An Overview of the Argument
Blackstone was almost certainly the source of the Double Jeopardy Clause language. The Supreme Court noted that a bar against double jeopardy “was carried into the jurisprudence of this Country through the medium of Blackstone.”1 Blackstone stated a centuries-old common law view of double jeopardy, which is (I argue) the most defensible conception today, more than two hundred years later. But the growing procedural and substantive complexity of modern law has made the application of the common law principles devilishly difficult. Blackstone did not have to deal with federal sentencing guidelines, for example, and the number of criminal offenses today is a thousand times larger than when English double jeopardy law was in its infancy. “At the end of the thirteenth century, apart from treason and three offenses which were fast falling into the category of misdemeanors, there were but six felonies.”2 The difficulties of discerning when two offenses are the “same offense” are markedly reduced when one has a universe of six or seven offenses, rather than one of 7,000.3
This chapter sketches the argument that the rest of the book makes in more detail. Ultimately, the argument requires understanding Black-stone’s conception of double jeopardy and how that conception applies to the radically different modern criminal procedure. First, I introduce three models of double jeopardy protection as a way of organizing the argument.
The Court has explicitly rejected the first of the three models, though commentators sometimes write wistfully as if the model were still viable. This model, attractive to those who fear legislative excess, finds in the Double Jeopardy Clause a substantive limitation on how the legislature can define crimes and fix punishments.
A double jeopardy protection can be primarily about punishment, limiting the number of punishments that the substantive law can authorize for particular wrongdoing. Indeed, as we see in chapter 2, Thomas Becket’s attempt to limit the king’s power to punish clerics already convicted in ecclesiastical courts was early evidence of a double jeopardy principle. But, as we also see, that limitation on the lawgiver’s power to define crimes and fix penalties was decisively rejected in the thirteenth century. The reasons used to reject it eight hundred years ago are still valid today.
In modern times, it is the legislature, not the monarch, that creates criminal blameworthiness. Model 1 thus views double jeopardy as a substantive limitation on the legislature. The legislature is free to create crimes and authorize punishments, under this model, as long as the result is not impermissible multiple punishment. Model 1 assumes that the Double Jeopardy Clause has a substantive component that tells us when punishments are impermissibly multiple. But what would that component be? One might argue that any time a defendant is punished “twice” for the same offense, he has suffered impermissible multiple punishment. Even that apparently obvious principle turns out to be controversial and complex. Legislatures often authorize both a fine and a jail term for a single criminal violation. If the judge imposes both penalties, has the defendant been punished in violation of the Double Jeopardy Clause?
Perhaps a Model 1 proponent would argue that a fine and jail term do violate double jeopardy. If so, what about a fine and community service? Or a fine and probation? Or probation and community service? It seems unlikely that the Double Jeopardy Clause would forbid the judge to impose probation and community service, yet there is no principled basis to distinguish those sanctions from a fine and a jail term. All four sanctions require the defendant to do something as a consequence of having been found guilty of a criminal violation.4 Suffering a judicially imposed consequence for a criminal violation, even community service, must be a punishment. If not, why would the legislature authorize it as a sanction for a criminal violation? Thus, the Model 1 proponent is forced to argue that the imposition of probation and community service is multiple punishment.
But our intuitions resist the argument that probation and community service for the same criminal violation is impermissible multiple punishment. The reason, I believe, has to do with long-standing allocation of institutional responsibility to design and impose appropriate penalties. We expect the legislature to define crimes and fix penalties. We expect judges to impose from a range of penalties those that are most appropriate. In accomplishing the task of fashioning the proper punishment, flexibility is a virtue. In some cases, community service is sufficient punishment. In others, a prison sentence is necessary. And in cases where the public purse is also harmed, both a fine and a prison sentence are necessary to achieve a proportional punishment.
To take an example that is not strictly about judicial discretion, President Ford decided that Richard Nixon had “suffered enough” by having to resign the presidency, a “sentence” that would be unavailable for the rest of us. Whether or not one agrees with President Ford’s pardon of Nixon, the point is that flexibility permits the sentencer leeway to construct a sentence that takes into account unusual or even unique characteristics of the offender. Any principle that says the legislature can authorize, and the judge can impose, only a single kind of penalty seems too arbitrary, too rigid.
Our intuitions here are manifested in sentencing theory. Different theoretical models of sentencing have been offered over the centuries, ranging from a pure retributive model to a pure rehabilitation model, including various combinations of retribution, deterrence, incapacitation, and rehabilitation. Retribution is probably the oldest sentencing theory; it can be found in the Old Testament (an eye for an eye) as well as in the Code of Hammurabi. Kant was perhaps its most passionate and best advocate. To show his dedication to retribution as the sole justification for punishment, Kant offered a hypothetical in which an island community was going to leave the island forever. The community has to decide what to do with convicted murderers in prison awaiting execution. One choice is to leave the murderers behind on the island. They can harm none other than themselves (Kant assumes they cannot escape). But he insists that justice requires that “the last murderer lying in the prison ought to be executed before the resolution was carried out. This ought to be done in order that every one may realize the desert of his deeds, and that blood guiltiness may not remain upon the people; for otherwise they might all be regarded as participators in the murder as a public violation of justice.”5
Deterrence is probably a more modern, and certainly a more utilitarian, justification for punishment. The theory is to set punishments at a level that will deter the individual offender from repeating his offense, and the general population from committing it for the first time. These goals are called, respectively, special and general deterrence. Incapacitation is related to deterrence, except that it focuses only on the dangerousness of the individual offender. The more likely he is to recividate, the longer the sentence should be.
Rehabilitation is a theory currently out of favor, in part because it permits the most flexibility on the part of the judge, and the public has become dissatisfied with excessive judicial discretion in sentencing. Determining what sanction will have the best chance of rehabilitating an offender is a wholly individualized determination. Probation might be the best rehabilitative tonic for a first-time offender, but a repeat offender might need a fine and a prison term.
With a single possible exception, all of these theories and combinations of theories require at least some flexibility in sentencing. Retribution might seem to offer a “one size fits all” approach—for each crime there might be a Kantian punishment that is uniquely appropriate—but this would work only if the substantive criminal law were narrowly tailored to reflect Kantian harms. Modern United States criminal law has broad offense definitions and expansive notion of accomplice liability, thus requiring individualized sentences even under the retributive model. For example, a youth who drives the getaway car for what he thinks is a burglary may be guilty of murder if his partners kill a janitor in the building. But it is difficult to argue that he is as blameworthy as the killers. Indeed, the Supreme Court has focused on lesser blameworthiness to derive limitations on accomplice liability from the Eighth Amendment’s ban on cruel and unusual punishment.6
Incapacitation, special deterrence, and rehabilitation all focus on the individual offender and thus, by definition, require flexibility in sentencing. The only theory that does not require attention to individualized sentencing is the general deterrence goal of discouraging those who have yet to commit crime. Presumably, the question of optimal general deterrence is unrelated to the characteristics of the particular offender; after all, general deterrence failed by definition in the case before the court. Thus, a single sentence for murder is presumably justified by general deterrence goals, without regard to the failure of that sentence to satisfy retributive, incapacitative, or special deterrence goals in the case of our young getaway driver. The general deterrence message is “Don’t drive a getaway car for a burglary because you might wind up guilty of murder.” Thus, a system based wholly on general deterrent goals would likely need only a single sentence for each crime.
But general deterrence is an unsatisfying justification if used in isolation. General deterrence requires penalties to be arranged according to the temptation to commit crimes. Because few of us are tempted to commit murder, and many of us are tempted to take recreational drugs or cheat on our income taxes, a sentencing system built solely on general deterrence would have higher penalties for drug possession and income tax evasion than for murder. We reject this system because our ideal of justice is driven in large part by retribution and, to a lesser extent, by incapacitation and special deterrence. Murder will always carry the greatest penalty, despite its relative rarity, because it is the most harmful to society—so harmful that Kant wanted to kill the convicted murderers rather than strand them on the island.
It is quite likely, therefore, that general deterrence adherents would accept at least some flexibility in sentencing to take account of individual distinctions in blameworthiness and the need for incapacitation. As long as retribution, incapacitation, and special deterrence play a role in our instincts about sentencing, legislatures should be encouraged to provide alternative forms of sentences (fines, jail, probation, community service), and judges should be encouraged (with or without formal sentencing guidelines) to concoct the appropriate sentence by using one, some, or all of the alternatives. To find a countervailing, absolute ban on this kind of sentencing flexibility is to read far too much into the Double Jeopardy Clause.
Indeed, the Supreme Court has never held that imposition of a legislatively authorized fine and prison term for the same offense violates the multiple punishment component of the Double Jeopardy Clause. But if this clear case of double punishment is not impermissible “multiple punishment,” why would two consecutive prison terms be “multiple punishment” if they are clearly authorized by the legislature? The answer, the Court provided in Missouri v. Hunter,7 is that no combination of clearly authorized penalties constitutes multiple punishment under the Double Jeopardy Clause. Thus, Missouri could constitutionally authorize a second conviction and consecutive sentence for using a weapon to commit robbery (one conviction and sentence for robbery; the second conviction and consecutive sentence for using a weapon to commit the robbery).
