (n.d.). Grand jury. Cornell Law School. law.cornell.edu
2. Legal Information Institute. (n.d.). Bail. Cornell Law School. law.cornell.edu
3. Wikipedia. (n.d.). Brady v. Maryland. en.wikipedia.org
4. Wikipedia. (n.d.). United States v. Salerno. en.wikipedia.org
5. Congressional Research Service. (n.d.). Constitution Annotated: Fifth Amendment. constitution.congress.gov
6. Bureau of Justice Statistics. (n.d.). Bureau of Justice Statistics. United States Department of Justice. bjs.ojp.gov

A forged cheque for eighty-eight dollars and thirty cents

Paul Lewis Hayes was indicted in Kentucky for uttering a forged instrument. The instrument was a cheque for $88.30. The offence carried two to ten years.

The prosecutor made an offer: plead guilty and I will recommend five years. He also said, on the record, what would happen if Hayes refused. He would return to the grand jury and seek a second indictment under the Kentucky Habitual Criminal Act. Hayes had two prior felony convictions, and under that Act a third conviction carried a mandatory sentence of life imprisonment.

Hayes refused. The prosecutor did exactly what he had said he would do. A jury convicted Hayes on the forged cheque, the habitual criminal count was proved, and he was sentenced to life.

Stop here and work the problem before reading the answer. Two descriptions of that exchange are available. On one, a prosecutor with probable cause for a greater charge offered a discount for saving the state a trial, and a defendant who fully understood the terms chose the risk. On the other, a man received a life sentence for exercising a constitutional right, and the eighty-eight dollar cheque is the tell: nothing about his conduct changed between the offer and the sentence except that he said no.

Bordenkircher v. Hayes, 434 U.S. 357 (1978), adopted the first description, five to four. In the give and take of plea negotiation there is no element of punishment or retaliation, Justice Stewart wrote, so long as the accused is free to accept or reject the offer, and the prosecutor had probable cause to bring the greater charge all along. The dissent answered that the state was penalising the assertion of a right, and that calling it a negotiation did not change what it was.

Whatever you conclude, notice what the case establishes: the threat to charge more, made explicitly to induce a plea, is constitutional. That is the legal foundation of the machine this lesson describes. And a standing note, more pointed here than anywhere else in the course: this is general legal education, not legal advice, and no one facing a plea decision should make it on the strength of a lesson rather than a lawyer.

How much of the system this is

In Missouri v. Frye (2012), the Supreme Court set out the figures it was working from: ninety-seven per cent of federal convictions and ninety-four per cent of state convictions are the result of guilty pleas. In the companion case, Lafler v. Cooper, Justice Kennedy drew the conclusion in a sentence that has been quoted ever since, describing criminal justice today as for the most part a system of pleas, not a system of trials.

Sit with what that means for the rest of this course. The confrontation right, the reasonable doubt standard, the jury, the rules of evidence, the cross-examination of the state's witnesses: those govern the small remainder. For the overwhelming majority of defendants, the operative procedure is a conversation between two lawyers, and the operative safeguard is a fifteen minute colloquy in open court.

So what?: If you want to know what American criminal procedure actually does to people, study the plea, not the trial.

Where it came from, and when the Court blessed it

Plea bargaining is not ancient. Early nineteenth century American court records show trials as the ordinary disposition, and judges of that era were often openly hostile to guilty pleas induced by promises. It grew after the Civil War alongside professional police forces, professional prosecutors, rising urban caseloads, and increasingly complex trials. By 1931 the Wickersham Commission was documenting it as routine and largely invisible.

The Supreme Court did not squarely approve it until 1970. Brady v. United States held that a guilty plea is not involuntary merely because it was motivated by the desire to avoid a heavier sentence, including the death penalty. A year later, in Santobello v. New York (1971), Chief Justice Burger went further and called plea bargaining an essential component of the administration of justice, to be encouraged if properly administered. Santobello also supplied a real protection: when a prosecutor makes a promise that induces a plea, the promise must be kept, and a broken one entitles the defendant to relief.

The same term produced North Carolina v. Alford (1970). Henry Alford pleaded guilty to second-degree murder while telling the court he had not done it, in order to avoid a capital trial. The Court held the plea valid: a defendant may plead guilty while protesting innocence where the plea is voluntary and the record contains a strong factual basis. The Alford plea is the sharpest illustration of what the system is actually asking people to do.

