Frederick Schauer - Formalism

The Yale Law Journal Company, Inc.
Formalism
Author(s): Frederick Schauer
Source: The Yale Law Journal, Vol. 97, No. 4 (Mar., 1988), pp. 509-548
Published by: The Yale Law Journal Company, Inc.
Stable URL: http://www.jstor.org/stable/796369 .
Accessed: 09/09/2014 02:21
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp
.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact [email protected].
The Yale Law Journal Company, Inc. is collaborating with JSTOR to digitize, preserve and extend access to
The Yale Law Journal.
http://www.jstor.org
This content downloaded from 147.9.241.57 on Tue, 9 Sep 2014 02:21:01 AM

All use subject to JSTOR Terms and Conditions
The Yale Law
Journal
Volume 97, Number 4, March 1988
Article
Formalism
Frederick Schauer*
Legal decisions and theories are frequently condemned as formalistic, yet little
* *
discussion has occurred regarding exactly what the term formalism* means.
In this Article, Professor Schauer examines divergent uses ofthe term to eluci-
date its descriptive content. Conceptions of formalism, he argues, involve the
notion that rules constrict the choice of the decisionmaker. Our aversion to
formalism stems from denial that the language of rules either can or should
constrict choice in this way. Yet Professor Schauer argues that this aversion to
formalism should be rethought: At times language both can and should re?
strict decisionmakers. Consequently, the term "formalistic" should not be used
as a blanket condemnation of a decisionmaking process; instead the debate
regarding decision according to rules should be confronted on its own terms.
With accelerating frequency, legal decisions and theories are con?

demned as "formalist" or "formalistic." But what is formalism, and what
is so bad about it? Even a cursory look at the literature reveals scant
agreement on what it is for decisions in law, or perspectives on law, to be
* Professorof Law,
University of Michigan. I am grateful to audiences at Brooklyn Law School,
Cornell Law School, DePaul University College of Law, Duke University School of Law, Indiana
University at Bloomington School of Law, New York University School of Law, and the American
Political Science Association for helping me to clarify some of my good ideas and jettison some of my
bad ones. I am also indebtedto Alex Aleinikoff, Bruce Frier, Leo Katz, James Krier, William Miller,
and Richard Pildes for commenting on earlier versions of this article with just the right blend of
hostility and sympathy.
509

510 The Yale Law Journal [Vol. 97: 509
formalistic, except that whatever formalism is, it is not good.1 Few judges
or scholars would describe themselves as formalists, for a congratulatory
use of the word "formal" seems almost a linguistic error. Indeed, the pejo-
rative connotations of the word "formalism," in concert with the lack of
agreement on the word's descriptive content, make it tempting to conclude
that "formalist" is the adjective used to describe any judicial decision, style
of legal thinking, or legal theory with which the user of the term
disagrees.
Yet this temptation should be resisted. There does seem to be descrip?
tive content in the notion of formalism, even if there are widely divergent
uses of the term. At the heart of the word "formalism," in many of its
numerous uses, lies the concept of decisionmaking according to rule. For?
malism is the way in which rules achieve their "ruleness" precisely by
doing what is supposed to be the failing of formalism: screening off from a
decisionmaker factors that a sensitive decisionmaker would otherwise take
into account.
Moreover, it appears that this screening off takes place
largely through the force of the language in which rules are written. Thus
the tasks performed by rules are tasks for which the primary tool is the
specific linguistic formulation of a rule. As a result, insofar as formalism
is frequently condemned as excessive reliance on the language of a rule, it
is the very idea of decisionmaking by rule that is being condemned, either
as a description of how decisionmaking can take place or as a prescription
for how decisionmaking should take place.
Once we disentangle and examine the various strands of formalism and
recognize the way in which formalism, rules, and language are concep-
tually intertwined, it turns out that there is something, indeed much, to be
said for decision according to rule?and therefore for formalism. I do not
argue that formalism is always good or that legal systems ought often or
1. See, e.g., H.L.A. Hart, The Concept of Law 124-30 (1961) (formalism as refusal to ac?
knowledgenecessityof choice in penumbralarea of rules); M. Horwitz, The Transformation of
American Law 254 (1977) (formalism as refusal to recognize instrumental functions of law); K.
Llewellyn, Jurisprudence: Realism in Theory and Practice 183-88 (1962) (formalism as
excessive reliance on canonicallywritten language of rules); R. Unger, The Critical Legal Stud?
ies Movement 1-2 (1986) (formalism as constrainedand comparativelyapolitical decisionmaking);
Kennedy,Legal Formality, 2 J. Legal Stud. 351, 355 (1973) (formalismas view that rule applica?
tion is mechanical and that mechanical rule application is just); Strauss, Formal and Functional
Approaches to Separation-of-Powers Questions?A Foolish Inconsistency?, 72 Cornell L. Rev.
488, 489 (1987) (formalism as refusal to acknowledge practical consequences of judicial decisions);
Tushnet, Anti-Formalismin Recent Constitutional Theory,83 Mich. L. Rev. 1502, 1506-07 (1985)
(formalism as artificial narrowing of range of interpretivechoices).
One can avoid the confusion of multiple usage by simply stipulating a meaning for the term "for?
malism." See, e.g., Posner, Legal Formalism, Legal Realism, and the Interpretation of Statutes and
the Constitution, 37 Case W. Res. L. Rev. 179, 181-82 (1986). This tack, however, evades most of
the interesting problems. Having stipulated that "formalism" means deductive logical reasoning,
Judge Posner proceeds easily to the conclusion that formalist reasoning has no application to the
interpretationof canonical texts. That conclusion, however, follows, if at all, only from the narrow-
ness of the stipulated definition. By not stipulating a meaning in advance of the analysis, I intend to
focus on a broaderrange of issues. In the process, I will explore the way in which deduction,even in
Posner's sense, may be related to the interpretationof canonical texts. See infra note 48.