Hunter, decided over only two dissents, is a clear refutation of a Model 1 substantive limitation on the power of the legislature to decree more than one conviction for the “same offense” in a single trial. If the legislature speaks clearly enough, two convictions in one trial (or three or four, for that matter) are not “multiple punishment” for purposes of the Double Jeopardy Clause. Oddly enough, after rejecting Model 1, the Court seems poised to accept a variant of this model based on little more than a jurisprudential reflex. The Justices who support a more expansive view of double jeopardy seem willing to find a limitation on the legislature in an amalgam of loose language in earlier cases, and the conservatives are attracted by what appears to be a literal reading of the constitutional text. But, as I show later, this Model 2 is flawed for the very same reason the Court rejected Model 1.
My view is that Hunter is the only defensible framework for a double jeopardy/multiple punishment framework. I argue in the next two sections for an expansion of the Hunter analysis to include successive prosecution theory and the civil/criminal distinction.
If the Double Jeopardy Clause does not forbid the legislature the power to decree the number of penalties, or convictions, that can attend the same offense in a single trial, the next question is whether it limits the power of the legislature in the context of successive prosecutions. At first glance (and, I argue, on deeper analysis as well), it would seem that the power to “define crimes and fix penalties” is the power to have as many substantive criminal offenses as the legislature wants for particular conduct. After all, by what standard could a court judge whether the legislature had divided conduct too finely, created too many offenses out of one “indivisible” course of conduct? What would an “indivisible” course of conduct look like? As then-Justice Rehnquist noted in rejecting this notion: “To the extent that this … thesis assumes that any particular criminal transaction is made up of a determinable number of constitutional atoms that the legislature cannot further subdivide into separate offenses, ‘it demands more of the Double Jeopardy Clause than it is capable of supplying.’”8
Once we know how many substantive offenses have been committed by the defendant’s conduct—and this question Hunter answers by deferring to legislative intent—any other limit on the prosecutor must be a limit on obtaining authorized penalties in more than one trial. One such limit is the joinder law of the jurisdiction. Joinder law, which is nonconstitutional in nature, is a procedural mechanism that sometimes requires trying in one trial offenses arising out of the same conduct. Model 2 essentially holds that the Double Jeopardy Clause overrides the legislature’s ability to define different substantive offenses. Stated differently, Model 2 finds in the Double Jeopardy Clause an additional, constitutionally based form of mandatory joinder.
There are two sets of arguments against the proposition that double jeopardy sometimes bars separate trials for legislatively separate offenses. Ironically, the arguments are held by Justices on opposite ends of the judicial philosophy spectrum (“liberal” to “conservative”). On the “liberal” end of the spectrum, we hear about the policy of not harassing criminal defendants by repeated prosecutions for the same conduct. This, to be sure, is the argument underlying nonconstitutional joinder provisions that sometimes require trying together offenses that arise from the same conduct. But the double jeopardy issue is whether that sound policy is embedded in the Double Jeopardy Clause. As chapter 2 shows, this policy argument is unpersuasive if we have a good account of when the legislative has truly authorized separate penalties. The antiharassment policy is also an untenable construction of the language of the Double Jeopardy Clause, which, after all, forbids more than one jeopardy for the same “offense,” not for the “same conduct” or “same transaction” or “same episode.”
Ironically, the reliance on constitutional language to avoid the liberal antiharassment argument leads the conservative side of the Court also to miss the essence of double jeopardy. Focusing on the language about “twice in jeopardy,” Justice Scalia argues that the Double Jeopardy Clause is wholly inapplicable in the single trial/multiple punishment context. While this is a permissible reading of the language, it is not the best reading of the history of double jeopardy, again as we see in chapter 2. More important, Scalia’s focus on carving the multiple punishment doctrine out of double jeopardy has led him to accept blindly the proposition that the clause must limit the number of trials the legislature can authorize. There is no reason in logic or history or language that the latter proposition follows from Scalia’s multiple punishment argument.
The Court has never held that the Double Jeopardy Clause forbids a prosecutor to seek two trials in the Hunter situation—when the legislature clearly meant to create different substantive offenses that would be the same offense under the Court’s mechanical same-offense test. Moreover, Scalia’s endorsement of that proposition is limited to a single footnote.9 Thus, the Court still has time to realize that the reason to reject Model 1 also applies to Model 2. The Double Jeopardy Clause has no substantive component that somehow denies to the legislature its prerogative to have as many substantive offenses, and trials, out of the same conduct as it wants. Policies about harassing prosecutions, while sound enough, do not transform the Double Jeopardy Clause into a rule of mandatory joinder.
If double jeopardy is not a substantive limitation on the legislature when it defines crimes and creates penalties in the single-trial context (Hunter), symmetry and logic dictate that double jeopardy is never a limit on the legislature. Of course, one can reject symmetry and logic and hold that the Framers meant to fasten double jeopardy on the legislature to prevent it from acting in derogation of the double jeopardy principle when “true” double jeopardy questions arise—that is, when one jeopardy has ended and another is threatened. This is Scalia’s view. Unfortunately, as the rest of this book details, there is no acceptable way to define either “offense” or “jeopardy” without ultimately relying on what the legislature intended. Our culture, legal and otherwise, gives us a rough idea of when a punishment is “cruel and unusual.” Culture gives us a benchmark of “unreasonable searches and seizures.” It tells us roughly when a witness has been “compelled to be a witness against himself.” It tells us precisely nothing about when offenses are the same and when a jeopardy has become “twice.” To arrive at a definition, we need the legislature. But if we need the legislature to flesh out our understanding of double jeopardy, we cannot avoid deferring entirely to the legislature when it speaks clearly to double jeopardy issues. That, at least, is the thesis of my book.
Consistent with Blackstone’s conception of double jeopardy, discussed in the next section, Model 3 limits the role of the Double Jeopardy Clause to ensuring the integrity of the legislative prerogative about “offense” and “jeopardy.” In effect, this model teaches that double jeopardy is a procedural limitation on prosecutors and judges, but not a substantive limitation on the legislature. This model takes seriously what the Court said in 1977 in Brown v. Ohio: “The legislature remains free under the Double Jeopardy Clause to define crimes and fix punishments.”10
Emphasizing that double jeopardy does not have a substantive component makes clear that it functions as a crude form of sentencing control (or guideline). Like modern statutory sentencing systems, double jeopardy provides legislative limits on how judges can sentence defendants. Double jeopardy provides a crude, either/or limitation—either the judge can sentence both for auto theft and joyriding, or the judge cannot. Despite its lack of fine gradations, which we expect in modern sentencing schemes, the Model 3 view of double jeopardy is remarkably similar to sentencing theory.
Like sentencing theory, double jeopardy applies uniformly whether penalties for violating different substantive offenses are sought in one trial or in two. Commentators and judges have perceived an independent concern with successive prosecutions11 because of the obvious (and trivial) limitation on prosecutors that follows from the multiple punishment limitation. If a judge cannot impose two convictions for offense A and offense B, prosecutors cannot seek two trials for A and B. While this is true, it tells us nothing more important than that prosecutors cannot seek any penalty if the legislature did not intend the defendant’s conduct to merit a penalty.
Consider statutory rape, defined as sexual intercourse with a person age 14–16 by someone who is at least two years older than the victim. A day before her sixteenth birthday, Allison has sex with Andy on his fourteenth birthday. Whatever our view of the morality involved, the legislature has spoken: Allison is not guilty of statutory rape, and the prosecutor may not seek her conviction. If the prosecutor does seek to convict Allison, the judge has no choice but to dismiss the indictment because Allison did not commit the substantive offense of statutory rape.
Now assume the legislature defines offense A so that it is an aggravated form of B, intending that actors who violate both A and B should be convicted of either offense but not both. If the prosecutor charges both in a single trial, and the jury convicts, the judge can impose but one conviction. This is the crucial multiple punishment limitation that derives from the Double Jeopardy Clause. It also follows that, once a conviction for either A or B exists, the prosecutor cannot prosecute the other, just as the prosecutor cannot prosecute Allison in the previous example—neither Allison nor our A-B offender has committed a crime for which the legislature intends prosecution.
This way of viewing the successive prosecution doctrine makes it derivative of the multiple punishment protection, precisely the opposite view from that of Justice Scalia. The Supreme Court seems to have recognized the derivative nature of successive prosecution theory in Brown, when it said:
If two offenses are the same under this [same-offense] test for purposes of barring consecutive sentences at a single trial, they necessarily will be the same for purposes of barring successive prosecutions. Where the judge is forbidden to impose cumulative punishment for two crimes at the end of a single proceeding, the prosecutor is forbidden to strive for the same result in successive proceedings.12
To be sure, the Court confused matters by saying, earlier in the opinion, “Where successive prosecutions are at stake, the [double jeopardy] guarantee serves ‘a constitutional policy of finality for the defendant’s benefit.’”13 I address the “constititutional policy of finality” in the next chapter. For now, suffice to say that the defendant who committed A-B has finality expectations no different from Allison. Allison can count on not being prosecuted for statutory rape because she did not commit statutory rape. Our A-B defendant can count on not being prosecuted for both A and B because he has not committed an act that makes him prosecutable for both crimes.
The task of defending Model 3 begins in the next section, where we meet the strongest contrary argument: that the Double Jeopardy Clause should be a substantive limitation on the legislature, like the Fourth Amendment, the Sixth Amendment, and the rest of the Fifth Amendment. The legislature cannot authorize unreasonable searches and seizures, or put limits on the constitutional right to counsel or confrontation, or permit defendants to be compelled to testify against themselves. Why, then, is the legislature the sole source of the substance of the Double Jeopardy Clause? Putting the question in less friendly terms, as some have done, why is the legislature free to act without the constraint of the Double Jeopardy Clause?
