1

Substance versus
Procedure

When you look at the law from a distance, you see a maze of rules.
This is the maze that ensnared the accused Joseph K. in Kafka's The
Trial when he tried to determine whether he was guilty of a crime.
If you look at the maze more carefully, you find that the rules break
down into two general categories, rules of substance and rules of pro-
cedure. The substantive rules define the crimes that are punished in
the particular state or country. If Joseph K. was guilty of a crime,
that crime would have been defined in the substantive rules of the local
criminal law. If those rules are secret or too complicated or too vague
to understand, then the legal system inhumanely drives people to
anxiety about whether they are guilty of a transgression against the
rules.
Being guilty is one thing; being prosecuted and punished another.
Whether one is ever held liable for a particular offense depends on the
rules of procedure. These rules determine how the state enforces the
criminal law by proving the occurrence of crime and convicting and
punishing those responsible for the crime.
In general terms, we can say that the substantive rules establish
"guilt in principle." The procedural rules determine whether individ-
uals are "guilty in fact." Whether guilt in principle becomes guilt in
fact depends on several factors—on the evidence available, on the rules
for introducing and evaluating this evidence, and on the personalities
and talents of those charged with making the decision of guilt. The

7
8 Basic Concepts of Criminal Law

agony of Joseph K. derives not only from the inscrutable rules of sub-
stance but from the torture of undefined procedures.
The rules of legal procedure allocate functions among the lawyers,
the judge, and the jury or the lay people who assist the judge in finding
the facts of the cases. They also determine the scope of admissible ev-
idence, prescribe provisions for appeal, and establish criteria for re-
versing judgments and starting all over again. These rules determine
the way the game is played. And the game is always played the same
way, whether in the particular case the rules lead to what appears to
be a just result or not.
In real games and sports, curiously, we rarely find procedures for
litigating disputes. The substantive rules of card games, chess, baseball,
hockey, and other games determine when one side scores a point, loses
a piece, or commits a foul. If there is a dispute about the facts to which
the rules apply (did the ball hit on this or that side of the line?), the
rules for settling disputes are typically no more complicated than "the
umpire decides" or "each side calls its own fouls." In most areas outside
of the law, we make do with informal processes that depend on the
good faith of all concerned.
Games assume the good faith of all participants. But the law as-
sumes rather that litigants are motivated by self-interest. To secure their
ends, they might well act in bad faith. For this reason, the procedures
for settling disputes are as important as the rules that determine, in
principle, who should win and who should lose.
We may understand the general points behind the distinction be-
tween substance and procedure, but do we understand how the dis-
tinction works in practice? Let us consider the problem more deeply.
Our reflections on establishing guilt under the law are summarized
in the following syllogism:
Major: Whoever intentionally kills another person is
guilty of murder.
Minor: On January 1, 1996, John Jones intentionally
killed Bruce Barnes.
Conclusion: John Jones is guilty of murder.
This is the "syllogism of legal guilt." The major premise is defined by
the rules of substantive law. The minor premise is a matter of fact, and
the facts are established by following the procedures laid down in pro-
cedural rules, namely, the rules for conducting a fair trial.
Note there is also a process or a procedure for determining the
major premises. The rules of substantive law are not self-evident. The
trial judge determines what these rules are by researching the law in
the books or by asking for briefs from the lawyers on questions of law.
Surprisingly, there are no fixed rules for fathoming the rules of sub-
 Substance versus Procedure 9

stantive law. The process is informal, and much depends on how par-
ticular judges like to work.
In a system based on jury trial, as in the United States, the judge
expresses the major premises of the law in his instructions to the jury.
The jury determines the factual issues in the minor premises, and then,
when the system works properly, the jury applies the law to the facts,
the major premise to the minor. Jury instructions also contain proce-
dural rules, such as one requiring the jury to be convinced beyond a
reasonable doubt that a fact relevant to the minor premise is true. If
the jury has doubts that it identifies as reasonable, then it may not
regard the fact as proven.
Most constitutions of the world are more concerned about proce-
dural rights than about rights to a substantive law of a certain sort. The
Fifth and Sixth Amendments to the U.S. Constitution list an array of
rights (e.g., right to counsel, jury trial) that are designed primarily not
to promote the efficiency of the trial but to protect the interests of the
accused.
With regard to substantive law, the most common constitutional
provision today bespeaks the liberal principle that states must advise
their citizens in advance of the substantive rules of conduct which
might trigger criminal liability. The U.S. Constitution expressly prohib-
its ex post facto laws [no legislation after the fact].1 The same rule is
made explicit in the 1949 German Basic Law and in virtually all modern
constitutions.2 This excludes a certain set of possible major premises,
namely, those rules that are legislated as statutory law after the facts
in the minor premise have occurred. It follows that the date of the law's
enactment is critical to whether the major premise is constitutionally
acceptable. A more complete version of the major premise in the ex-
ample would read, therefore:

As of January 1, 1996 (the date mentioned in the minor prem-

ise), it was the law of this state (or country) that:
Whoever intentionally kills another person is guilty of murder.

Adding one complication invites another. Now that we have tied down
the law to a particular date and place, we must add the qualification
the crime occurred in the place (or under other circumstances) that
give the court "competence" over the alleged crime. Adding the re-
quirement of judicial competence changes both the major and minor
premises of the syllogism of legal guilt. The full statement becomes:

Major: As of January 1, 1996, it was the law of this state

(or country) that: Whoever intentionally kills an-
other person within the competence of the court
is guilty of murder.
10 Basic Concepts of Criminal Law

Minor: On January 1, 1996, John Jones intentionally

killed Bruce Barnes within the competence of the
court.
Conclusion: John Jones is guilty of murder.