1988] Formalism 511
even ever be formalistic. Nevertheless, I do want to urge a rethinking of

the contemporary aversion to formalism. For even if what can be said for
formalism is not in the end persuasive, the issues should be before us for
inspection, rather than blocked by a discourse of epithets.
I. Formalism as the Denial of Choice
A. Choice Within Norms
Few decisions are charged with formalism as often as Lochner v. New

York.2 But what makes Justice Peckham's majority opinion in Lochner
formalistic? Surely it is not just that the Court protected an unrestricted
privilege of labor contracting against the first stirrings of the welfare state.
For the Court to make such a political decision under the rubric of broad
constitutional clauses like "liberty" is a far cry from what seems to be
meant when decisions are criticized as being formal. To the extent that
the charge of formalism suggests narrowness, Lochner is hardly a candi-
date. We criticize Lochner not for being narrow, but for being excessively
broad.
Although Lochner is criticized for the length of its reach, a closer look
reveals that it is not the result that is condemned as formalistic but rather
the justification for that result. The formalism in Lochner inheres in its
denial of the political, moral, social, and economic choices involved in the
decision, and indeed in its denial that there was any choice at all. Justice
Peckham simply announced that "[t]he general right to make a contract in
relation is part of the liberty of the individual
to his business protected by
the Fourteenth Amendment"3 and that "[t]he right to purchase or to sell
labor is part of the liberty protected by this amendment."4 To these pro-
nouncements he added the confident statement that "[o]f course the liberty
of contract relating to labor includes both parties to it."5
Justice Peckham's language suggests that he is explaining a precise
statutory scheme rather than expounding on one word in the Constitution.
It is precisely for this reason that his opinion draws criticism. We con?
demn Lochner as formalistic not because it involves a choice, but because
2. 198 U.S. 45 (1905). For condemnationsof Lochner (and the era of which it is taken to be
archetypal) as formalistic, see Gordon, Critical Legal Histories, 36 Stan. L. Rev. 57, 99 (1984);
Peller, The Metaphysicsof American Law, 73 Calif. L. Rev. 1151, 1193, 1200-01 (1985); Seidman,
Public Principle and Private Choice: The Uneasy Case for a Boundary Maintenance Theory of
Constitutional Law, 96 Yale L.J. 1006, 1006-07 (1987); Developments in the Law?Immigration
Policy and the Rights of Aliens, 96 Harv. L. Rev. 1286, 1292 (1983); Note, The Constitutionalityof
Rent Control Restrictions on Property Owners*Dominion Interests, 100 Harv. L. Rev. 1067, 1077
(1987); Note, Formalism, Legal Realism, and Constitutionally Protected Privacy Under the Fourth
and Fifth Amendments,90 Harv. L. Rev. 945, 951 (1977); Powers, Book Review, 1985 Duke L.J.
221, 232; Rotenberg,Politics, Personality and Judging: The Lessons of Brandeis and Frankfurter
on Judicial Restraint (Book Review), 83 Colum. L. Rev. 1863, 1875 n.60 (1983).
3. 198 U.S. at 53.
4. Id.
5. Id. at 56.