What makes a plea valid

A guilty plea waives three constitutional rights at once, and Boykin v. Alabama (1969) requires the record to show affirmatively that the waiver was knowing and voluntary. The three are the privilege against compelled self-incrimination, the right to trial by jury, and the right to confront one's accusers.

In federal court that showing is made through the Rule 11 colloquy, in which the judge, addressing the defendant personally in open court, establishes that he understands the charge, the maximum penalty and any mandatory minimum, and the rights he is giving up, that the plea is not the product of force or of promises outside the agreement, and that there is a factual basis for it.

Two limits on that ritual are worth naming honestly. First, the colloquy asks whether anyone threatened the defendant, and the threat that actually produced most pleas, the prospect of a much longer sentence after trial, is not what the question means. Second, under United States v. Ruiz (2002), the government need not disclose impeachment evidence before the plea, so the defendant answers the judge's questions without knowing how weak the state's witnesses may be.

When counsel gets the negotiation wrong

If the plea is where cases are decided, then effective assistance at the plea stage matters more than effective assistance at trial. The Court took twenty-seven years after Strickland to say so clearly, and then did it in three cases.

• Hill v. Lockhart (1985) applied Strickland to pleas: the prejudice question is whether, but for counsel's errors, the defendant would have gone to trial.
• Padilla v. Kentucky (2010). Jose Padilla had been a lawful permanent resident for over forty years and had served in the United States armed forces in Vietnam. His lawyer told him he did not have to worry about immigration status. In fact the drug offence made deportation virtually mandatory. The Court held that counsel must advise a noncitizen client whether a plea carries a risk of deportation, rejecting the old distinction between direct and collateral consequences for this purpose.
• Missouri v. Frye (2012). The prosecutor sent two written offers, one of them a misdemeanour with ninety days. Frye's lawyer never told him they existed, and they expired. The Court held that counsel has a duty to communicate formal offers.
• Lafler v. Cooper (2012). Counsel advised rejecting an offer because he wrongly believed the state could not prove intent to murder where the victim had been shot below the waist. Cooper went to trial, was convicted, and received a sentence several times longer than the offer. The Court held that a defendant may show prejudice from advice that caused him to reject a plea, and approved a remedy of ordering the offer reoffered.

What matters here: A defendant now has a constitutional right to competent advice about an offer, but the right runs to the advice, not to the offer. The prosecutor is under no obligation to make one.

The trial penalty, and where the leverage comes from

The gap between the sentence offered for a plea and the sentence imposed after a conviction at trial is usually called the trial penalty. In 2018 the National Association of Criminal Defense Lawyers published a book-length study of it under that name, arguing that the differential has grown large enough that the right to trial is exercised only by defendants who can afford to gamble years of their life on it.

Where does that differential come from? Three sources, all of them lawful:

1. Charge stacking. One incident can support several counts, and each count adds exposure. Dropping counts costs the prosecutor nothing she was ever going to prove separately.
2. Mandatory minimums. A statute that fixes a floor moves the decision from the judge to the prosecutor, because the choice of charge now determines the sentence. In federal drug cases, filing an information about prior convictions under 21 U.S.C. section 851 raises the mandatory minimum, and the decision whether to file it belongs entirely to the prosecutor.
3. Acceptance of responsibility. Sentencing schemes reduce the sentence for pleading guilty. Read from the other end, they increase it for going to trial.

Defenders of the system reply that the discount is what a defendant is paying for, that the post-trial sentence is the real one, and that a system in which pleading guilty produced no benefit would produce no pleas and therefore no functioning courts. That reply is serious. Notice, though, that it concedes the structural point: the system depends on making trial expensive.

The innocence problem

An innocent defendant should refuse any plea, on the theory that he cannot be convicted. The theory assumes he is confident about the outcome, indifferent to risk, and free while he waits. Often none of that is true.

Consider a person held on a bond he cannot post, charged with a misdemeanour, offered a plea to time served. Pleading guilty means walking out today. Refusing means weeks in jail awaiting trial, which for many people means losing a job, a tenancy, or custody of a child. The rational choice for an innocent person in that position is frequently to plead guilty, and this is a well documented feature of lower criminal courts rather than a hypothetical. Malcolm Feeley's 1979 study of a Connecticut misdemeanour court made the point in its title: the process is the punishment.