1.1 The Philosophical Problem:

Substance versus Procedure
It seems as though we have a good idea of the difference between
substantive rules and procedural rules. In many borderline cases, how-
ever, this distinction is hardly obvious. Take, for example, the statute
of limitations, which prescribes the time limit within which the state
may prosecute a particular crime. This looks like a procedural rule, but
it could be interpreted as substantive by redrafting the major premise.
Suppose that the limitation period for murder is twenty years. Then
the major premise of our example might read:
As of January 1, 1996 (the date mentioned in the minor prem-
ise), it was the law of this jurisdiction that:
Whoever intentionally killed another person on or after January
1, 1976 is guilty of murder.
Note that this formulation shifts the tense of the major premise
from the present to the past. The prohibition is transformed from one
against murder in the abstract to one that exposes the offender to lia-
bility for a period of twenty years. It is as though the major premise
read: if you kill someone, you are guilty of murder for twenty years
and no longer. But what is wrong with this formulation? The question,
I suppose, is whether we desire to have the norms of the criminal law
express general moral principles or whether they should define the
conditions under which the state may deprive an individual of his or
her liberty. If you take the view that the criminal law should state moral
rules, the prohibition should be against murder in general; if the pur-
pose is to define the conditions of liability, the latter approach is pref-
erable.
It turns out, then, that in borderline cases the distinction between
substance and procedure raises philosophical issues. We cannot clarify
the distinction without a theory both about the nature of substantive
law and the particular issue we are trying to classify, in this case, the
statute of limitations.
Assessing the nature of the statute of limitations became a burning
political issue in Germany after World War II. The question was how
long the West German government would be able to prosecute con-
centration camp murders under their homicide statute, which carried
 Substance versus Procedure 11

a prescriptive period, a statute of limitations, of twenty years. The initial

German position was that the twenty-year period began running in
May 1945 when the Third Reich collapsed and prosecution became
politically feasible. When the statutory period was about to run out in
May 1965, the Bundestag [Parliament] of the Federal Republic ex-
tended the period for ten years. Before the prescriptive limit took hold
in 1975, the legislature abolished it altogether. When alleged war crim-
inals were prosecuted after 1965 or after 1975, could they legitimately
claim that they were being subject to an ex post facto law, namely, a
rule on prosecuting homicide that was enacted after they committed
their offenses?
There is something unsettling about prosecuting concentration
camp killers on the basis of the German homicide statutes in force at
the time. There is no doubt that if they killed innocent inmates, they
violated the statute. They could claim an exception, perhaps on the
basis of administrative regulations or military orders. Contemporary
German courts reject defenses of this sort on the ground that the im-
plicit instructions to kill were themselves secret and therefore unlaw-
ful.3 What remains is the statute prohibiting homicide. There is no dif-
ference, in the view of German courts, between killing someone in a
1943 Berlin robbery or killing someone in a 1943 Auschwitz gas cham-
ber.
The legislature's extending the statute of limitations differs argua-
bly from the courts' disregarding unjust orders to kill. By extending the
prescriptive period, the legislature changes the time period in which
the alleged criminal is subject to liability. That requires us to answer
the question whether the twenty-year prescriptive period enters in the
definition of the crime that the guards committed. There are two in-
terpretations, one substantive, one procedural:
The substantive If you intentionally kill an innocent per-
interpretation: son, you are guilty of murder for
twenty years. [After the twenty year
period has run, you are no longer
guilty.]

The procedural If you intentionally kill an innocent per-

interpretation: son, you are guilty of murder. You are
subject to prosecution for a period of
twenty years. [After the twenty-year
period has run, you are still guilty but
you cannot be prosecuted.]
Note that in the substantive interpretation, the time period enters
into the definition of guilt; in the procedural case, the time period ap-
12 Basic Concepts of Criminal Law

plies merely to rules for prosecuting the offense. Is it coherent and

plausible to interpret the concept of guilt to include a time period? The
conventional answer requires us to decide whether the statute of lim-
itations is substantive or procedural. How do we decide that ques-
tion?
The general prohibition against retroactive criminal legislation (ex
post facto laws) provides some guidance to answering the question.4
The principle behind this prohibition is that individuals have a right to
know what the "law" is at the time that they supposedly violate it. The
principle is expressed as well in the Latin maxim: nullum crimen, nulla
poena sine lege [There is no crime, no punishment, without prior legis-
lative warning]. While the 1787 U.S. Constitution contains a prohibi-
tion against ex post facto legislation, the 1949 German Constitution
enacts the broader prohibition against punishing in the absence of prior
legislative warning.5 The basic principle is this:
Individuals have a right to know what the "law" is at the time
that they are said to violate it.
But how much of the "law" is included in this principle? Do in-
dividuals have the right to know all aspects of the procedural as well
as the substantive law? Does the individual have the right to know
precisely what evidence might be introduced against him at trial? If,
for example, O.J. Simpson is guilty of murdering his former wife and
Ron Goldman, did he have the right to know at the time he committed
the offense that the prosecution would use evidence of prior spousal
abuse against him? Suppose the law at the time of the killings was that
evidence of the defendant's spousal abuse was not admissible. Suppose
further that after the murder, the legislature intervened and changed
the law to make the evidence of spousal abuse admissible. Would this
have been unfair to Simpson as a criminal defendant?
Whether the evidence of spousal abuse is admissible or not has little
to do with the definition of murder. Simpson had a right to know how
murder was defined in California at the time he allegedly acted, but it
would seem odd to say that he also had a right to know what evidence
the prosecution might use to try to convict him. After all, if he was
guilty, he was guilty of murder—not murder as it could be proved by
admitting evidence of prior spousal abuse.
Some courts would solve this problem simply by saying that the
evidence of prior spousal abuse is an "evidentiary" or "procedural"
matter and therefore there would be nothing wrong with changing the
rule after the date of the suspected murder. In other words, the clas-
sification as procedure would settle the issue.
But the classification is not always so easily made. On borderline
issues, such as the statute of limitations, we have to reach back to the
principle that motivates the classification. What is the intuition that
 Substance versus Procedure 13