I can imagine a skeptic who would say my legislative-prerogative thesis puts too much power in legislatures who pander to the crime fear in the populace.14 Alternatively, the skeptic might deny the positivism that is at the heart of my thesis, opting instead for some natural law understanding of double jeopardy.15 The skeptic would say that to defer completely to the legislature somehow denies the Double Jeopardy Clause the independence befitting a constitutional right. The skeptic might concede that the legislature controls what counts as a completed jeopardy. The end of jeopardy has always been understood as a verdict and, in this century, certain kinds of mistrials. Moreover, the legislature has little lattitude in defining verdicts because of the overarching right to a jury trial and the right to be found guilty beyond a reasonable doubt.
But the skeptic would draw the line at defining “offense.” She would insist that the power to define double jeopardy offense belongs to judges. Judges are best situated to uncover “the consciousness and spirit of a community which, in turn, is embedded in its customs and practices.”16 These “customs and practices” give shape to the definitional inquiry. Judges are charged with defining “search,” “seizure,” “cruel and unusual punishment,” and “coercion,” to name a few difficult constitutional concepts. Why not “offense”?
The double jeopardy argument set out in this book does not deny judges their interpretative function. Indeed, the account offered here is the only one that provides judges an overarching, hard-edged principle to use in discharging that function. But it limits the range of the interpretive function to determining how the legislature meant to assign criminal blameworthiness.
The defense of this limit on judicial interpretation is partly linguistic: “offense” has no referent other than the legislative enactment. Concepts like “coercion” and “cruel” have cultural referents that exist independent of the legislature. A law seeking to define compulsion, for example, could not supplant judicial interpretations of what it means to be compelled to be a witness against oneself. “Offense” is different; there is no referent beyond what the offense-creating institution has done. We might say that it is immoral to lie or cheat. But we cannot say that it is an “offense,” in the double jeopardy sense or the criminal law sense,17 until the appropriate institution has made it an offense.
The linguistic argument intertwines with the historical and functional argument. With rare exception, the double jeopardy principle has always operated as a limitation on prosecutors and judges but not on the lawgiver; it is only the legislature (the lawgiver) that is institutionally competent to create legal blameworthiness. My argument crucially depends on the premise that the purpose of the Double Jeopardy Clause is to limit the number of blameworthiness determinations. Both Martin Friedland and Jay Sigler relegate prevention of multiple convictions to a secondary status because both believe that the principal function of double jeopardy is to prevent “the unwarranted harassment of the accused by multiple prosecutions.”18 That premise might cause one to reject the legislature as the ultimate source of double jeopardy limitations because one might naturally view the legislature as just another source of potential harassment.
Chapter 2 considers, and rejects, harassment as a double jeopardy policy that exists independent of the policy limiting the number of judgments to those authorized by the legislature. If I am right to reject harassment as an independent double jeopardy policy, no institution other than the legislature can give meaning to jeopardy and to same offense. To be sure, common-law judges were principally responsible for these tasks in the pre1792 world. But common-law judges performed many functions that we now assign to the legislature. More significant, even in the world where judges gave meaning to jeopardy and same offense, an occasional statute usurped the judicial role by permitting second trials or punishments in particular categories of cases,19 and none of these statutes were overruled by courts or criticized by any of the commentators, including the oft-critical Blackstone.
Using legislative intent to determine the right number of penalties is easy to defend when the issue arises in a single trial. The question of how many penalties the particular conduct deserves is otherwise beyond the capacity of judges to answer. Begin with the related question of the quantity of punishment a particular defendant or offense deserves (without regard to the appropriate number of penalties). In the Eighth Amendment cases involving a single penalty, the Court has been hard pressed to answer the desert question. This is the reason the Court has essentially limited the Eighth Amendment proportionality analysis to cases involving the death penalty.20 It is possible, though not easy, to develop a metaphysical answer to the question of whether a particular crime deserves the death penalty.21 It is almost impossible to defend a desert position on penalties less than death. Does possession of 650 grams of cocaine deserve forty years in prison? No, I do not believe so (even though the Court has approved life in prison without parole for this offense).22 Can I demonstrate that forty years is cruel and unusual punishment? Probably not. Even if I could, what would the answer be to thirty-nine years? Or thirty-eight years? And so forth.
Shifting to multiple penalties, the desert question is no easier to answer. Consider entry into a federal bank with intent to rob it, and aggravated bank robbery (robbery with a dangerous weapon). The first offense is punishable by fifteen years, the second by twenty-five years. If the judge imposes consecutive sentences totaling forty years, does this defendant have a valid claim of multiple punishment? Assuming forty years for aggravated bank robbery is not an excessive single penalty, it would be an excessive multiple penalty only if Congress did not intend both penalties to apply.23 From a functional perspective, therefore, the multiple punishment analysis begins with legislative intent.
Legislative intent is the right analytic approach to multiple punishment, both for the pragmatic reason that no other analytic structure exists and because it is the best functional structure. Though accounts of general moral or ethical blameworthiness can be developed and defended, the only blameworthiness that concerns the criminal justice system is wholly contained within the criminal prohibition. The legislature is the author of that prohibition and thus of that blameworthiness.
To see that this is so, consider what a metaphysical blameworthiness account would say about mala prohibita offenses—those that are thought to prohibit morally neutral but potentially harmful acts, such as driving without a license. There are dozens, if not hundreds, of criminal statutes that proscribe conduct that would be considered morally neutral on mainstream accounts of morality—engaging in homosexual conduct, sale of obscenity to adults, public drunkenness, vagrancy, gambling, and (perhaps) prostitution. If conduct is morally neutral, the statutory criminal blameworthiness in these offenses must necessarily be found in what Michael Moore calls the “bad state of affairs” that the legislature intended to prevent.24 Thus, gambling is a crime because it may harm the family of the gambler by depriving them of necessary money. Public drunkenness is a crime because it upsets the sensibilities of upstanding citizens. But if morally neutral criminal blameworthiness is found not in the conduct itself but in a social problem the legislature seeks to address, on what theory is traditional criminal blameworthiness (e.g., murder, robbery, theft) found in the conduct rather than in the legislative description and prohibition of that conduct?
Driving without a license is a blameworthy, forbidden act because the legislature wants to promote highway safety, and prohibiting this morally neutral conduct is an overinclusive way of accomplishing that goal. Drunk driving is a blameworthy act for the same reason and irrespective of the moral status of driving drunk (which is surely different from driving without a license). The argument seems irresistible that criminal blameworthiness, as opposed to moral fault, is exclusively the function of legislative creation. Thus, the criminal blameworthiness of engaging in homosexual conduct, for example, is constructed in the same manner as the criminal blameworthiness of murder—by consulting the statutory language forbidding the conduct, and without regard to metaphysical notions of wrongful or harmful acts.
This is but another way to say that legislatures today do what judges did in Blackstone’s day. In 1769 most criminal offenses were the product of judicial decision. Common-law judges created and refined the criminal “code” over the centuries, carving out subcategories of offenses as necessary—creating manslaughter as a crime distinct from murder, for example, to distinguish its lesser blameworthiness.25 Judges eventually relinquished to the legislature the role of defining crimes. With the role of defining crimes comes the role of defining jeopardy offenses. It is that simple.
To claim that all double jeopardy theory ultimately collapses into the “black hole” of legislative intent is not to suggest that this intent is easy to uncover. Nor do I have (or need) a new method of ascertaining the relevant intent. Indeed, part of the argument here is that the legislature so rarely makes its intent known on double jeopardy issues that courts must resort to presumptions to decide these questions. Chapters 6 and 8 develop and defend a set of presumptions about when offenses are the same and when jeopardy ends. But, if the legislature makes its intent clear, this intent overcomes any presumptions offered in this book or previously created by courts. Requiring the legislature to speak clearly to rebut double jeopardy presumptions solves the problem of how to ascertain legislative intent that is ambiguous, vague, or nonexistent. The practical value of this approach has been recognized in the Australian Crimes Act, which “requires that the legislature specifically intend double punishment” for the same act before it can be imposed.26
The theory offered here is simply that the legislature creates substantive blameworthiness and thus inevitably controls how it is to be parceled out (same offense) and when one determination procedurally stands for that blameworthiness (the end of jeopardy). Once the confusion about double jeopardy having a substantive component is cleared up, the analysis can profitably focus on how best to determine legislative intent on “same offense” and on “twice in jeopardy.”
I detail the specific doctrinal evidence in favor of the legislative-prerogative thesis in chapter 3. Here I offer more general doctrinal support, first drawing from areas outside double jeopardy. Then I use two somewhat unusual double jeopardy cases to show that my Model 3 account has more explanatory power than the Model 2 view, which limits successive prosecutions even when the legislature meant to create separately punishable offenses.
Outside the double jeopardy context, the Court has recognized that the legislature operates with only minimal limitations on its power to define crimes and fix penalties; these limitations are embedded in due process and in the Eighth Amendment prohibition of cruel and unusual punishment. For example, the issue in Patterson v. New York27 was whether in a murder prosecution the state had to prove the absence of “extreme emotional disturbance,” a general affirmative defense created by a different code section. Patterson argued that homicide consisted of the elements of homicide plus the proof that he had not acted under extreme emotional disturbance. To permit the legislature to characterize extreme emotional disturbance as a defense, rather than as a (negation of) one element of the crime itself, Patterson argued, had the effect of undermining the due process right to be proven guilty of the offense beyond a reasonable doubt. The Court responded that Patterson’s “offense” was what the legislature defined in the homicide code section and did not include the negation of affirmative defenses defined elsewhere in the criminal code. The key to Patterson is that the legislature retains the prerogative to decide what constitutes a criminal offense and what constitutes a defense to that crime.