it attempts to describe this choice as compulsion.6 What strikes us clearly

as a political or social or moral or economic choice is described in Lochner
as definitionally incorporated within the meaning of a broad term. Thus,
choice is masked by the language of linguistic inexorability.
When I say that pelicans are birds, the truth of the statement follows
inexorably from the meaning of the term "bird." If someone disagrees, or
points at a living, breathing, flying pelican and says "That is not a bird,"
she simply does not know what the word "bird" means.7 We criticize
Lochner as formalistic because it treats
"liberty" (or the words
the word
"life, liberty, or property, without due process of law") as being like the
word "bird" and the privilege of contracting as being like a pelican, i.e.,
subsumed in the broader category. According to the reasoning in Lochner,
if you don't know that contracting for labor without governmental control
is an example of liberty, then you just don't know what the word "lib?
erty" means.
Lochner is condemned as formalistic precisely because the analogy be?
tween pelicans (as birds) and unrestricted contracting (as liberty) fails.
One can understand much about the concept of liberty and about the word
"liberty" and yet still deny that they include the privilege of uncon-
strained labor contracting.8 Thus, a decisionmaker who knows or should
6. This was noted by Holmes in his now-famousobservation,"General propositionsdo not decide

concrete cases." Id. at 76 (Holmes, J., dissenting).
7. Of course when I use the term "inexorable,"I do not mean that the world and our language
could not have been otherwise;the word "bird"could have referredto frogs instead of pelicans, or to
only puffins, robins, and sparrows, but not pelicans, ostriches,and condors. Definitions are contingent
and subject to change, and thereforethe word "bird"might yet come to be the word that speakers of
English use to refer to frogs, or only to small and not to large birds. Yet although there remains a
possibilitythat the word "bird"will come to mean these things, this is only a possible world?it is not
our world. In our world, the exclusion of frogs and the inclusion of large birds is definitionallypart of
the meaning of the word "bird."As I argue below, see infra notes 56-57 and accompanyingtext, the
contingencyof definition hardly entails the view that it is within the provinceof any one actor, legal
or otherwise, to change it. Neither you nor I have the power to make it proper to use the word "bird"
to refer to a frog, even though the word "bird" could in another world be used to refer to frogs.
8. The extent to which this is true for morally and politically loaded words such as "liberty" is
likely to vary with time, place, and culture. Take, for example, the transformationof the "honor
codes" at various venerable universities. These codes were phrased in quite general terms at their
inception in the 18th and 19th centuries because these schools contained homogeneousstudent bodies
who shared a common conceptionof the type of conduct definitionally incorporatedwithin the word
"honor."If a person thought that purchasinga term paper from a professionalterm paper service was
consistent with being honorable, then that person simply did not know what "honor" meant. As
values have changedand as student bodies have become less homogeneous,however, shared definitions
of terms such as "honor"have brokendown. Some people now do think that buying a term paper can
be honorable,and this breakdownin shared meaning has caused general referencesto "honor"to be
displacedin such codes by more detailed rules. There may now be little shared agreementabout what
the precept "be honorable"requires, but there is considerableagreementabout what the rule "do not
purchase a term paper" requires.
Thus, the criticism of Lochner and its ilk as "formalistic"in the sense discussed in the text is
ambiguous. The critic could mean that the term we now take to be susceptible to debate was not as
debatableat the time of the relevant decision. But this would hardly explain the pejorative,unless we
want to condemn an entire era and the conceptual and linguistic apparatus that reflected its under?
standings. The alternative is that the term "formalism"charges that there was at the time room for
debate about the applicationof the general term to the particularcase, but the relevant decisionmakers
either did not recognize that fact (perhaps because they refused to look outside their own socioeco-

