Concept
Calder's Four Categories
Ex Post Facto Laws
Calder's Four Categories
Justice Chase’s opinion in Calder v. Bull gives the Ex Post Facto Clauses their working definition. He named four kinds of law that fall within the words and intent of the prohibition:1
- A law that makes criminal, and punishes, an action that was innocent when done.
- A law that aggravates a crime, or makes it greater than it was when committed.
- A law that changes the punishment and inflicts a greater punishment than the law attached to the crime when committed.
- A law that alters the legal rules of evidence and receives less, or different, testimony than the law required at the time of the offense in order to convict the offender.
The Supreme Court still treats the list as the core of the doctrine. “Ex post facto law” was a term of art with an established meaning when the Constitution was framed, and Chase’s opinion expounded the legislative acts that implicate the Clause’s core concern.2
Two clauses carry the prohibition. Article I, § 9 forbids Congress to pass an ex post facto law, and Article I, § 10 forbids the states to pass one.3 Each limits a legislature that enacts a penal law with retrospective effect,4 so Chase’s list governs federal and state enactments alike.
The list also marks the line between ex post facto laws and retrospective laws generally. Every ex post facto law is retrospective, Chase observed, but not every retrospective law is ex post facto. A law that impairs vested rights may be unjust or oppressive without offending the Clause, and a statute of pardon that mitigates punishment is a proper retrospective law.5 The Court’s later statement of the limit is blunter: although the Latin phrase literally covers any law passed after the fact, the prohibition applies only to penal statutes that disadvantage the offender.6 A claimant who cannot identify such a statute has no ex post facto claim, though the section on related retroactivity limits shows other doctrines that may still apply.
The four categories sort into two groups. The first three restrict the substantive criminal law, which decides what counts as a crime and how heavily it is punished. The fourth reaches the law of evidence, so it restricts procedure.7 It belongs on the list because a legislature barred from redefining or repricing a crime could otherwise reach the same defendant by lowering what the prosecution must prove. Later cases extended the prohibition to some procedural changes beyond evidence rules, when the accused loses a substantial right;8 the section on procedural changes takes up those cases.
Each category receives its own subsection: the rule, the reason for it, a leading case, and the neighboring situations that fall outside it. The first category is the clearest violation and the one legislatures least often attempt.
Criminalizing Innocent Conduct
The first category forbids a legislature to make an act criminal after the act is done. Blackstone’s description, which Collins v. Youngblood treated as an authoritative statement of the term’s technical meaning, captures the rule: a law is ex post facto when, “after an action (indifferent in itself) is committed, the legislator then for the first time declares it to have been a crime, and inflicts a punishment upon the person who has committed it.”9
Three requirements follow from that description. The act must be complete before the law exists, it must have been lawful when done, and the law must punish it. Where all three are present, the defendant had no warning, because a prohibition not yet enacted cannot be read, and the legislature may have shaped the prohibition with the defendant already in view. Those are the interests the prohibition protects: it assures that legislative acts give fair warning of their effect, and it restrains arbitrary and potentially vindictive legislation.10
Legislatures seldom attempt the plain version. The clearest ex post facto law creates a new crime and applies it to conduct that was not criminal when committed, and legislation seldom if ever tries to do so.11 The doctrine does its work against two equivalents.
The first equivalent deletes an element of the offense. Elements draw the line between crime and lawful conduct, and a person who stays on the lawful side of that line has committed no crime. A legislature that moves the line after the act pulls lawful conduct across it. Omitting “knowingly” from a statutory crime, or “in the nighttime” from a burglary statute, and applying the change to earlier acts has exactly that effect.12 Someone who entered a dwelling by day to steal committed no burglary as the statute then defined it, and applying the amended statute to that entry would punish an act that was no crime when done.