Even closer to the double jeopardy context, the issue in Schad v. Arizona28 was whether a guilty verdict could stand if the state presented two different theories of liability, one requiring direct proof of mens rea (premeditated murder) while the other presumed mens rea once the actus reus was proven (felony murder). It was possible, Schad argued, that six jury members thought he committed the underlying felony but lacked premeditation and six others thought he premeditated but did not commit the underlying felony. In that situation, half the jury would have acquitted of felony murder and half would have acquitted of premeditated murder. How, then, Schad argued, could his verdict be constitutional?
As the Court recognized, Schad was not really arguing about jury unanimity or proof beyond a reasonable doubt. The jury unanimously found Schad guilty of first-degree murder beyond a reasonable doubt because Arizona defined first-degree murder in the alternative (premeditated or felony murder). Thus,
petitioner’s real challenge is to Arizona’s characterization of first-degree murder as a single crime as to which a verdict need not be limited to any one statutory alternative, as against which petitioner argues that premeditated murder and felony murder are separate crimes as to which the jury must return separate verdicts. The issue in this case, then, is one of the permissible limits in defining criminal conduct.29
The limits to the legislature’s prerogative to define criminal conduct is to be found in a “distillate of the concept of due process with its demands for fundamental fairness and for the rationality that is an essential component of that fairness.”30 In determining whether a particular statute satisfies this due process “distillate,” the Court began “with a threshold presumption of legislative competence to determine the appropriate relationship between means and ends in defining the elements of a crime.”31
Why this heavy dose of deference? The answer is allocation of institutional responsibilities and competence. The difficulty is in trying to determine whether there are “really” two offenses when the legislature said it was creating only one offense. As the Court put it: “Judicial restraint necessarily follows from a recognition of the impossibility of determining, as an a priori matter, whether a given combination of facts is consistent with there being only one offense.”32 The same difficulty infects the double jeopardy same offense analysis—how can a court know “as an a priori matter” whether what a defendant did is “really” only one crime even though the legislature has defined it as two? Schad suggests that, for due process purposes, there is a presumption that the legislature has made the right choice. My argument in this book is that, for double jeopardy purposes, the presumption is conclusive as long as the legislature expresses itself clearly.
If the legislature satisfies due process “fundamental fairness” and “rationality,” Schad holds that it may define one crime as it pleases. The legislative deference rationale of Schad suggests the prerogative to define two (or more) crimes to permit successive prosecution and cumulative punishment, again as long as the legislature satisfies the Eighth Amendment and the Schad standard of “fundamental fairness” and “rationality.” Because those standards are found outside the Double Jeopardy Clause, however, we can state the principle derived from Schad as follows: the Double Jeopardy Clause contains no substantive limitation on the legislative prerogative. If any given combination of statutes satisfies the Eighth Amendment and due process, Model 3 double jeopardy requires only that courts have good reason to believe that the legislature meant both statutes to apply to the defendant’s conduct. In that event, both penalties can be applied consistent with the Constitution.
Two examples from the Court’s double jeopardy jurisprudence show the explanatory power of Model 3. Oddball cases often make good tests of the concepts behind doctrine, and Ohio v. Johnson33 is an oddball case. The state charged two pairs of greater- and lesser-included offenses, murder/manslaughter and robbery/theft. At arraignment, the defendant offered to plead guilty to the lesser charges in each pair. Though the prosecutor objected, the judge accepted the guilty pleas to the lesser charges and dismissed the greater charges as violative of double jeopardy.
The Court held, 7-2, that double jeopardy did not forbid the state from proceeding on the greater charges. But finding a rationale in traditional doctrine, which tracks Models 1 and 2, was not easy. The Court had to distinguish Brown v. Ohio,34 its seminal same-offense case that held that a lesser-included offense is always the same as the greater offense. It distinguished Brown by limply arguing that Johnson had “only offered to resolve part of the charges brought against him, while the State objected to disposing of any of the counts against respondent without a trial.”35 This is true enough, but the Court did not explain why the defendant’s actions should change the meaning of “same offense.” If a greater- and a lesser-included offense are always the same offense, as Brown certainly suggests, Johnson should have won (this was the argument of Justices Stevens and Marshall in dissent).
And if greater- and lesser-included offenses are not always the same offense, which is implicit in Johnson, we need an account of when to ignore the included-offense test. The Court’s weak account in Johnson echoes dicta from a Justice Black opinion (inapposite on the merits) about preventing the harassment of defendants. Presumably, Johnson had not been harassed, because it was his idea to plead guilty to the lesser charges and the prosecutor objected on the spot. While this works to explain Johnson as a gloss on Model 2’s view that “same offense” has a content independent of legislative intent, harassment is what I call a “soft-edged” principle—one that does not provide much guidance to courts. I also argue that harassment is a due process concept, based on what the Schad Court termed a “distillate” of “fundamental fairness” and “rationality.” Doctrines that are indistinguishable from due process should be removed from double jeopardy and relocated to the Due Process Clause.
Model 3 operates much more cleanly here. We ask whether the legislature authorized judges to accept plea bargains over the prosecutor’s objection, thereby preventing the state from proceeding with trial. In this country, the decision of whether to proceed with trial belongs to the prosecutor, not the judge. Suppose the judge had told the prosecutor, “I don’t think you should bring these greater charges, though I have not heard the evidence and cannot enter a judgment of acquittal. So I will dismiss with prejudice the greater charges.” It seems clear to me that the legislature did not authorize this kind of judicial oversight, which is the effect of what the judge did in Johnson.
If a review of the state rules of procedure bears out this assessment, then the judge has acted in derogation of the legislative prerogative that prosecutors decide which charges to bring to trial. Since double jeopardy, under Model 3, operates only to enforce legislative judgments (actual or presumed) about “offense” and “jeopardy,” it cannot be deployed in derogation of the legislative scheme. Johnson loses under Model 3, just as he does under Model 2, but he loses because double jeopardy is simply inapplicable to oppose legislative judgments, rather than because of some vague notion that he was not harassed.
The difference between this model and the first two is fundamental: under Model 3 defendants receive, as double jeopardy protection, no more convictions than the legislature has authorized, either by virtue of the way the legislature has defined “offense” or by virtue of how the legislature meant for one exposure to jeopardy to end. But there is no substantive component of double jeopardy that limits the legislature. If the legislature created an offense that always required reprosecution of conduct underlying another offense, and made it clear that reprosecution is what it intended, Model 3 says that the double jeopardy inquiry is at an end. We do not have to agonize over whether the legislature is permitted to create two cumulatively punishable substantive criminal law offenses.
The Supreme Court considered a statute very close to the one just hypothesized in Garrett v. United States.36 The unusual federal offense of “continuing criminal enterprise” (CCE) requires proof of predicate felonies (along with other elements). While these felonies do not have to be manifested in prior convictions, they usually are (because that makes the prosecutor’s job easier and because it is usually the prior charges and convictions that tip off the authorities to the existence of the criminal enterprise). Garrett was convicted in federal court in the state of Washington for importing marijuana. He was later prosecuted in Florida for CCE. The Government offered the Washington felony conviction as part of its CCE proof. These are the same offense under the included-offense test because CCE always requires proof of other felonies, here the Washington felony.37
In analyzing the double jeopardy same-offense question, the majority began by asking whether Congress intended to create “separate” offenses, with neither being a “substitute” for the other.38 This is the right way to phrase the legislative-intent question. The question, as the Court realized, is not whether Congress meant to create an offense in addition to the predicate felony offenses; of course Congress meant to do that. The key question is whether Congress meant for the CCE offense to operate separate from the predicate felonies, rather than as a substitute for them. One way Congress could have envisioned the problem is as follows: if someone commits a sufficient number of predicate felonies, then we want that person convicted of CCE in lieu of the other, presumably lesser, predicate felonies. If Congress meant that, of course, it did not mean to create separate offenses.
The answer to the separate-offense question is the answer to my Model 3 question about legislative intent to create different double jeopardy offenses. Having gotten off to a good Model 3 start, the Court in Garrett quickly complicated matters, rephrasing the inquiry as three separate questions: (1) whether Congress intended to create offenses that could be prosecuted separately; (2) whether the Double Jeopardy Clause bars separate prosecution of these offenses; and (3) whether the Double Jeopardy Clause bars cumulative punishment of these offenses. On a Model 3 view, this is the same question asked three different ways. Once a court determines that Congress did (or did not) create separate criminal law offenses, we also know how many prosecutions and punishments are permissible under the Double Jeopardy Clause. The Court’s opinion in Garrett is, by comparison, a Gordian knot of different strands of analysis.
Read one way, Garrett suggests that legislative intent is the only relevant criterion.39 But after holding that Congress intended to create offenses that could be prosecuted separately, the Court then went on to discuss at length whether the Double Jeopardy Clause would permit the authorized separate prosecutions and cumulative punishments. The opinion is puzzling even on its own terms. After concluding that the Double Jeopardy Clause permitted separate prosecutions, because CCE is an exception to the normal Brown v. Ohio application of the lesser-included-offense test, the Court treated as undecided the issue whether cumulative punishments were permitted. But on any coherent view of double jeopardy, deciding that two offenses may be prosecuted separately necessarily answers the question of whether the Double Jeopardy Clause permits cumulative punishments. What kind of constitutional scheme would permit separate prosecutions and then forbid the penalty sought in the second trial?