enables us to say with confidence that purely procedural matters do

not enter into the "law" that individuals have a right to know before
they act. I suppose the answer is that the rules of procedure do not
bear on the morality of acting. Whether evidence of prior spousal abuse
is admissible against O.J. Simpson has nothing to do with the morality
of killing his wife. We could formulate a principle this way:
Individuals have a right to know that which could make a moral
difference in their choosing to engage in the action or not.
We should remind ourselves that the topic is the permissibility of
retroactive legislation. Ex post facto laws are not permissible if they
infringe on what individuals have a right to know when they act. They
have a right to know, the principle holds, those matters and only those
matters that bear on the morality of their actions. For example, a phy-
sician has the right to know the local definition of death before he treats
a body as dead and begins to remove an organ for purposes of trans-
plantation. Whether society perceives a moribund patient with a flat
EEG reading as dead surely does indeed make a moral difference in
deciding whether to harvest organs from the body. As a result, it would
clearly be unfair to a physician who relied on the definition in force at
the time of his action to have the definition of death changed retro-
actively. Doing so would convert an action that was morally indifferent
into a homicide punishable as murder.
A physician might properly rely on the local definition of death in
reaching a decision whether to make an incision into a body and re-
move its organs; but could you imagine someone calculating whether
to commit murder or not on the basis of whether evidence of prior
spousal abuse could be admitted against him? If the culprit decided to
kill because the evidence of prior abuse would not be admissible against
him, he would hardly be relying on a factor that could make a moral
difference in choosing to engage in the action. Now how do these re-
flections assist us in classifying the statute of limitations as substantive
or procedural?
What do we think of the person who reflects upon the possibility
of killing in the following way: "If I commit this crime now, I am subject
to prosecution, at most, for the next twenty years. This is a risk worth
running." Deciding to kill on the basis of this consideration would hard-
ly be morally superior to killing on the assumption that evidence of
prior spousal abuse would not be admissible at trial. Engaging in highly
immoral acts in the calculated hope of getting away with them is hardly
worth the protection of the law. It would be equally suspect for the
actor to adopt the substantive interpretation of the statute of limitations
and conclude that if he commits the crime, he would be guilty for only
twenty years. The statute of limitation has many purposes, including
setting a limit on the state's power of investigation and prosecution and
14 Basic Concepts of Criminal Law

avoiding trials on the basis of stale, unreliable evidence. It would be

difficult to say that among these purposes was providing an incentive
to commit murder in the hope of getting away with it.
On the basis of these reflections we can conclude that the statute
of limitations is procedural and that, therefore, it was constitutionally
permissible for the German legislature retroactively to abolish the
twenty-year statute of limitations on murder.6 It is worth noting, how-
ever, that the German Constitutional Court distinguished between
cases in which the twenty-year period of limitation had run and those
in which it had not. In cases where the period of limitations had run,
the suspect had the right to rely on the new state of affairs created by
the passage of time. He had no right to rely on the statute at the time
of acting, but after twenty years of exposure to the state's punitive
authority, he was entitled to resume his life without fear of prosecution.
Some courts might disagree with this distinction developed by the
German Constitutional Court. For example, the Hungarian Constitu-
tional Court concluded in 1992 that the post-Communist Parliament
had no authority whatsoever to alter the statute of limitations in force
during the Communist period. It did not matter whether in the partic-
ular case the prescriptive period had run or not. In whatever form it
took, legislative intervention in this area appeared, at least to the judges
of the Hungarian Constitutional Court, to violate the constitutional
provision entrenching the rule of law in the post-Communist legal
system.7

1.2 The Burden of Proof: Half A Loaf

The distinction between substance and procedure comes into play in
many contexts other than the permissible retroactivity of legislation. It
is not easy to fathom, for example, whether the burden of proof should
be treated as a matter of substance or procedure. First, we have to pause
to think about the meaning of the term "burden of proof" and closely
related concepts.
The burden of proof addresses the question: Who wins the trial on
a particular issue in the event the jury (or other trier of fact) cannot
decide one way or another on that issue. Suppose the defendant asserts
self-defense in a homicide case. The jury is convinced that the defen-
dant killed the victim and did so intentionally, but it cannot resolve the
question whether it was done in self-defense. The evidence on that
evidence is simply inconclusive. In cases of this sort, where there is no
way to decide clearly one way or the other, the burden of proof resolves
the tie. If the state has the burden of proving the absence of self-
defense, then the defendant should under these circumstances be
found not guilty. If the defendant has the burden of proof on the issue,
then the result is just the opposite: self-defense is regarded as not
 Substance versus Procedure 15