The second equivalent abolishes a defense that was available when the act occurred. A justification or excuse is the law’s statement that conduct within the literal terms of an offense is permitted or not blameworthy. A statute that removed self-defense from battery, or lack of pregnancy from abortion, and applied to earlier acts would punish a person for doing what the law then allowed.13 In dicta in Hall, Justice Washington reasoned that if “unavoidable accident” was an affirmative defense to a charge of failing to deliver cargo when the owner forfeited it, a later law adding the requirement that the vessel or cargo actually be lost at sea would take away a defense available when he acted and would be an invalid ex post facto law.14
The defense point has a limit, and the limit follows from the category’s premise. The category assumes the act was lawful or excused when done, so it reaches only the withdrawal of defenses related to the essential elements of a crime or to matters a defendant might plead as justification or excuse.15 The Court drew that line in Collins, which overruled Kring v. Missouri, and the line sits at the last prong of the shorter formulation that traces to Beazell v. Ohio: a law that deprives the accused of a defense available when the act was committed.16 A lost procedural advantage that says nothing about whether the act was lawful falls outside the category, and the section on procedural changes takes up those losses.
The leading illustration is Cummings v. Missouri, and it shows that the category reaches a law that defines no crime at all. Missouri’s constitution made the enjoyment of certain rights depend on an expurgatory oath disclaiming specified past acts, and the Court held the requirement ex post facto.17 The deprivation was punishment, the Court said, and it was no less punishment because the oath offered a way out. Whole classes of people were unable to take it, and a right made to depend on an impossible condition is denied outright. Such a denial, “enforced for a past act, is nothing less than punishment imposed for that act.”18
The oath also shows how the categories overlap. Some of the acts it reached had been serious offenses carrying fine, imprisonment, or other heavy penalties, and the Court said a further penalty for those acts “impose[d] additional punishment to that prescribed when the act was committed,” which is the third category. It added that the clauses subvert the presumption of innocence and alter the rules of evidence, a point that anticipates the fourth.19 The same reasoning applies with more force to any act that had carried no penalty at all. Whether a disability counts as punishment is a threshold question taken up in the section on whether a law is punitive.
The category depends on the act being complete before the law exists. A legislature may reach conduct, a condition, or an omission that continues after the law takes effect: a statute that increases the penalty for a conspiracy, for example, may be applied to a conspiracy that began earlier but carried on past the effective date.20 Once the law is on the books, the person can stop, and the law then operates on continuing conduct and not on a completed act.
The second question does the real work. A defendant who acted lawfully under the earlier law, with its elements and defenses taken as they stood, is the person the first category protects. A procedural advantage lost without touching the offense’s definition raises a different problem, and later sections treat it separately.
Aggravating the Offense
Chase’s second category reaches “every law that aggravates a crime, or makes it greater than it was, when committed.”22 Read alone, the words suggest a legislature that makes an existing crime worse after the fact. The Supreme Court gave the category a wider reach in Stogner v. California, which held that a law enacted after a limitations period has expired violates the Clause when applied to revive a time-barred prosecution, because such a law falls within Chase’s second category.23
California’s courts had rejected the defendant’s ex post facto claim, relying on People v. Frazer, and the Supreme Court reversed.24 The Court’s objection was that California’s statute “retroactively withdraws a complete defense to prosecution after it has already attached,” and does so “in a manner that allows the State to withdraw this defense at will and with respect to individuals already identified.”25 That objection tracks the Clause’s purposes: a legislature able to withdraw a defense at will, and as to persons it has already identified, has the opening for arbitrary and potentially vindictive lawmaking that the Clause restrains.26
The second category fits because Chase’s concern ran to laws that enabled punishment where it was not otherwise available “in the ordinary course of law,” and the Court grouped California’s statute with them.27 After a limitations period runs, the state can no longer punish the conduct in the ordinary course; a revival statute opens a path to punishment that the law had closed. The conduct was criminal when committed, and a revival statute changes neither the offense nor its penalty, so the first and third categories do not fit. The Court had already identified this feature as the vice relevant to Chase’s purpose in Carmell v. Texas, a fourth-category case discussed below.28
Earlier decisions point the same way. Courts that upheld extensions of unexpired periods consistently distinguished periods that had already expired, often by saying that an extension is not ex post facto “provided,” “so long as,” “because,” or “if” the prior period has not expired. That manner of speaking suggests a presumption that reviving time-barred prosecutions is not allowed.29
The holding is narrow. A state remains free to extend time limits for future offenses and for prosecutions not yet time barred.30 While a period is still running, no complete defense has attached, so an extension withdraws nothing. The line is the date the original period expired.