The Court’s failure to apply a Model 3 analysis to the successive-prosecution question led it to a convoluted, unsatisfying analysis. The Court distinguished “typical” applications of the included-offense test because Garrett’s criminal conduct continued after the first conviction. Though the majority is right, the point is both obvious and trivial. If a defendant is convicted of conspiracy, and the conspiracy continues after the conviction, the defendant can be tried again for the subsequent conspiratorial conduct. This distinction is less apt in Garrett, as the dissent pointed out, because all the conduct for which Garrett was prosecuted was over by the time of the first trial.40
Rather than respond to the dissent’s point on conduct, Justice Rehnquist wrote for the Court, “One who insists that the music stop and the piper be paid at a particular point must at least have stopped dancing himself before he may seek such an accounting.”41 This analogy hardly qualifies as an answer to the dissent and only demonstrates the emptiness of the conceptual Model 2 double jeopardy universe. The Court seemed to mean its “paying the piper” metaphor to answer some kind of generalized fairness question—whether it was fair for the government to charge these offenses in separate indictments. Six members of the Supreme Court found no unfairness, citing the relatively small part the Washington felony played in the five-and-one-half-year continuing criminal enterprise.42 Three members of the Court found unfairness, at least in part because the CCE case could have been proven using other offenses alleged in the CCE indictment.43 This kind of analysis is precisely why double jeopardy is in such a mess today. On what basis could anyone judge which position best manifests a vague double jeopardy policy against unfairness?
It is better to start over. Starting over, in my view, means returning to the history and policies underlying double jeopardy. Though Blackstone was far from the beginning of double jeopardy history, his 1764 summary is the most complete, satisfying statement of the doctrine and policy yet written.
Blackstone wrote at length about double jeopardy in his Commentaries. Though some of the meaning will be unclear at present, in part because of the differences between the eighteenth-century legal system and procedure and those in use today, we can begin by quoting Blackstone on double jeopardy. A particularly jarring anachronism must be noted: the use of “appeal” in Blackstone’s day meant a prosecution by a private citizen, to be distinguished from the king’s indictment. It is difficult to keep out of one’s mind the modern usage of appealing to a higher court, and I thus occasionally use brackets to remind the reader of Blackstone’s meaning. To make Blackstone’s contribution more apparent, I break it into parts and pause to show how it fits with a modern conception of double jeopardy. All of this, however, is by way of summary. The rest of the book explains and defends a blameworthiness conception of double jeopardy more fully. Blackstone begins:
Special pleas in bar … are of four kinds: a former acquittal, a former conviction, a former attainder, or a pardon. There are many other pleas which may be pleaded in bar of an appeal [a private prosecution]; but these are applicable to both appeals and indictments.
I. First, the plea of autrefoits acquit, or a former acquittal, is grounded on this universal maxim of the common law of England, that no man is to be brought into jeopardy of his life or limb more than once for the same offence. And hence it is allowed as a consequence, that when a man is once fairly found not guilty upon any indictment or other prosecution, before any court having competent jurisdiction of the offence, he may plead such acquittal in bar of any subsequent accusation for the same crime. Therefore an acquittal on an appeal [private suit] is a good bar to an indictment on the same offence. And so also was an acquittal on an indictment a good bar to an appeal, by the common law; and therefore, in favor of appeals, a general practice was introduced not to try any person on an indictment of homicide till after the year and day, within which appeals may be brought, were past; by which time it often happened that the witnesses died, or the whole was forgotten. To remedy which inconvenience, the statute 3 Hen. VII. c.1 enacts, that indictments shall be proceeded on immediately at the king’s suit, for the death of a man, without waiting for bringing an appeal; and that the plea of autrefoits acquit on an indictment shall be no bar to the prosecuting of any appeal.44
In Blackstone’s day, the private prosecution (the appeal) was a criminal trial, simply brought by a different prosecutor. Indeed, at common law it was the preferred method of prosecuting crime, as it had been for a thousand years or more. This can be seen in the common-law rule that the king had to wait a year and a day before bringing an indictment in order to let the private prosecutor have the first opportunity. At least one thirteenth-century commentator (Britton) treated the private appeal and the king’s indictment as separate realms, permitting both to go forward.45 But by the time of Coke (mid1600s), the law was clear that a verdict in either type of prosecution was a bar to the other, as the year-and-a-day rule makes plain. Blackstone notes the single exception—the Henry VII statute permitting the king to proceed on a homicide indictment immediately without an acquittal barring the private appeal. This suggests that by the time of Henry VII, the monarch was sufficiently interested in controlling prosecutions that the private appeal was beginning to lose its prominence.
Over time, the private appeal withered—today it seems odd even to think of a private citizen prosecuting crime.46 Its place was taken by a more prominent role for the king’s indictments and the civil tort action. Today, as suggested earlier and detailed in chapter 4, there is no overlap between civil tort and crime, just as there was no overlap in Blackstone’s day when a homicide indictment produced an acquittal. Today, as long as an action is civil—making this distinction is the subject of chapter 4—there is no bar to a criminal prosecution based on the same conduct.
Returning now to Blackstone:
2. Secondly, the plea of autrefoits convict, or a former conviction for the same identical crime, though no judgment was ever given, or perhaps will be, (being suspended by the benefit of clergy or other causes) is a good plea in bar to an indictment. And this depends upon the same principle as the former, that no man ought to be twice brought in danger of his life for one and the same crime. Hereupon it has been held that a conviction of manslaughter, on an appeal or an indictment, is a bar even in another appeal, and much more in an indictment for murder; for the fact prosecuted is the same in both, though the offences differ in coloring and in degree. It is to be observed that the pleas of autrefoits acquit and autrefoits convict, or a former acquittal and former conviction, must be upon a prosecution for the same identical act and crime.47
Blackstone’s view of greater- and lesser-included offenses is manifested today in the Court’s same-offense test, most clearly articulated in Brown v. Ohio. The idea that conviction of a lesser offense bars prosecution of the greater has been accepted in a line of cases extending back in time between Brown and Blackstone.48 Notice that the plea of autrefois convict is actually broader in one way than that of autrefois acquit:the Henry VII exception to the year-and-a-day rule did not apply to autrefois convict. The common-law view of double jeopardy was roughly as follows: if a defendant was convicted of the same act and crime by either a private citizen or the king, it was an end to the matter (more on this in relation to autrefois attaint). If, on the other hand, a defendant was acquitted, there was the chance of an erroneous acquittal (because “witnesses died, or the whole was forgotten”). This risk of error justified an exception when the king proceeded first in a homicide case and an acquittal resulted.
With that single difference to one side, Blackstone noted the similarity between the two pleas: “It is to be observed that the pleas of autrefoits acquit and autrefoits convict, or a former acquittal and former conviction, must be upon a prosecution for the same identical act and crime.” Of course, he is using “identical act and crime” in a loose way because he gave as an example the offenses of murder and manslaughter. They are identical only in that manslaughter is necessarily included in murder. Again, this suggests the Court’s greater- and lesser-included offense test from Brown. Though I will suggest modifications to the greater- and lesser-included offense test, it is far better than any other test, certainly far better than the “identical-offense” test recently offered by Akhil Amar and Jonathan Marcus.49 Once again, Blackstone had the right idea about double jeopardy.
In the next section, Blackstone contrasts autrefois acquit and autrefois convict with a much broader plea. Recall that he ended the prior discussion by noting that the pleas of former acquittal and former conviction “must be upon a prosecution for the same identical act and crime.” Now he continues:
But the case is otherwise, in
Thirdly, the plea of autrefoits attaint, or a former attainder, which is a good plea in bar, whether it be for the same or any other felony. For wherever a man is attainted of felony by a judgment of death, either upon a verdict or confession, by outlawry, or heretofore by abjuration, and whether upon an appeal or an indictment, he may plead such attainder in bar to any subsequent indictment or appeal for the same or for any other felony. And this because, generally, such proceeding on a second prosecution cannot be to any purpose; for the prisoner is dead in law by the first attainder, his blood is already corrupted, and he hath forfeited all that he had; so that it is absurd and superfluous to endeavor to attaint him a second time. But to this general rule, however, as to all others, there are some exceptions; wherein cessante ratione, cessat et ipsa lex [the reason ceasing, the law ceases]. As, 1. Where the former attainder is reversed for error, for then it is the same as if it had never been. And the same reason holds where the attainder is reversed by parliament, or the judgment vacated by the king’s pardon, with regard to felonies committed afterwards. 2. Where the attainder was upon indictment, such attainder is no bar to an appeal [private prosecution], for the prior sentence is pardonable by the king; and if that might be pleaded in bar of the appeal, the king might in the end defeat the suit of the subject by suffering the prior sentence to stop the prosecution of the second, and then, when the time of appealing is elapsed, granting the delinquent a pardon. 3. An attainder in felony is no bar to an indictment of treason; because not only the judgment and manner of death are different, but the forfeiture is more extensive and the land goes to different persons. 4. [omitted].50 And from these instances we may collect that a plea of autrefoits attaint is never good but when a second trial would be quite superfluous.51
Much of this discussion is anachronistic. There is no modern analog to the plea of autrefois attaint because its very premise is no longer viable—that a felony judgment corrupts the blood and causes forfeiture of all property and rights (becoming “dead in law,” to use Blackstone’s arresting metaphor). With that kind of draconian, permanent change in status, little wonder that a single attaint could be pled in bar of an indictment for any other felony, not just the same offense. As Blackstone explained, “it is absurd and superfluous to endeavor to attaint him a second time” if the attaint is still in effect.