proven and therefore the defendant should be guilty of murder. What

is at stake, therefore, is not proof but persuasion. The question is: Who
has the burden of persuading the judge or jury on a particular issue. If
you bear the burden and you fail to persuade, you lose on the point.
Therefore, the burden of persuasion is more aptly labeled "the risk of
nonpersuasion." It is obvious that allocating the risk of nonpersuasion
of proof between the state and the defendant can have a radical impact
on the outcome of the trial.
Some European lawyers might maintain that the risk of nonper-
suasion is irrelevant to European legal thinking, the reason being that
the judge—and not the parties—always bears the burden of investi-
gating and establishing the facts. Yet in the nineteenth century, the
same mode of "inquisitorial" trial prevailed in Europe and it was quite
common to allocate the risk of nonpersuasion to the defense.8 The bur-
den of investigation and of fact-finding does not dictate any particular
decision about whether the state or the defendant should bear the risk
that a particular issue, such as self-defense, remains unclarified at the
end of trial.
At first blush, the burden of persuasion appears to be a purely pro-
cedural institution. A more careful look at the various burdens as they
are known in common law trials suggests, however, that they carry
substantive meaning. For the sake of clarity, we should distinguish be-
tween "the burden of persuasion" and two related concepts—the "bur-
den of going forward" and the "standard of proof."
The "burden of going forward" imposes either on the state or on
the defendant the duty of providing sufficient evidence for the court to
take the question seriously as subject to debate. For example, if the
defendant comes into court and simply says, "The decedent tried to kill
me," that would hardly be enough. The defendant must raise enough
evidence to generate a debatable issue, one on which reasonable people
might disagree. If the defendant does not meet this burden, the judge
will simply rule against the defendant on the matter; for example, self-
defense will not be considered an issue in the case. The state also has
the burden of going forward on those matters that it must prove to the
satisfaction of the jury. In a homicide case, for example, the prosecution
must raise a debatable issue about whether the defendant directed his
deadly attack against a living victim. If the state does not meet this
burden, the judge will dismiss the indictment and terminate the trial.
Both jury trials in common law systems and judge trials in Conti-
nental civil law jurisdictions recognize a burden of going forward.9 In
a jury trial, the burden finds its practical expression in the judge's de-
cision to grant or deny jury instructions on the particular question. If
the defendant does not meet the burden of going forward on self-
defense, the judge will deny jury instructions, which means that for all
practical purposes the issue is regarded as not existing in the trial. In
16 Basic Concepts of Criminal Law

bench trials on the European Continent, the judge must write an opin-
ion analyzing the issues raised in the case. If the defendant does not
bring forth evidence to support a particular claim, the judge need not
analyze that issue in the opinion finding the defendant guilty or not
guilty. Silence in the opinion, then, is equivalent to silence in the jury
instructions. In both cases, the judge's silence testifies to a decision that
the defendant has not produced sufficient evidence to raise a debatable
issue for resolution at trial.
The second related concept, "the standard of proof," differs further
from the burden of proof. When the prosecution bears the burden of
persuasion, it must prove these facts "beyond a reasonable doubt." If
the jury has a reasonable doubt on the particular question, it should
find for the defendant—in other words, the prosecution bears the risk
of a reasonable doubt at the end of the case. This is the strictest standard
of proof.
Less strict than "proof beyond a reasonable doubt," the burden of
proving the issue "by a preponderance of the evidence" requires a
showing that the fact is probably more true than false. Proof "by clear
and convincing evidence" requires something more than proof by a
preponderance and less than proof beyond a reasonable doubt. The
assumption behind these diverse standards is that we can speak of proof
in quantitative terms.
A sports analogy might be helpful (American lawyers like to think
of trials as competitions akin to sporting matches). If the different stan-
dards were arrayed on a football field with yard lines numbered from
one to 100, and we thought of bearing the burden of proof as analogous
to moving up the field with the ball, the strictest standard of "beyond
a reasonable doubt" would require taking the ball at least to the 99th
yard line. The standard of "clear and convincing evidence" might be
equivalent to the 70th yard line. And the lowest standard of "prepon-
derance of the evidence" would coincide with the 51st yard line. This
metaphor is useful as well in explaining the duty of going forward,
which designates a rather low standard of bringing the ball, say, to the
10th yard line.
In civil cases [private legal disputes], the standard of proof for both
plaintiff and defendant is usually no more than a preponderance of the
evidence and occasionally proof by clear and convincing evidence. The
idea that the standards of proof in criminal and civil [private law] trials
lend themselves to quantitative measurement marks an important dif-
ference between common law jurisdiction and most Continental courts.
Beginning with the French revolution, European lawyers have come
to use the standard of intime conviction—a subjective standard requiring
the judge's personal conviction. Consequently, Continental lawyers
purport to use the same standard of proof in criminal and civil [private
law] trials.
 Substance versus Procedure 17

This accounts for a common law procedural option that strikes

Continental lawyers as odd. In common law courts and only excep-
tionally in Continental courts, a finding of not guilty in a criminal trial
in no way hinders a subsequent action at private law for damages. This
is precisely what happened in the trials of Bernhard Goetz and O.J.
Simpson, both of whom were acquitted of the serious criminal charges
levied against them, but were then forced to stand trial again for tort
damages.10
The reason that this phenomenon puzzles Continental Europeans
is that they typically use the same standard of proof for both criminal
and tort cases. A finding of not guilty in the criminal trial represents a
final judgment of the legal system that the defendant is not liable for
the acts charged. The finding is considered res judicata—"a thing deter-
mined"—for purposes of the private law trial. The failing of the pros-
ecution is binding, therefore, on the injured victim who might wish to
sue for damages. This is not true in the common law system. A judg-
ment of not guilty in the criminal trial means merely that the prose-
cution had not carried the ball to the 99th yard line. If the jury thought,
say, that the proof carried the case only to the 80th yard line, another
jury in a tort case could well find that the injured plaintiff introduced
sufficient evidence to pass the required 51st yard line. This is all the
plaintiff must do to recover for tort damages.
What matters, in the Continental view of the trial, is not imaginary
lines on a football field, but the judge's response to the evidence and
personal conviction of guilt. As a result of this single concept of proof,
most Continental lawyers sense a contradiction if the defendant is ac-
quitted on criminal charges and then tried again on a complaint for
compensation in tort law.
Another major difference between the common law and the Con-
tinental traditions is that the common law has struggled for centuries
with the question: Who should bear the burden of proof on issues
raised by the defense? Notable among these controverted issues are
self-defense and insanity. Many jurisdictions in the United States re-
quire the defense to bear the burden on either or both of these issues.11
This shifting of the burden to the defense on "defenses" almost never
occurs in Continental courts.
Now what are the implications in common law courts of shifting
the burden of persuasion? Is this purely a procedural rule? Or does it
also have substantive implications? There are at least two ways that the
legal system can strengthen or weaken a particular defense. It can add
qualifications that make it more difficult for the defendant to prevail
on the defense. For example, in a case of self-defense, suppose the
existing rule on self-defense holds that any time the defendant believes,
in good faith, that he is about to be attacked, he may use deadly force
in his defense. The courts tighten the defense, as they did in the Goetz
18 Basic Concepts of Criminal Law