A revival problem therefore turns on one comparison of dates: the day the original period ended against the day the new statute took effect. If the statute came first, the extension is permissible. If the period had already run, applying the statute violates the second category.
Increasing Punishment
Chase’s third category covers every law “that changes the punishment, and inflicts a greater punishment, than the law annexed to the crime, when committed.”32 The first two categories ask whether the defendant could be punished at all. Here the act was a crime from the start, and the question is whether the legislature has raised the price after the fact. Answering it requires comparing two statutory scales: the one attached to the offense when it was committed, and the one the prosecution now asks the court to apply.
Increased punishments appear in no explicit historical source, but the Court has never questioned their prohibition, apparently on the theory that “the enhancement of a crime, or penalty, seems to come within the same mischief as the creation of a crime or penalty.”33 The mischief has two parts. A person weighing an offense prices the risk by the penalty the law announces, and a heavier penalty adopted afterward punishes reliance on that announcement. And a legislature that knows who has been caught can raise the penalty for exactly that offense.34
The measuring stick is the standard of punishment the statute prescribes, not the sentence the judge imposed. In Lindsey v. Washington the defendants received fifteen years, a term the old statute also permitted, and the Court rejected the argument that this saved the new law. The Clause, it said, “looks to the standard of punishment prescribed by a statute, rather than to the sentence actually imposed,” and it forbids applying “any new punitive measure to a crime already consummated, to the detriment or material disadvantage of the wrongdoer.”35 Because the comparison is between the two scales, a defendant may object to an increase even though the sentence received falls within the old limits. A court that asked what the old law’s sentencing judge would have done would be guessing; comparing the scales asks only what each statute prescribed.
Dobbert v. Florida marks the boundary of that rule. Dobbert raised three ex post facto claims: that Florida changed the roles of judge and jury in imposing death sentences between his acts and his trial, that no valid death penalty statute was in effect when he acted, and that the new statute attached more stringent parole requirements to a life sentence.36 Only the third concerns the measure of punishment; the first two return in the section on procedural changes. The third claim rested on the new rule that anyone sentenced to life imprisonment must serve at least 25 years before parole eligibility, a limit the prior statute lacked.37 The Court held that Dobbert, having been sentenced to death, could not complain of burdens attached to a life sentence he did not receive. Both life imprisonment and death remained available under the new law.38
What separates Lindsey from Dobbert is the effect of the new law on the punishment the defendant faced. The new law in Lindsey prescribed a heavier punishment for the offense of which the defendants stood convicted; the parole rule in Dobbert attached only to an alternative sentence he did not receive. A challenger must therefore point to a change that bears on the punishment applied to the challenger, though the challenger need not prove that the sentence imposed exceeds what the old law would have produced.
Violations of the third category recur in a few familiar shapes:
- A higher range, as when a statute making grand theft punishable by one to five years is amended to two to ten years and applied to an earlier theft.39
- A mandatory death penalty substituted for a statute that allowed life imprisonment or death.40
- An added sanction, as when a penalty of a fine or imprisonment becomes a fine and imprisonment.41
Each shape removes an outcome the old scale allowed or adds a burden it did not carry. A fourth shape is less obvious: an amendment that lowers the ceiling but raises the floor. Under the standard-of-punishment test, such an amendment removes the lenient end of the old range, and that loss counts as a disadvantage even though the maximum fell.42
The federal Sentencing Guidelines build the same comparison into their text. A sentencing court must use the Guidelines Manual in effect on the date of sentencing unless doing so would violate the Ex Post Facto Clause, in which case it must use the Manual in effect on the date of the offense.43
The third step is where the harder cases arise, because a new scale does not always sit plainly above the old one. The comparison runs in one direction. A law that lessens punishment leaves the offender no worse off, so a change from the death penalty to life imprisonment may be applied retroactively, and statutes of pardon that mitigate punishment are not ex post facto.44 Whether an earlier offender actually receives the lighter penalty depends on the statute and any saving provision. A saving provision can keep the penalties of a repealed act in force for earlier offenses, and one court, relying on Bradley, upheld sentences imposed under the repealed act as saved.45
The same question governs changes that alter how a penalty operates rather than its stated measure: does the new measure leave the offender worse off? Rescinding good-time credits, for example, can lengthen the time served before release, and a court will look to that effect, not to the legislature’s intent, in deciding whether the Clauses are violated.46 The later sections on sentencing guidelines, good-time credits, parole timing, and recidivist enhancements apply the question in detail.