The reader might wonder why the common law developed a plea of autrefois convict if it had the generally broader plea of autrefois attaint. The reason was procedural; several mechanisms permitted a convicted felon to avoid having judgment imposed. In these cases, the proper plea was autrefois convict, former conviction (without judgment). When autrefois attaint faded from use, autrefois convict became the proper plea for all cases that sought to use a former conviction to bar a second trial. Like the plea of former acquittal, former conviction was a good plea only if the offenses were the same.
Two wrinkles in the law of attaint still have relevance, however. First, an attaint upon an indictment barred another indictment but not a private appeal; again, this demonstrates the separate nature of the private suit, which still exists today, albeit in modified form, in the distinctiveness of the tort and the criminal systems. Blackstone’s rationale for this exception also explains the distinct tort/criminal systems: if the king pardons the conviction after the time for appeal has passed, the private citizen is left without a remedy. This explains why, for example, the families of Nicole Simpson and Ronald Goldman could obtain a multimillion-dollar civil judgment against O. J. Simpson despite his acquittal in the criminal case.
Second, notice the effect of the attaint being reversed—the same effect today of almost all appellate reversals: the defendant may be tried again “[w]here the former attainder is reversed for error, for then it is the same as if it had never been.” The difficulty of getting a former attainder reversed explains Blackstone’s preference for when pardons should be pleaded in bar:
4. Lastly, a pardon may be pleaded in bar; as at once destroying the end and purpose of the indictment by remitting that punishment which the prosecution is calculated to inflict. There is one advantage that attends pleading a pardon in bar, or in arrest of judgment, before sentence is passed, which gives it by much preference to pleading it after sentence or attainder. That is, that by stopping the judgment it stops the attainder and prevents the corruption of the blood, which when once corrupted by at tainder cannot afterwards be restored otherwise than by act of parliament.52
The influence of Blackstone can be seen in Supreme Court opinions and early American law review commentary. In 1883 the law review commentary summarized the same offense universe in a way that both Blackstone and the current Supreme Court would understand and endorse. Charles Batchelder recapitulated the rules of former jeopardy as follows:
1. That a conviction or acquittal of the whole crime bars further prosecution for the whole or any part of the crime.
2. That a conviction of an included offence on an indictment for the greater crime bars further prosecution for the whole or any part of the crime.
3. That a conviction or acquittal of an included offence on an indictment for the included offence only is a bar to the further prosecution of the whole or any part of the crime.53
Batchelder’s summary is manifested in the holding and mode of analysis in Brown v. Ohio, almost one hundred years after Batchelder wrote his summary and more than two hundred years after Blackstone published his Commentaries. The consistency of double jeopardy law from Blackstone to the present suggests that “same offense” is captured in some kind of greater-lesser-offense relationship. To explain the precise nature of that relationship requires the first seven chapters of this book. The next section shows the explanatory power of Model 3, informed as it is by Blackstone’s double jeopardy principles.
The question of when offenses are the same can usefully be viewed as a substantive criminal law problem—a threshold double jeopardy issue. There is no reason to be concerned about the twice-in-jeopardy question if the offenses are not the same.
Jay Sigler asserts that the “same offense” question is the “greatest puzzle in state double jeopardy law,” one that also “seems to bedevil the federal courts.”54 One law student comment perhaps said it best in its title: Two Centuries of Judicial Failure in Search of a Standard.55 Though Sir Matthew Hale presented dozens of examples of when offenses were or were not the same offense,56 Blackstone spent little time on this point. Indeed, he seemed to view the question as almost self-evident when he noted that murder and manslaughter were “the same identical crime” because “the fact prosecuted is the same in both, though the offenses differ in coloring and in degree.” The “same offense” part of Blackstone’s conception seems to turn on whether the crimes involved the same “fact.” Nineteenth-century courts were attracted to this idea. Many sought to determine “same offense” by trying to decide whether “the same act” was prosecuted more than once.57
To the extent this inquiry demanded an account of “same act” existing in the universe, it was doomed to failure. In a highly influential student note about double jeopardy, Larry Simon drew on J. L. Austin to state, “Whether any span of conduct is an act depends entirely upon the verb in the question we ask. A man is shaving. How many acts is he doing? Is shaving an act? Yes. Is changing the blade in one’s razor an act? Yes. Is applying lather to one’s face an act? … Yes, yes, yes.”58
This critique is effective only against a metaphysical account of “act.”59 But there is no reason to use a metaphysical account in double jeopardy analysis. A more defensible account is to locate “act” in the legislative description of the prohibited conduct. Indeed, it is possible that Blackstone saw “fact prosecuted” as “fact prohibited by the criminal law.” This account makes it possible to defend a “same act” definition of same offense.
When deciding how to understand “acts” prohibited by the criminal law, the most obvious source would be the statutory language. If a statute were to make “shaving” a crime, for example, courts would have to determine what the legislature meant by “shaving.” The legislative history might be helpful, or the context in which the word appears in the statute. Presumptions (for example, the so-called “rule of lenity”)60 may have to be pressed into service. Whatever the outcome, whatever difficulties it entails, it is a very different inquiry from the one Simon had in mind, and it is not subject to Austin’s metaphysical critique.
The Supreme Court has adopted at least two presumptions about “same offense,” as we see in chapter 2. The Court has also held that these presumptions are rebuttable by clear evidence of contrary legislative intent, at least when the issue is whether offenses tried together in a single trial are the same. While the Court’s conception is a reasonable starting point, it is radically incomplete and sometimes produces answers that cannot be justified. The Court held in 1958 that selling the same quantity of narcotics can be three double jeopardy offenses: selling not from the original package, selling not pursuant to a written order, and selling with knowledge of unlawful importation.61 This is a rather puny conception of legislatively decreed “same act.” The Court held in 1980 that rape and murder are the same offense.62 This seems too robust.
The task in chapter 5 is to offer a better justification for the Court’s fuzzy same-offense picture by showing that legislative intent, and presumptions about legislative intent, are the ultimate source of guidance on the same offense question. Chapter 5 also refocuses the same-offense picture by including some additional offense pairs and excluding others that the Court has included. Finally, it improves the clarity of the picture by unifying the successive prosecution problem with that of multiple punishments in a single trial.
The same offense problem is, of course, not the only double jeopardy problem. Many other problems can be lumped together as twice-in-jeopardy double jeopardy issues, discussed in the next section.
In Blackstone’s day, questions about second jeopardies had to do with the kind of trial outcome—a conviction with judgment, conviction without judgment, or acquittal. To shift Blackstone into modern times, we have but to rephrase the outcome question. Did the defendant receive an outcome (verdict, sentence, guilty plea) that the legislature intended to count as a final determination of blameworthiness? If so, that defendant has completed a procedural jeopardy, and a new jeopardy for the same offense cannot be instituted.
The focus on the legislative definition of what constitutes a final outcome parallels the view offered so far that privileges the legislative definition of offenses in terms of proscribed acts. There is much work yet to be done to defend this view of double jeopardy as premised on legislative prerogative. In sum, the argument is that double jeopardy protects procedurally what the legislature creates in the substance of its criminal law and its criminal procedure.
Here again, the Supreme Court has made a good start in understanding double jeopardy based on presumed legislative intent, holding that a dismissal bars a new trial for the same offense only if the dismissal is equivalent to an acquittal.63 This makes sense because the legislature would not have intended a defendant to face a second trial after the first trial has ended in what amounts to an acquittal. Keeping the focus on what the legislature would have intended permits easy resolution of cases in which the defendant procures an acquittal by fraud. Finding virtually no case law, David Rudstein turned to the treatises, which suggest that double jeopardy does not protect when the acquittal is obtained by fraud.64 Bishop, for example, was clear that “[t]here is direct English authority, and there are numerous judicial dicta, English and American” supporting that proposition.65 Though the Supreme Court has not intimated an answer to the fraud question, a legislative intent theory suggests that the treatises are correct.
In the absence of fraud, acquittals always bar a second trial, and the Court has been sensitive to the reality that “acquittals” can occur in contexts other than a not-guilty verdict at trial. For example, consider an appeal that produces a finding in the defendant’s favor on guilt—a reversal on the ground of insufficient evidence.66 In this context, the Court has held that an outcome that is equivalent to an acquittal (a finding that the state produced insufficient evidence to convict) should be treated as an acquittal. The Court has also held that an acquittal of criminal offense #1 can create a collateral estoppel bar against criminal offense #2 even though the two offenses are not the same offense.67 (A moment’s reflection discloses that of course the two offenses are not the same offense; if they were, there would be no need for a separate doctrine of collateral estoppel.) Collateral estoppel is nothing more than an application of the rule that a substantive determination as to criminal blameworthiness always bars a new proceeding. One way to think of outcomes that are not strictly verdicts but nonetheless resolve the criminal blameworthiness question in the defendant’s favor is that they are “acquittal-equivalents.” If the jury acquits S of robbing victim #1 when S claims that the victim was mistaken about the robber’s identity, that jury has implicitly acquitted S of robbing victim #2 at the same time and place.