proceedings, by insisting that the defense apply only if the defendant

maintains a reasonable belief that he is about to be attacked.12 If his
belief is unreasonable, he has no valid claim of self-defense. Adding the
requirement of reasonableness undoubtedly makes it more difficult for
the defendant to prevail on a claim of self-defense.
Alternatively, the state has the option of tightening the defense by
shifting the burden of proof on the issue to the defendant. This would
mean that the defendant would have to prove by a preponderance of
the evidence that he did act in self-defense. Combining these options,
we can list, in order of preference to the defendant, the following ver-
sions of self-defense:

Versions of Self-Defense
I. The best option for the defense is a "subjective" standard that
enables the defense to prevail anytime he or she acts in
good faith coupled with the requirement that the prosecution
disprove claims of self-defense by proof beyond a reasonable
doubt.
II. The worst option for the defense is an "objective" standard
that limits self-defense to cases in which the defendant acts
with reasonable belief in the conditions for the defense cou-
pled with a rule requiring the defense to establish the de-
fense by a preponderance of the evidence.
III. Arrayed between these two extremes are two middle posi-
tions:
A) An "objective" standard that limits self-defense to
cases in which the defendant acts with reasonable belief
in the conditions for the defense coupled with the require-
ment that the prosecution disprove claims of self-defense
by proof beyond a reasonable doubt.
B) A "subjective" standard that enables the defense to
prevail anytime he or she acts in good faith coupled with a
rule requiring the defense to establish the defense by a
preponderance of the evidence.
One departure from the best standard for the defendant implies a
shift in the burden of persuasion (IIIB). The other requires the defen-
dant to have a reasonable belief that he is about to be attacked (IIIA).
There is no way to rank these two versions because whether one or
the other is worse for the defense depends entirely on the facts of the
particular case. There are some situations—events that occur behind
closed doors—about which it is hard to know exactly what happened
 Substance versus Procedure 19

and therefore the shift in the burden of persuasion would be devastat-

ing for the defense. There are other cases, such as the Goetz case itself,
where the shooting occurs in public and therefore the primary problem
is not what happened but assessing whether the defendant's fear of
violent attack was reasonable or not.
Now we restate the question with which we began this inquiry: Is
the allocation of the burden of persuasion a procedural rule or a sub-
stantive rule? The answer is not easy, for we see that the allocation of
the burden has an impact on the balance of advantage between pros-
ecution and defense that resembles toughening or weakening the sub-
stantive rule of self-defense. From a functional point of view, rules
allocating the burden of persuasion have the same impact as changes
in the substantive law.
There are some situations in which allocating the burden is clearly
a technique for modifying and therefore softening substantive legal re-
forms. For example, the Model Penal Code advocates a new defense of
mistake of law and to soften the blow for those opposed to the inno-
vation, the Code recommends imposing the burden of persuasion on
the new defense on the defendant.13 There is no apparent reason for
this shift except the politics of law reform. Another example is the
proposal to move from strict liability, in which proof of the defendant's
negligence is not required, to insisting on proper evidence of the de-
fendant's negligence, but with the added twist of requiring the defen-
dant to bear the burden on the issue.14
Does it follow that by analogy to changes in the statute of limita-
tions (before the prescriptive period has run), the state may retroac-
tively change the burden of persuasion? Recall the question we asked
previously: Does the defense have a right to rely on law in force at the
time of his action? It would seem odd to say that the defendant may
not rely on the statute of limitations but he may rely on the require-
ment that the prosecution bear the burden of proof. On the other hand,
we could make the argument of functional equivalence with the sub-
stantive law in regard to the statute of limitations as readily as we can
in regard to the burden of proof. I confess that I am not sure of the
right answer to this quandary. To be sure, we are beginning to see how
difficult and subtle the distinction between substance and procedure
turns out to be in these borderline cases.