Lowering the Quantum of Proof
Chase’s fourth category covers every law “that alters the legal rules of evidence, and receives less, or different testimony, than the law required at the time of the commission of the offense, to convict the offender.”48 The first three categories govern what counts as a crime and how heavily it is punished. This one governs how much the state must show before it may convict.
Treating a lowered threshold of proof like a new crime follows from what a proof requirement does. A rule that the state may not convict without a certain amount or kind of evidence is part of the law’s answer to when guilt exists. A legislature that cuts the requirement after the offense lets the state convict on a showing the law of the day declared insufficient. The change also runs one way: the requirement restrains only the prosecution, so lowering it can help the state and never the accused.
The cases sort along that line. A retroactive statute that moves the prosecution’s burden from proof beyond a reasonable doubt to a preponderance of the evidence is ex post facto, while a statute that enlarges the class of persons competent to testify is not, because the new rule may be used for or against criminal defendants as a class.49 The second kind of law neither increases the defendant’s punishment nor lowers the type or amount of evidence needed to establish guilt.50
Hopt v. Utah shows the second kind. When the homicide occurred, a convicted felon could not have been called as a witness. The law changed before trial, a convicted felon took the stand, and his testimony implicated Hopt. Although the change plainly hurt the defendant, the Court found no violation, because it neither made an innocent act criminal, nor aggravated a crime already committed, nor increased punishment, nor changed the proof necessary to convict.51 In Thompson v. Missouri, a conviction resting solely on circumstantial evidence was reversed because certain evidence was inadmissible, and the law was changed before the retrial to make that evidence admissible. The Court held the change procedural and not violative.52 Later cases make clear that Calder’s language was not meant to prohibit applying new evidentiary rules in trials for earlier crimes.53
The harder cases involve a rule that looks evidentiary but works on the quantum of proof. It is sometimes difficult to tell whether a law that nominally changes only the rules of evidence has in fact changed what is needed to convict, much as a law removing a defense would.54 Carmell shows the difference. When some of the defendant’s offenses occurred, Texas law treated a victim’s testimony as inadequate to convict unless she was under 14 at the time of the assaults or had told another person of the assault within six months. Before trial Texas abolished the told-another requirement for victims under 18.55 The Court distinguished the competent-witness decisions and held that applying the amendment to earlier offenses violated the Clause.56 The amendment did not widen the circle of people who could speak. It lowered the showing the state needed: a victim’s account that the law at the time of the offense treated as insufficient to convict became enough, which is a conviction on less testimony than the law then required.57
The label a legislature chooses decides nothing. By calling a law procedural, a legislature does not immunize it from scrutiny under the Clause.58 The oath in Cummings illustrates the point from the other side. The Court said its clauses assumed the parties guilty and called on them to prove their innocence through an expurgatory oath, which subverted the presumption of innocence and altered the rules of evidence.59 A device that shifts the burden onto the accused shares the vice of one that lowers the state’s burden.
The third step does the real work, because the question is whether the required showing fell, and the fact that the change touches evidence is not enough. A defendant who loses the benefit of a corroboration rule stands where Carmell stood; a defendant who faces one more competent witness stands where Hopt stood. Procedural changes that leave the amount of proof untouched receive separate treatment later in the chapter.