The National Registry of Exonerations records exonerations of people who had pleaded guilty, concentrated in drug cases and in high volume misdemeanour courts. Every one of them went through a Rule 11 style colloquy and told a judge, under oath, that they were guilty and that nobody had pressured them.

The best case for the system, and the best reply

The strongest defence is contractual, argued most fully by Robert Scott and William Stuntz in 1992. A plea is an exchange from which both parties gain: the state saves the cost and risk of trial, the defendant buys certainty and a lower sentence. Forbidding the exchange would not give defendants trials they want; it would give them trials they would rather have traded away, and would leave them worse off by their own lights. On this view, restricting plea bargaining is paternalism dressed as protection.

The strongest reply, argued by Stephanos Bibas in 2004, accepts the contract frame and attacks its assumptions. A well functioning bargain requires parties who know the odds and can act on them. Real plea negotiation is distorted by information asymmetry, since the defence often has not seen the state's evidence and is not entitled to impeachment material; by agency costs, since defence counsel carrying two hundred cases has different incentives from the client; by resources, since retained counsel with time to investigate produces different offers; by pretrial detention, which imposes a cost on refusing that has nothing to do with the merits; and by ordinary risk aversion in the face of decades. Prices set under those conditions do not track the probability of conviction, so the bargain does not do the work the defence claims for it.

The core of it: The dispute is not about whether trades should be allowed. It is about whether the conditions under which these particular trades are made can be called consent.

Common misconceptions

• Plea bargaining is a modern shortcut that could simply be banned. It has been the ordinary disposition for over a century, and the Court has treated it as essential since 1971.
• Threatening a harsher charge to induce a plea is coercion. Bordenkircher holds it constitutional where probable cause exists for the greater charge.
• You cannot plead guilty if you say you are innocent. An Alford plea allows exactly that, given voluntariness and a strong factual basis.
• The judge's colloquy screens out coerced pleas. It screens for threats outside the bargain; the pressure that produces most pleas is the bargain itself.
• A defendant has a right to a plea offer. He has a right to competent advice about an offer that is made, not to receive one.
• Innocent people would never plead guilty. Detention, caseloads, risk, and time-served offers make it a rational choice, and documented exonerations follow guilty pleas.

Pulling it together

• Bordenkircher permits a prosecutor to threaten and then bring a much greater charge when a defendant refuses to plead.
• The Court's own figures put guilty pleas at ninety-seven per cent of federal and ninety-four per cent of state convictions.
• Brady v. United States and Santobello approved bargaining; Santobello also requires the prosecution to keep its promises.
• Boykin and Rule 11 require a record showing a knowing and voluntary waiver of three specific rights.
• Hill, Padilla, Frye, and Lafler apply the Sixth Amendment to the negotiation itself.
• The trial penalty is built from charge stacking, mandatory minimums, and sentencing credit for acceptance of responsibility.
• The contract defence of plea bargaining is strong on its own terms; the reply attacks the conditions under which the bargains are struck.

Sources

1. Legal Information Institute. (n.d.). Plea bargain. Cornell Law School. law.cornell.edu
2. Wikipedia. (n.d.). Bordenkircher v. Hayes. en.wikipedia.org
3. Wikipedia. (n.d.). North Carolina v. Alford. en.wikipedia.org
4. Bureau of Justice Statistics. (n.d.). Bureau of Justice Statistics. United States Department of Justice. bjs.ojp.gov
5. Scott, R. E., and Stuntz, W. J. (1992). Plea bargaining as contract. Yale Law Journal, 101(8).
6. Feeley, M. M. (1979). The Process Is the Punishment: Handling Cases in a Lower Criminal Court. Russell Sage Foundation.

Ten votes out of twelve

In 2016 a Louisiana jury found Evangelisto Ramos guilty of a serious crime by a vote of ten to two, and he was sentenced to life in prison without the possibility of parole. In forty-eight states and in federal court those two dissenting votes would have produced a hung jury. In Louisiana they produced a conviction.

Louisiana had permitted non-unanimous felony verdicts since its constitutional convention of 1898, a convention whose own stated purpose included establishing the supremacy of the white race in the state. Oregon adopted its rule in the 1930s in a period of open nativist agitation. When the question reached the Supreme Court in 1972 the Justices had split so oddly that the rule survived on a single concurrence, and it stood for another forty-eight years.