1.3 The Distinction in Context

To add to our difficulties in fathoming the distinction between sub-
stance and procedure, we must note that the chameleon-like distinc-
tion takes on different hues in the context of different legal problems.
Common law judges must decide, for example, which issues to decide
for themselves and which issues to send to the jury. The guiding prin-
20 Basic Concepts of Criminal Law

ciple is that the judge resolves questions of law, and the jury decides
questions of fact. The distinction between law and fact correlates with
that between the major and minor premise in the syllogism of legal
guilt, and these two premises dovetail, more or less, with the distinction
between substance and procedure.
This view of "substance" turn out to be broader than any theory
we have encountered so far. Judges decide what the statute of limita-
tions and the burden of persuasion should be, and juries decide
whether the factual preconditions are present for concluding either that
the prescriptive period has run (when did the crime occur?) or that the
burden has been met (how much evidence is there and how convincing
is it?). This approach to the concept of substantive law—namely, as
equivalent to the "law" the judge must determine—sweeps too wide
and therefore provides little guidance to the quandaries we considered
earlier.
The legal context influences our perception of substance and pro-
cedure. Consider a notable example from the field of private legal dis-
putes: the Erie doctrine in the federal courts of the United States.15 To
understand this doctrine, a few words about the jurisdiction of the U.S.
federal courts are in order. In the area of private disputes, the federal
courts hear cases either arising under federal law, including the Con-
stitution, or cases based on diversity of citizenship, which for these
purpose means suits between residents of different states. In these so-
called diversity cases, the courts apply the "common law" of torts, con-
tracts, and other fields of private law. At a certain point in the early
twentieth century, the courts became skeptical about whether the com-
mon law remained a unified whole. If the federal "common law"
turned out to be different from the "common law" in the state courts,
many plaintiffs would go to federal court just to take advantage of the
federal rule. The loss of unity in the common law resulted, eventually,
in the Supreme Court's holding that in diversity-of-citizenship cases,
the federal courts should apply the "substantive" law of the state in
which they were sitting. This made sense intuitively. The purpose of
diversity jurisdiction was not to provide an alternative body of law but
merely to guarantee a neutral forum for citizens of different states.
Accordingly, the court would use its own procedures but would hear
the case as it arose under the local law of the state.
Applying this distinction between state substantive law and federal
procedural law required, of course, the courts to contemplate the dis-
tinction between substance and procedure. A few years after its Erie
decision the Supreme Court had to decide whether the New York stat-
ute of limitations should be classified as one or the other. The Supreme
Court took the occasion of that decision to formulate a general theory
about resolving these problems of classification.16 Because the purpose
of the Erie decision was to ensure the same outcome in the state and
 Substance versus Procedure 21

federal courts, the purpose of classifying issues as substance or proce-

dure should be the same.
The question is whether such a statute concerns merely the manner
and the means by which a right to recover, as
recognized by the State, is enforced, or whether such statutory limi-
tation is a matter of substance . . . namely, does it significantly affect
the result of a litigation for a federal court to disregard a law of a State
that would be controlling in an action upon the same claim by the
same parties in a State court?17
There we have it. Procedural rules are those that concern "merely
the manner and the means by which a right... is enforced" and sub-
stantive rules are those that "significantly affect the result." A few lines
later in the opinion, the Court coins the expression that has caught on
as the "holding" of the case: substantive rules are those that "determine
the outcome of a litigation."18 That has come to be known as the "out-
come-determinative" test.
Unfortunately, virtually all rules "determine the outcome" of liti-
gation. To be sure, the statute of limitations and the burden of persua-
sion fall under this expansive concept of substance. But so do many
minor rules bearing, say, on the law of evidence. Yet this test might be
right for its context—namely, reconciling federal and state law and
avoiding the temptation of "forum shopping," but it would not work
well as a test for the scope of the prohibition against ex post facto leg-
islation or as a medium for distinguishing the scope of the "law" that
the judge rather than the jury should decide.
The distinction between substance and procedure comes into play
as well in the conflicts of laws (the field known in Europe as "private
international law"). If a dispute between private parties arises in one
state or country and then is heard in another, the forum that hears the
dispute will apply its own procedural rules. That makes sense: you come
to my court, you use my rules. Also true, however, is that under the
traditional approach to the conflicts of laws in fields of torts and con-
tracts, the forum would apply the substantive law of the state or coun-
try where the dispute had arisen. It is perfectly normal, therefore, for
California to try a case under its own procedural law but under the
substantive law of New York or France.
It is worth noting that in criminal law, the jurisdiction of the court
determines the applicable substantive law. It is very rare that one court
will apply the substantive law of another. Why is it possible in contract
and tort disputes, but not in criminal cases, to determine liability under
the law of another jurisdiction? The reason, I believe, is that in criminal
cases, the state where the crime occurs is intimately involved with the
resolution of the suspect's guilt or innocence. Crimes typically leave
victims and social anxiety in their wake. Resolving the consequences
22 Basic Concepts of Criminal Law

of the crime invokes local interests in a way that accidents and con-
tractual problems do not. For this reason, criminal trials do not lend
themselves to export.19 They stay in the community where the crime
occurred.20
When legal disputes are exported, however, the receiving court
must decide how much of the substantive law of the foreign jurisdiction
it is willing to apply. The same problems arise with regard to classifying
issues like the statute of limitations and the burden of proof that we
have seen in other contexts. Here, however, the tendency of the re-
ceiving court—in contrast to the Erie jurisprudence—is to take a very
constricted view of substantive law. With few exceptions, the statute
of limitations and the burden of persuasion are treated as procedural:
the receiving court applies, therefore, its own law. The "outcome-
determinative" test does not fare well in this context because the pos-
sibility of an outcome in California, different from, say, in New York or
Mexico, is not so disturbing. The principle is that the courts will rec-
ognize the norms of a foreign jurisdiction as governing the plaintiff's
claim, but it need not recognize every foreign rule that would influence
the trial, had it been held where the claim arose.
These, then, are the five different contexts for seeing the distinction
between substance and procedure at work:
1. Legality: The scope of the prohibition against ex post facto laws
2. Proof: The allocation of the burden of persuasion between the
prosecution and the defense
3. Law and fact: Decisions about which issues should be determined
by a judge and which by a jury
4. Erie doctrine in the United States: Federal courts hearing cases aris-
ing under the substantive law of the state in which they sit
5. Conflicts of law: One court hearing a dispute arising under the
substantive law of a foreign jurisdiction
To summarize our thinking about substance and procedure, let us
recall the different ways of thinking about the the statute of limitations:

Classification of Statute
Context of Limitations
Legality Procedure (for sure)
Erie doctrine Substance (for sure)
Conflicts of law Procedure (generally)

As we have seen as well, the burden of persuasion poses its own

problems of classification because of the functional similarity between
shifts in the burden and changes in the substantive rule. Deciding what
 Substance versus Procedure 23

is law for the judge to resolve and what is fact for the jury to find,
provides it own distinctive take on the distinction between substance
and procedure.
This study in the way in which one fundamental legal distinction
plays itself out in different legal contexts could generate the wrong
message. One might think these decisions of classification were simply
arbitrary and politically motivated. But that would be the wrong lesson
to grasp from this inquiry. The impact of context on legal analysis is
not arbitrary. There are good reasons why the distinction comes out
one way when the concern is legality and another way when the in-
quiry is achieving harmony between state and federal courts sitting in
the same city.
The deep message that unites this chapter with the others that fol-
low is that the basic distinctions of criminal justice transcend the en-
acted law of particular states and countries. The local statutory law does
not determine the boundary that runs between substance and proce-
dure or the way the boundary adapts to the changing context of the
inquiry. The message of this chapter illustrates the overarching theme
of this book. The basic distinctions of criminal justice require philo-
sophical and conceptual analysis. On these matters, you cannot simply
look up the law in the books. You have to think about the problem and
clarify in your own mind the construction of the concepts that makes
the most sense.

Notes
1. U.S. Constitution, art. I, sec. 9, cl. 3
2. German Basic Law §103(II); Mexican Constitution art. 14; Portugese
Constitution art. 29(1); European Convention on Human Rights §7(I).
3. See Judgment of Constitutional Court, October 24, 1996, 1997 Neue
Juristische Wochenshrift 929. For a critical assessment of the Court's allowing
a breach of the prohibition against retroactive liability, see Jorg Arnold, Ein-
schrdnkung des Ruckwirkungsverbot, 1997(5) Juristische Schulung 399.
4. See authorities cited in note 2 supra.
5. It is worth noting the difference between the two Latin maxims. Ex
post facto laws are statutes enacted after the occurrence of the crime. The maxim
nulla poena sine lege is violated if the court, imposes liability, in the absence of
a statute, simply as a matter of common law development. Common law
crimes, so far as they still exist in England and the United States, violate the
maxim nulla poena sine lege but do not infringe on the constitutional prohibition
against ex post facto laws.
6. BVerfGE 25, 269 (Decision of the Constitutional Court 1969).
7. Resolution of the Constitutional Court of Hungary, No. 11/1992 (III.5)
AB.
8. George P. Fletcher, Two Kinds of Legal Rules: A Comparative Study of
Burden-of-Persuasion Practices in Criminal Cases, 77 Yale L. J. 880 (1968).
24 Basic Concepts of Criminal Law

9. It is important to keep in mind the difference between the terms "civil

trial" and "civil law jurisdiction." "Civil trials" litigate matters of private law.
Today virtually every state or country has a "civil code" and this code regulates
private legal transactions, such as torts, contracts, property, etc. "Civil law"
systems are those found on the European Continent. The terminology is mis-
leading and I would not use it if it were not so well entrenched.
10. On the trial of Bernhard Goetz, see generally George P. Fletcher, A
Crime of Self-Defense: Bernhard Goetz and the Law on Trial (1988). There are
many books on the Simpson trial. Two of the most informative and legally
insightful are Alan Dershowitz, Reasonable Doubts (1996), and Jeffrey Toobin,
The Run of His Life: The People v. O.J. Simpson (1996).
11. For a detailed discussion of this problem, see sections 6.1-6.2.
12. See A Crime of Self-Defense, chap. 3.
13. MPC §2.04.
14. The leading case in Canada on this point is Regina'v. City of Saulte
Ste. Marie, 3 Dominion Law Reports 3d 161 (1978) (in cases of strict liability
the defendant must bear the burden of persuasion on the defense of "due
diligence").
15. Erie R.R. Co. v. Tompkins, 304 U.S. 64 (1938)
16. Guaranty Trust v. York, 326 U.S. 99 (1945).
17. Id. at 109.
18. Id.
19. Another exception to the principle of "exporting" disputes to other
jurisdictions is divorce law. Apparently, the community and the state are in-
volved in the maintenance of marriages in the same way they must attend to
the aftermath of crime.
20. See U.S. Constitution, Sixth Amendment, which provides: "the ac-
cused shall enjoy the right to a speedy and public trial, by an impartial jury of
the State and district wherein the crime shall have been committed." For an interpre-
tation of this amendment in line with the argument of the text, see George P.
Fletcher, With Justice for Some: Victims' Rights in Criminal Trials, 169-176
(paperbacked. 1996).
 2