On Blackstone’s account (and Justice Story’s as well), double jeopardy protects only against new trials following verdicts, which the Court has extended to include any determination that the defendant was not guilty. The implication of this principle is that outcomes that are neither verdicts nor acquittal-equivalents should not be treated as jeopardy bars. Thus, the typical mistrial—an order terminating a trial because of prejudicial errors in the proceeding—could never be a jeopardy bar because it does not resolve the issue of defendant’s guilt. But the Court so far has refused to follow this logic or to follow Blackstone and Story. In an effort to prevent procedural unfairness to defendants where mistrials were granted too hastily, the Supreme Court has attempted to identify a standard by which a mistrial can be judged a jeopardy bar even when it does not resolve the defendant’s guilt. This attempt has not gone well. The test currently in use is whether the first trial ended because of a “manifest necessity,” a standard that Stephen Schulhofer describes as a “thoroughly deceptive misnomer, perhaps not rivaled even by the Holy Roman Empire.”68
Moreover, to define the end of jeopardy without drawing on an understanding of “verdict” forces one to draw an arbitrary line about when jeopardy begins. If we are going to investigate the fairness of the prosecutor’s mistrial motion, we have to know when the trial has proceeded far enough that the unfairness becomes a double jeopardy problem. That arbitrary line is when the jury is sworn (or, in a trial to the judge, when the first witness is sworn).69 If, on the other hand, we understood verdicts as the only way jeopardy ends (Blackstone’s and Story’s understanding), there would be no need to worry about when jeopardy begins. If the criminal process produces a finding in the defendant’s favor on the facts of the criminal offense, it is a verdict and an end to jeopardy without regard to when in the process the finding occurred.
Keeping the focus on the actual nature of the outcome (without regard to its label) helps provide coherence to a confused mistrial case law. Oddly enough, the Court permitted a defendant to bar a second trial because the prosecutor failed to subpoena an important witness70 but did not permit defendants to show that a hung jury (one that cannot agree on a verdict) was based on insufficient evidence.71 Why should a defendant be in a better position vis-à-vis the Double Jeopardy Clause if the prosecutor failed to subpoena a witness than if the state presented its entire case and failed to introduce sufficient evidence even to submit the case to a jury? It makes no sense. Indeed, the Court seemed to acknowledge the lack of logic in the hung jury case. The mistrial rule, the Court intoned, “has its own sources and logic.”72 Moreover, “Justice Holmes’ aphorism that ‘a page of history is worth a volume of logic’ sensibly applies here.”73
If, rather than follow the Court’s unique logic, we stay with Blackstone and Story, the Court has been both too generous and too strict in its mistrial cases. The Court has been too generous in creating a mistrial doctrine separate from acquittal equivalence, causing great doctrinal incoherence in the bargain as courts struggle to put a meaning on “manifest necessity.” The Court has been too strict in refusing to permit a defendant to show that a hung jury is based on legally insufficient evidence, while permitting convicted defendants to gain double jeopardy protection if they show insufficient evidence underlying the conviction. Why should a hung-jury defendant be in an inferior double jeopardy position when compared to a convicted defendant?
With procedural jeopardy understood as turning on an accurate characterization of the initial outcome, many of the complexities of modern criminal procedure become irrelevant, and what appear to be difficult cases are quite simple. For example, the Model 3 view of double jeopardy functions as a crude limitation on sentencing, by limiting the number of convictions. But Model 3 double jeopardy has no other implications for sentencing, despite sentencing procedures that might raise jeopardy issues under the fairness approach that animates the other two models. The federal sentencing guidelines permit a defendant’s sentence to be enhanced on the basis of uncharged conduct. A double jeopardy question is whether this uncharged conduct can later be prosecuted as a criminal offense. It is easy to see why this raises double jeopardy concerns: the government is using the same conduct twice, once to enhance the defendant’s sentence and once to obtain an additional conviction. Is this fair? Is it a form of prosecutorial harassment?
Model 3 operates more cleanly here. If a court’s lodestar were fairness (preventing harassment), who knows whether double jeopardy permits the uncharged conduct to be prosecuted later? But Model 3 clearly permits a later prosecution for uncharged conduct because Model 3 requires a verdict (or acquittal-equivalent) to trigger double jeopardy protection. Nothing about increased sentencing is a verdict that is separate from the verdict which serves as the basis for the sentence. The Court agreed,74 though it used a slightly different rationale because it has yet to accept explicitly Model 3.75
Now consider a variation. Suppose the defendant is tried for two offenses, acquitted of one, convicted of the other, and the federal prosecutor seeks to prove the “acquitted” conduct to enhance the sentence on the conviction. That this would pass the fairness test seems unlikely. After all, the defendant could legitimately claim that she is being made to prove her innocence a second time.76 What is fair about that? But Model 3 permits the government to prove this conduct. Sentencing is not a verdict—cannot lead to a verdict—and thus it is not a jeopardy that is forbidden by the acquittal. The defendant has been found guilty of only one offense; nothing about being found guilty of one offense, and sentenced for that offense, can ever be double jeopardy under Model 3. Of course, this is just another way to say that the legislature can have whatever sentence and however many convictions that it wants (consistent with due process and the Eighth Amendment). The Court agreed, again using a somewhat different analysis.77
Model 3 also explains why there is no jeopardy bar to a new trial following a typical appellate reversal (on grounds other than sufficiency of evidence)78 and why double jeopardy does not prohibit resentencing to a higher sentence following a retrial and new conviction.79 Assume a defendant received a ten-year sentence, appealed the conviction, and secured a reversal on grounds that have nothing to do with the sufficiency of the evidence. As the Court has recognized, this defendant has no double jeopardy claim against a new trial. There is no extant verdict that would give rise to a jeopardy claim. Moreover, if this defendant is reconvicted, double jeopardy does not prohibit a longer sentence. The Court’s metaphor in justifying this result is instructive: the “slate has been wiped clean.”80 Precisely. There is no verdict manifesting blameworthiness for the conduct in question. The state may, therefore, retry the defendant and sentence her to any sentence permitted by law—at least as far as double jeopardy is concerned.81
Guilty pleas should be treated no differently from any other kind of verdict. A guilty plea is a conviction that is an end to further jeopardy for that offense (assuming it is not overturned on appeal). If, however, a defendant wishes to make his guilty plea conviction conditional on providing testimony against another defendant, there is nothing in the Double Jeopardy Clause that prevents the state from enforcing the contract.82 By vacating the first conviction, the state puts the defendant where he would have been if there had been no guilty plea in the first place; there is no verdict to manifest his blameworthiness and thus no bar to the state proceeding with a trial.
So far, then, the Model 3 argument is quite straightforward, at least in its outlines. If a criminal prosecution ends in a verdict, or an acquittal-equivalent, the defendant may not be charged again for the same offense as long as the first outcome remains undisturbed by the defendant. Though much work remains to be done to fill in the details of the “same-offense” picture, it is roughly whether the legislature meant the two statutes to proscribe the same criminal act. As the legislature almost never discloses its intent on these matters, we have to develop a series of presumptions about likely legislative intent. These presumptions draw heavily on the philosophy of action.
One more category of questions awaits us, chameleon issues that could be described as same offense or as twice-in-jeopardy issues. Rather than try to force these issues into one or the other category, let us think of them as separate from the issues of same offense and jeopardy.
One perplexing issue in criminal law generally is finding the dividing line between civil sanctions and criminal penalties. An example that is close to the line is an action denominated as civil in which the government seeks treble damages for an intentional tort.83 Outside double jeopardy, the issue has purely procedural implications. If a proceeding is civil, the defendant is not entitled to a jury or to the standard of proof beyond a reasonable doubt. Moreover, an indigent defendant is not entitled to counsel at government expense.
But as double jeopardy cannot arise until the first jeopardy is ended, the civil/criminal double jeopardy issue has deeper implications. If the first proceeding—the intentional tort action, for example—is determined to be criminal, the defendant cannot be tried for the parallel criminal offense (assuming the tort and criminal offense meet the relevant test for same offense). In this way, the civil/criminal issue has substantive implications, as it controls the total sanctions that can be imposed for the conduct that gives rise to potential civil and criminal liability.
This problem has proved particularly nettlesome for the Court in the last two decades. The Court made a brave start on the problem in 1989,84 but the dicta from that decision have caused a flood of new cases seeking to use civil sanctions to forestall criminal trials or punishments. In 1994 the Court withdrew some of the robust dicta from the earlier case, insisting that it is only the truly rare civil sanction that constitutes double jeopardy punishment.85 And in 1997 the Court essentially reversed the troublesome 1989 precedent.86
A Model 3 account of double jeopardy solves this problem easily and along the lines the Court finally adopted in 1997. As the legislature determines both what counts as a verdict and when offenses are the same, a civil and criminal offense can never be double jeopardy for the same offense, no matter how much overlap in the offense definitions. The legislature, in enacting both a civil and a criminal sanction, must have wanted both to apply.
An issue that is controversial, but hardly perplexing, is the Court’s “dual sovereignty” doctrine. Under this formalistic view of the relationship between the states and the federal government (as well as the relationships between the states and between Indian tribes and the federal government), each sovereign retained the power to punish criminal acts without regard to whether the other sovereigns have tried or punished the same crimes.87 This, then, is a truly easy double jeopardy issue for the Court: each sovereign may proceed, and no issues even arise about same offense or whether the first trial ended in a criminal verdict.