Ramos v. Louisiana, 590 U.S. 83 (2020), ended it. The Sixth Amendment right to a jury trial, as incorporated against the states, requires a unanimous verdict to convict of a serious offence. A year later Edwards v. Vannoy (2021) held that the ruling does not apply retroactively on federal collateral review, so people convicted by divided juries before 2020 whose appeals were finished did not get new trials.

That sequence sets up the whole lesson. Trial rights are specific, they are the product of contested history rather than of abstraction, and what happens after conviction is governed by a separate body of law that most courses treat as an afterthought. This is general legal education rather than legal advice, and it is describing the federal constitutional floor; every state adds to it.

The jury: when you get one, and who is on it

Duncan v. Louisiana (1968) incorporated the jury trial right against the states. It does not cover everything. Under Baldwin v. New York (1970) the right attaches to serious offences, meaning those carrying a potential sentence of more than six months, which leaves a very large volume of misdemeanour prosecutions to be tried by a judge alone.

Size and unanimity took fifty years to settle. Williams v. Florida (1970) approved six-member juries; Ballew v. Georgia (1978) held that five is too few; Ramos requires unanimity whatever the size.

Composition is governed by two separate doctrines that students routinely merge.

• The fair cross-section requirement applies to the venire, the pool from which jurors are drawn. Taylor v. Louisiana (1975) struck down a scheme that effectively excluded women. No defendant is entitled to a jury of any particular composition; he is entitled to a pool from which distinctive groups have not been systematically excluded.
• Batson applies to the strikes used on the panel. In Batson v. Kentucky (1986) the Court held that a prosecutor may not use peremptory challenges to remove jurors on account of race, and set out three steps: the objecting party makes a prima facie showing of discrimination; the striking party offers a race-neutral explanation; the judge decides whether purposeful discrimination has been proved. J.E.B. v. Alabama (1994) extended it to sex.

Batson's weakness is step two, because almost any explanation qualifies as facially neutral, and step three asks a trial judge to find that a lawyer in front of him lied. Flowers v. Mississippi (2019) shows what it takes to win: the same prosecutor tried Curtis Flowers six times for the same crime, and across those trials the pattern of strikes against Black jurors was extreme enough that the Court reversed.

Remember: The cross-section right governs the pool and asks about systematic exclusion; Batson governs the strikes and asks about purposeful discrimination in this case.

Confrontation, and the end of the reliability test

Michael Crawford stabbed a man who he said had tried to assault his wife. His wife Sylvia gave a tape-recorded statement to police that undercut his self-defence claim. She did not testify, because a marital privilege kept her off the stand, so the prosecution played the tape.

Under the test then in force, from Ohio v. Roberts (1980), out-of-court statements were admissible if the declarant was unavailable and the statement bore adequate indicia of reliability. The Washington courts found the tape reliable.

Crawford v. Washington, 541 U.S. 36 (2004), discarded that framework. Justice Scalia held that the Confrontation Clause is not a rule about reliability at all. It commands a procedure: testimonial hearsay is inadmissible unless the declarant is unavailable and the defendant had a prior opportunity to cross-examine. Reliability is what cross-examination is for, and a judge's finding of reliability is not a substitute for it.

The obvious follow-up question was what makes a statement testimonial, and the answer arrived in stages.

• Davis v. Washington (2006) set the primary purpose test. A 911 call made to summon help during an ongoing emergency is not testimonial; statements given to officers after the emergency has passed, to establish what happened, are.
• Melendez-Diaz v. Massachusetts (2009) held that a forensic laboratory certificate is testimonial, so the analyst must be available for cross-examination. Bullcoming v. New Mexico (2011) added that a surrogate analyst who did not perform the test will not do.
• Smith v. Arizona (2024) closed a workaround. When an expert testifies to an absent analyst's statements in support of his own opinion, and those statements support the opinion only if they are true, they are being offered for their truth and the Confrontation Clause applies.
• Giles v. California (2008) preserved forfeiture by wrongdoing: a defendant who makes a witness unavailable, intending that result, loses the objection.

Proof beyond a reasonable doubt, and who carries what

In re Winship (1970) held that due process requires proof beyond a reasonable doubt of every fact necessary to constitute the crime charged. Two consequences follow that are worth stating precisely.