Punishment versus
Treatment

All systems of criminal law represent a shared commitment to acquit-

ting the innocent and punishing the guilty. This shared commitment
confers upon them a single unifying purpose that centers on the insti-
tution of punishment. Without punishment and institutions designed
to measure and carry out punishment, there is no criminal law. It is
fair to say, then, that the institution of punishment provides the distin-
guishing features of criminal law.
The problem is: What is punishment? Not every form of coercion,
not every sanction, constitutes punishment. Not even coerced confine-
ment provides an adequate signal that the criminal law has come into
play. One can lock people up for many reasons—for example, quar-
antine for disease, commitment for mental illness. Not all seizures of
the person are equivalent to the old fashioned punishment of flogging.
Grabbing a person to prevent him from committing suicide is neither
assault nor punishment but rather beneficial coercion. Understanding
criminal law, therefore, requires that we probe the distinction between
punishment and forms of coercion, expressing a benevolent desire to
aid the person affected. With some risk of oversimplification, I refer to
all these alternative, beneficial uses of coercion as "treatment."
The elaboration of the difference between punishment and treat-
ment depends largely on the context and purpose of legal analysis. The
argument tracks, therefore, the analysis in chapter 1 on the distinction
between substance and procedure. Fathoming the contours of punish-

25
26 Basic Concepts of Criminal Law

ment depends not on the positive law of particular states but on the
results of philosophical and conceptual inquiry.

2.1 Two Constitutional Perspectives: Impact versus Motive

For purposes of constitutional analysis, the concept of punishment is
of great importance. Most contemporary constitutions provide greater
procedural protection in criminal trials than in civil or administrative
hearings. In the United States, a proceeding is criminal in nature if and
only if the defendant faces "punishment" as a sanction.
In the United States, the special protection for criminal trials in-
cludes the provisions of the Fifth and Sixth Amendments to the Con-
stitution: among others, the privilege against self-incrimination, pro-
tection against double jeopardy, the right to assistance of counsel, and
the right to a jury trial. As a test for when a threatened sanction is
criminal in nature, the Supreme Court unhesitatingly invokes the con-
cept of "punishment" as the relevant criterion.1 That a sanction is in-
flicted in the criminal courts for a violation of a state or the federal
criminal code is sufficient to classify the sanction as "punitive," but
there are recurrent problems in assessing the punitive nature of other
sanctions, such as administrative commitment, expatriation, deporta-
tion, fines for custom violations, and the deprivation of social security
benefits.2 That Congress or another legislative body has labeled these
sanctions as civil in nature does not control the constitutional issue.
The question is ultimately conceptual or philosophical. The courts must
answer the question whether, no matter what the legislature has said,
the sanction is inherently "punitive." If it is, then regardless of the
legislative label, the process is criminal and the constitutional guaran-
tees apply. As the statutory law cannot demarcate the line between
substance and procedure, it cannot resolve the question whether a
sanction constitutes "punishment" and therefore requires a criminal
proceeding, with full constitutional protection.
One of the best candidates for punishment is physical confinement.
Since the early nineteenth century, we have used prisons as our stan-
dard mode of punishment. Sometimes confinement is dictated for social
protection, sometimes for the purpose of treatment, for the therapeutic
benefit of the person confined. The juvenile court movement early in
the twentieth century conceived of homes for juveniles as a form of
benevolent intervention, designed only to help wayward young people
avoid a life of crime. Because the purpose of confining juveniles was
considered to be treatment, for the good of the juvenile, rather than
punishment, the proceedings that led to the confinement of juveniles
was thought to be exempt from the constitutional protection that sur-
rounded criminal trials.
In the 1960s, civil libertarians mounted a broad attack against the
 Punishment versus Treatment 27

idea that beneficent motives could exempt coercive techniques from

constitutional controls. The argument was that a locked door is a locked
door, whatever the motive. Specifically, with regard to the juvenile
court movement, the Supreme Court concluded that however "eu-
phemistic the title . . .," a home for juveniles was "an institution of
confinement in which the child is incarcerated for greater or lesser
time."3 In the leading case of In re Gault,4 the Supreme Court took
significant steps to expand the constitutionally required protection in
state juvenile court proceedings. Henceforth, the Fourteenth Amend-
ment due process clause would require that juveniles enjoy the privi-
lege against self-incrimination, that they receive the assistance of coun-
sel, and that they be able to exercise the right to confront witnesses
against them. The euphemisms of treatment and rehabilitation paled
in contrast to the reality of incarceration.5 Yet the Court resisted the
arguments of those who sought to equate the confinement and treat-
ment of juveniles with the punishment of criminals. Though many el-
ements of due process apply to juvenile courts, not all of the protections
of the Sixth Amendment are required. The right to a jury trial, for
example, is not included within the constitutionally required proce-
dures in juvenile cases.6
Also, in the 1960s, the prison rights movement mounted a general
attack against indeterminate sentencing, a general practice of sentenc-
ing convicted persons to undefined terms in prison, with parole au-
thorities entrusted with the responsibility of setting a release date.7 This
practice was thought to be justified by the need to provide a program
of treatment that would meet the individual needs of the individual
confined. The attack on indeterminate sentencing stressed two values:
(1) the importance of equality in sentencing from court to court, and
case to case, and (2) the right of the prisoner himself to know at the
beginning of his term when he is likely to be released. The outcome of
this campaign is that most states abolished indeterminate sentencing
and the federal government issued its sentencing guidelines to achieve
greater equalization of sentences in the federal courts.8
Implicit in this critique of indeterminate sentencing was an attack
on the rehabilitative ideal, the aspiration of treating and reforming pris-
oners instead of punishing them. Because the goal of rehabilitation
cloaks the coercive power of the state in benevolent motives, the denial
of liberty is considered less problematic.9 Good motives by the state,
then, can generate a low-visibility threat to individual autonomy and
liberty.
In the debate about juvenile court and indeterminate sentencing,
two different perspectives vie for supremacy. According to one point of
view, if the state's motive is therapeutic, the confinement is nonpunitive
and its imposition is exempt from the procedural niceties of jury trials,
the participation of counsel, and the confrontation of witnesses. Ac-