A Model 3 account, premised as it is on legislative prerogative, agrees that each sovereign has this power, but the relevant question is whether each sovereign intended to use the power to punish in addition to the other sovereign(s). If the sovereign stated its intent to proceed (or not), that would be the end of the matter. Since an expression of this intent is almost always missing, the task for Model 3 is to formulate presumptions about that intent. A useful way to begin is to examine the early opinions of the Supreme Court. The most persuasive argument in those early opinions presumes that neither the federal government nor the states would want to proceed in addition to the other.88 Moreover, the best evidence on the original intent of the Framers, as Paul Cassell has shown, is that they would have understood the Double Jeopardy Clause to forbid overlapping state and federal prosecutions.89
The argument presented in this book is that double jeopardy is a procedural overlay on the substantive criminal law that defines blameworthiness and on the law of procedure that defines criminal verdicts and dictates when those verdicts can be sought by prosecutors. The legislature has the power to define crimes, set punishments, and create a procedural system in which some outcomes (but not others) are final. This legislative prerogative requires that judges and prosecutors respect the substantive law of crimes and the rules of criminal procedure but nothing more.
“Legislative prerogative” is not a new double jeopardy idea. In Brown v. Ohio, the Court held that joyriding and auto theft defined the same Ohio offense.90 The Court defended its holding by noting that it resulted from the way the state courts construed the crimes defined by the legislature. The Court hastened to add: “The legislature remains free under the Double Jeopardy Clause to define crimes and fix punishments; but once the legislature has acted courts may not impose more than one punishment for the same offense and prosecutors ordinarily may not attempt to secure that punishment in more than one trial.”91 Unfortunately, this language is ambiguous. The Court might have meant a robust legislative prerogative that gives the legislative carte blanche simply to state that offense A is different from offense B. Or the Court could have meant a more restrained kind of prerogative, one that merely states a truism: legislatures draft statutes, containing elements that, when construed by courts, lead to a result that may or may not be what the legislature wanted.
Even if Brown meant legislative prerogative in the robust sense, the theory sketched here, and defended throughout the book, is the first attempt to use legislative prerogative as a unifying conception for all of double jeopardy. I hasten to stress what I have said already in this chapter: legislative prerogative is the theoretical construct within which I create a series of presumptions about likely legislative intent. Because the legislature almost never makes its intent known, these presumptions are the principal focus of the book, and decide almost all the cases. Thus, the role of courts is no less central in a double jeopardy world built on legislative prerogative than a world in which courts create the concepts of “same offense” and “jeopardy” out of thin air. The difference is that a double jeopardy conception built on legislative prerogative has a sturdy foundation that can repel attacks easily directed at other double jeopardy theories.
The balance of the book is a defense of a view of double jeopardy based on legislative prerogative, as well as the set of presumptions necessary to make it work. The defense necessarily involves comparisons with other conceptions of double jeopardy. Because comparisons require value judgments, it is useful to briefly sketch the jurisprudence that underlies these value judgments. That is the subject of the next section.
Legal doctrine requires locating what can be called the “edges” of law—the beginning and the ending of the law’s reach. Some rules have harder “edges” than others; hard-edged rules have relatively few close cases at the margin of the law’s reach. The law against driving more than fifty-five miles per hour has completely hard edges. Problems of proof and potential defenses aside, a driver is either going more than fifty-five mph, or she is not. There are zero cases at the margin. A law against traveling at an excessive speed, or against reckless driving, has softer edges because a larger universe of contested cases can be imagined. The Fourth Amendment requirement that a search or seizure be “reasonable” has very soft edges indeed, leading to a jurisprudence that can fairly be characterized as baroque.92
Some have argued that soft-edged laws present judges the opportunity to create individual outcomes responsive to the underlying policy or goals manifested in a particular law. The “legal realists” from the 1920s and 1930s made policy analysis the centerpiece of their judicial approach. The legal realist method, which is still taught in law schools and heavily influences judges, is to identify explicitly the policies and values manifested in the law and then seek to decide the case in a way most consistent with these policies and values. Indeed, some legal realists argued that judges can get along better with policy analysis than by following a set of hard-edged rules. Felix Cohen said in 1935: “The age of the classical jurists is over.… The ‘Restatement of the Law’ by the American Law Institute is the last long-drawn-out gasp of a dying tradition.”93
But Cohen’s view has not prevailed. It seems unlikely that even the most exacting analysis of policies and values can replace rules. If a statement of policy could replace rules, then all criminal procedure rules could be replaced by Kafka’s policy command inscribed by the needles of the Harrow as a means of executing its last victim. The command was “Be Just.”94 If one rejects “Be Just” as a replacement for the Bill of Rights and the rules of criminal procedure, it must be because there is value in having hard-edged rules that both guide and limit judges in dispensing justice.
The critical legal studies scholars from the 1970s and 1980s undermined the part of the legal realist enterprise that manifested faith in judges. Critical theory partisans agreed with the realists that law has surprisingly soft edges—is largely “indeterminate,” in their terminology—permitting judges often to decide cases for reasons other than the law’s rule. Some of these writers seemed to suggest that law is wholly indeterminate, that it never compels a particular outcome. For example, Gary Peller and others argued that the constitutional provision requiring that the President be at least thirty-five years of age is indeterminate.95
Critical theory parted company with most legal realists in arguing that this indeterminacy is (or can be) pernicious because it permits judges to impose their own personal biases or vision of justice.96 “All law is masked power” is Owen Fiss’s summary.97 On this account, to make judges the ultimate arbiter of justice is to invite injustice.
Critical theory suggests a value in making law as determinate as possible. Law can be made more determinate—more predictable in application—by making its edges harder. Because predictability is a virtue in law, if all other factors are roughly equal, constitutional interpretations should favor hard-edged doctrine. All other factors equal, therefore, courts should eschew balancing tests that leave the outcome of cases in doubt until a judge weighs various policies or values to arrive at the right result for each unique case.
To be sure, many Supreme Court opinions make explicit use of balancing tests; Justice O’Connor seems particularly enamored of them.98 This is, in part, because few laws can have edges as sharp as the fifty-five-mph law. To fill in the gaps, to answer the questions left in doubt in what Richard Posner calls the “open” areas of law,99 courts must identify and apply the underlying policy goals of the law they are interpreting. Fourth Amendment doctrine currently permits police to search a residence without a warrant if they have sufficient reason to believe that a life is in jeopardy.100 This kind of doctrine has very soft edges, and we would not want it any other way. We want police to err on the side of searching if they believe they can save a life, and we are perfectly content to let courts review these decisions with a very soft standard.
Most judges and scholars would agree, however, that if everything else is roughly equal, hard-edged rules or doctrines are preferable. In the Fourth Amendment area we might, for example, prefer a harder-edged rule to govern when police can conduct body cavity searches of school children for drugs. In controlling the risks created by reckless driving, a speed limit may be a more appropriate tool than an offense of reckless driving. The issue is always whether the soft edges are necessary to accomplish what the legislature intends.
One critique easily made of double jeopardy doctrine is that some of it has unnecessarily soft edges—for example, much of the “manifest necessity” doctrine that determines when the first jeopardy has ended. As the Court implicitly observed in Arizona v. Washington,101 “manifest necessity” either means what it literally says (that there was no choice but to terminate the trial) or has no real content beyond what the Court has tried to supply in a series of cases. Washington rejected the literal meaning, and courts continue to struggle to decide how to apply “manifest necessity” cut off from its literal meaning. What does “necessity” mean if not “necessary”?
But the preference for hard-edged rules is not limitless. The Supreme Court has, since 1824, held that a hung jury always permits a second trial. This is a very hard-edged rule; it has no exceptions. Rules, however, can be both hard-edged and wrong. This is the limitation in my stated preference for hard-edged rules: “everything else” must be “roughly equal.”
What is not “equal” in the Court’s hung-jury rule is that the universe of outcomes does not correspond with the best statement of double jeopardy policy. This reliance on policy is not the rejection of rules espoused by the legal realists but is, rather, the use of policy to choose among various hard-edged rules. A second kind of critique, therefore, depends on comparing the outcomes of the Court’s doctrine with double jeopardy policy. Later chapters make this kind of critique not only of the Court’s hung jury rule but also of the Court’s mechanical, hard-edged test for same offense, as well as some scholarly conceptions of double jeopardy.
Choosing one rule over another thus requires a judge to consider the interplay between the value of hard edges—predictability—and the degree to which the outcomes under each rule seem to be consistent with the rule’s underlying policy and goals. The interplay is complex and itself subject to a claim of indeterminacy. A judge might, for example, prefer a rule with slightly softer edges if it avoids a subset of outcomes that are contrary to the policy that supports the rule. This is my argument against both the Court’s hung-jury rule and its same-offense test. But is there a principle to justify any particular trade-off of predictability for a slightly more satisfying set of outcomes?
I believe that such a principle or set of principles can be developed. Indeed, the jurisprudence sketched here is similar to that of Ronald Dworkin and can probably be defended as Dworkin defended his.102 But my book is about the history, policy, and doctrine of double jeopardy, not about jurisprudence. I am content to rest the remainder of my argument on the following general principles. First, all things being roughly equal, hard-edged rules are to be preferred over rules with softer edges. Second, in deciding whether all things are roughly equal, judges must understand as completely as possible the policy and values manifested in the rule. Using policy to evaluate a rule is a tricky business, of course, because statements of policy themselves must be reasonably hard-edged to be helpful. The call to “Be Just” does not permit a judge to evaluate very many rules. We must therefore seek a relatively hard-edged policy statement to guide our double jeopardy inquiry.
“Legislative prerogative” as tentatively described in this chapter is a policy statement that is relatively hard-edged, but is it the best policy statement that can be articulated? That is the subject of the balance of the book.