First, the judge may not lighten that load with a presumption. Sandstrom v. Montana (1979) struck down an instruction that the law presumes a person intends the ordinary consequences of his voluntary acts, because a jury could read it as shifting the burden on intent. Permissive inferences the jury may draw are fine; mandatory presumptions are not.

Second, the rule attaches to elements, not to defences. Under Patterson v. New York (1977) a state may place the burden of proving an affirmative defence on the defendant, so long as it has not simply relabelled an element as a defence. That is why, as Module 4 showed, the allocation of the burden on self-defence or insanity varies so much across states.

One more trial rule belongs here. Under Griffin v. California (1965), neither the prosecutor nor the judge may invite the jury to draw an adverse inference from a defendant's decision not to testify.

What sentencing is for, and why the answers conflict

Conviction ends one inquiry and starts another with a different logic. Five purposes are conventionally offered, and it is worth seeing that they point in different directions on the same case.

Purpose	The question it asks	Where it points
Retribution	What does this offender deserve for what he did?	Backward-looking; proportional to blameworthiness and harm
General deterrence	What sentence will discourage others?	Can justify punishing one person more than he deserves
Specific deterrence	What will discourage this offender?	Turns on his responsiveness, not on the offence
Incapacitation	How long must he be confined to prevent further offences?	Turns on predicted future conduct, which is what habitual offender laws use
Rehabilitation	What will change his behaviour?	Argues for treatment and for indeterminate release decisions

Restorative approaches add a sixth question about repairing the harm to the victim and the community. Federal judges are directed by statute, at 18 U.S.C. section 3553(a), to consider several of these at once, which is a real instruction rather than a rhetorical one: a judge weighing deterrence and rehabilitation on the same facts is being asked to reconcile aims that do not reconcile.

From discretion to guidelines, and partway back

Until the 1970s most American sentencing was indeterminate. A judge picked a wide range, and a parole board decided the actual release date based on progress in prison. Judge Marvin Frankel's 1973 book Criminal Sentences: Law Without Order attacked the system for lawlessness, arguing that the same defendant could receive wildly different sentences depending on which judge he drew, and that this was not discretion but arbitrariness.

The reform coalition that followed was unusual, joining critics who wanted less disparity with critics who wanted more severity. The Sentencing Reform Act of 1984 created the United States Sentencing Commission, abolished federal parole, and produced guidelines effective in 1987 that computed a narrow range from the offence level and the defendant's criminal history. Mistretta v. United States (1989) rejected a challenge to the Commission's constitutionality.

Then the jury right caught up with the guidelines.

The Apprendi line

Charles Apprendi fired shots into the home of a Black family in New Jersey and pleaded guilty to a firearms count carrying five to ten years. At sentencing the judge found, by a preponderance of the evidence, that the shooting had been racially motivated, and applied a hate crime enhancement to impose twelve years.

Apprendi v. New Jersey, 530 U.S. 466 (2000), reversed. Other than the fact of a prior conviction, any fact that increases the penalty for a crime beyond the prescribed statutory maximum must be submitted to a jury and proved beyond a reasonable doubt. The principle then worked through the system case by case.

Case	Year	What it added
Almendarez-Torres v. United States	1998	The prior conviction exception, decided before Apprendi and left standing by it
Apprendi v. New Jersey	2000	Jury and reasonable doubt for facts raising the statutory maximum
Ring v. Arizona	2002	Aggravating factors that make a defendant death-eligible are for the jury
Blakely v. Washington	2004	The relevant maximum is the maximum on the jury's verdict alone, not the top of a guideline scheme
United States v. Booker	2005	The federal guidelines are advisory rather than mandatory, reviewed for reasonableness
Alleyne v. United States	2013	Facts that raise a mandatory minimum are elements and go to the jury too

Gall v. United States (2007) confirmed that appellate courts review sentences outside the advisory range for abuse of discretion rather than requiring extraordinary justification. The practical result is a hybrid: judges must calculate the guideline range and consider the statutory purposes, and may then depart, subject to deferential review.

In short: A fact that raises the ceiling or the floor of a sentence is an element in everything but name, and the jury must find it beyond a reasonable doubt.

What the Eighth Amendment forbids

The cruel and unusual punishments clause is interpreted, since Trop v. Dulles (1958), by reference to evolving standards of decency that mark the progress of a maturing society, measured largely by counting state legislative choices and jury practice.

It has done the most work in two places.

Capital punishment. Furman v. Georgia (1972) invalidated the death penalty as then administered, on the ground that it was imposed arbitrarily. Gregg v. Georgia (1976) upheld revised statutes that guided discretion through bifurcated proceedings and aggravating and mitigating factors. Since then the Court has drawn categorical lines: Coker v. Georgia (1977), no death penalty for the rape of an adult woman; Kennedy v. Louisiana (2008), none for the rape of a child where no death occurred; Atkins v. Virginia (2002), none for defendants with intellectual disability; Roper v. Simmons (2005), none for offenders under eighteen at the time of the crime.

Juvenile sentences. Graham v. Florida (2010) barred life without parole for juveniles in non-homicide cases. Miller v. Alabama (2012) barred mandatory life without parole for juveniles in any case, requiring individualised consideration of youth. Montgomery v. Louisiana (2016) made Miller retroactive, and Jones v. Mississippi (2021) held that a sentencer need not make a separate factual finding of permanent incorrigibility.

Outside those categories, proportionality review of prison terms is very weak. Solem v. Helm (1983) once struck down a life sentence for a series of minor property offences, but Harmelin v. Michigan (1991) upheld a mandatory life sentence for possessing a large quantity of cocaine, and Ewing v. California and Lockyer v. Andrade (both 2003) upheld sentences of twenty-five years to life under California's three strikes law for shoplifting offences that were the third strike. The operative standard is gross disproportionality, and it almost never bites.

The part of the sentence nobody reads out

The judgment states a term of custody and supervision. It does not state that the conviction may also bar the person from public housing, from many occupational licences, from serving on a jury, from possessing a firearm, and, in many states, from voting for some period; that it may trigger registration requirements; or that for a noncitizen it may make removal from the country effectively automatic, which is why Padilla in the last lesson mattered so much. These collateral consequences are imposed by statutes scattered across a code, and they often outlast the sentence itself.

The point: The sentence pronounced in court is a subset of the punishment imposed by law.

Common misconceptions

• Every criminal defendant gets a jury. The right attaches to offences carrying more than six months, leaving most misdemeanours to bench trials.
• A defendant is entitled to a jury reflecting the community's composition. The fair cross-section right governs the pool, not the panel that is finally seated.
• Batson ended discriminatory strikes. It requires proof of purposeful discrimination, and almost any facially neutral reason satisfies step two.
• Confrontation is about whether the statement is reliable. Crawford rejected exactly that approach in favour of a procedural rule about testimonial statements.
• The federal sentencing guidelines are binding. Booker made them advisory; judges must calculate and consider them, then may depart.
• The Eighth Amendment requires sentences to be proportionate. For terms of years the standard is gross disproportionality, and challenges nearly always fail.

Looking back

• Duncan incorporated the jury right for serious offences; Ramos requires unanimity, and Edwards v. Vannoy denied it retroactive effect on collateral review.
• Taylor governs the venire through the fair cross-section requirement; Batson governs peremptory strikes through its three-step procedure.
• Crawford replaced reliability with a procedural rule for testimonial hearsay, refined by the primary purpose test and the forensic analyst cases.
• Winship requires proof beyond a reasonable doubt of every element; Sandstrom forbids mandatory presumptions and Patterson permits burdens on affirmative defences.
• The five purposes of sentencing point in different directions, and statutes ask judges to weigh them together.
• The Apprendi line moved facts that raise a sentencing ceiling or floor to the jury and made the federal guidelines advisory.
• The Eighth Amendment draws categorical lines around capital and juvenile sentences and almost none around terms of years.

Sources

1. Legal Information Institute. (n.d.). Sixth Amendment. Cornell Law School. law.cornell.edu
2. Legal Information Institute. (n.d.). Eighth Amendment. Cornell Law School. law.cornell.edu
3. Wikipedia. (n.d.). Crawford v. Washington. en.wikipedia.org
4. Wikipedia. (n.d.). Apprendi v. New Jersey. en.wikipedia.org
5. Congressional Research Service. (n.d.). Constitution Annotated: Sixth Amendment. constitution.congress.gov
6. Frankel, M. E. (1973). Criminal Sentences: Law Without Order. Hill and Wang.