1988] Formalism 513
know that such a choice is open, but treats the choice as no more available
than the choice to treat a pelican as other than a bird, is charged with
formalism for treating as definitionally inexorable that which involves
nondefinitional, substantive choices.9
Lochner is merely one example in which a false assertion of inexorabil-
ity is decried as formalistic. Much contemporary criticism of Blackstone,
Langdell, and others of their persuasion attacks their jurisprudence on
similar grounds.10 They stand accused of presenting contestable applica?
tions of general terms as definitionally incorporated within the meaning of
the general term. It is important, however, to understand the relationship
between the linguistic and the ontological questions for those of Black-
stone's vision. Blackstone's view that certain abstract definitionally terms
incorporate a wide range of specific results is tied intimately to his percep?
tion of a hard and suprahuman reality behind these general terms. If the
word "property," for example, actually describes some underlying and
noncontingent reality, then it follows easily specific embodi-
that certain
ments are necessarily part of that reality, just as pelicans are part of the
underlying reality that is the universe of birds. These instantiations might
still followeven if the general term is not a natural kind whose existence
and demarcation is beyond the control of human actors. There is nothing
natural or noncontingent about the term "basketball," but it is neverthe?
less an error in this culture at this time to apply that word to a group of
people hitting small hard balls with one of a collection of fourteen differ?
ent sticks. Still, linguistic clarity and rigidity are both facilitated insofar as
the words track the natural kinds of the world. To the extent that Black?
stone and others believed that
categories like liberty, property, and con?
tract were natural kinds ratherthan human artifacts, they were less likely
to perceive the choices we would now not think to deny. When one be?
lieves that a general term reflects a deep reality beyond the power of
human actors, the view that certain particulars are necessarily part of that
reality follows with special ease.
Thus, one view of the vice of formalism takes that vice to be one of
deception, either of oneself or of others. To disguise a choice in the lan-
nomic and political class) or intentionally chose to hide it.

9. Formalismmay be more broadlyviewed as extending to any justificationthat treats as inexora-
ble a choice that is not. In this broader sense, the claimed inexorability might come from something
other than rule formulations.To mask, for example, a political, moral, or social choice in the lan?
guage of "original intent" when original intent in fact does not provide a uniquely correctanswer to
the issue might be consideredformalistic in the same way that masking a political, moral, or social
choice in the language of the meaning of a rule is consideredformalisticwhen that language does not
provide a uniquely correct answer. Similarly, masking choice in the language of mathematicaleco?
nomic derivation or in the language of a unique solution to some "balance" might be considered
formalisticif these methods are in fact comparativelyindeterminate.
10. See, e.g., Grey, LangdelVs Orthodoxy,45 U. Pitt. L. Rev. 1 (1983); Hart, Positivism and
the Separation ofLaw and Morals, 71 Harv. L. Rev. 593, 610 (1958); Kennedy, The Structure of
Blackstone''sCommentaries,28 Buffalo L. Rev. 205 (1979); Lyons, Legal Formalism and Instru-
mentalism?A Pathological Study, 66 Cornell L. Rev. 949, 950 (1981).

guage of definitional inexorability obscures that choice and thus obstructs

questions of how it was made and whether it could have been made differ?
ently. Use of the word in this sense hinges on the existence
"formalism" of
a term (or phrase, sentence, or paragraph11) whose contested application
generates the choice. Some terms, like "liberty" and "equality," are per-
vasively indeterminate. It is not that such terms have no content whatso-

ever; it is that every application, every concretization, every instantiation
requires the addition of supplementary premises to apply the general term
to specific cases.12 Therefore, any application of that term that denies the
choice made among various eligible supplementary premises is formalistic
in this sense.13
More commonly, however, the indeterminacy to be filled by a deci-

sionmaker's choice is not pervasive throughout the range of applications of
a term. Instead, the indeterminacy is encountered only at the edges of a
term's meaning. As H.L.A. Hart tells us, legal terms possess a core of
settled meaning and a penumbra of debatable meaning.14 For Hart, for?
malism derives from the denial of choice in the penumbra of meaning,
where applying the term in question is optional. Thus, Hart conceives of
formalism as the unwillingness to acknowledge in cases of doubtful appli?
cation, such as the question of whether a bicycle is a vehicle for purposes
of the prohibition on vehicles in the park, that choices must be made that
go far beyond merely ascertaining the meaning of a word.
Hart's conception of formalism15 is closely aligned with that under-

girding those who criticize both Blackstone and Lochner.16 Hart's formal-
ist takes the penumbra to be as clear as the core, while the Lochner for-
malist takes the general term to be as determinate as the specific. Both
deny the extent of actual indeterminacy, and thus neither admits that the
application of the norm involves a choice not determined by the words of
the norm alone.
11. See infra note 85.

12. For a discussion of the often-ignorednecessity of relying on such supplementarypremises in
the application of the term "equality," see Westen, The Empty Idea of Equality, 95 Harv. L. Rev.
537 (1982). I disagree, however, with Westen's argument that the necessity of adding those supple?
mentary premises to give the primary term meaning renders the primary term superfluous. Because a
term is not self-standingdoes not mean that it serves no purpose, even if it needs external assistancein
order to serve that purpose.
13. On the choices necessitated(but often denied) by general terms, see Cohen, Transcendental
Nonsense and the Functional Approach, 35 Colum. L. Rev. 809 (1935); Dewey, Logical Method
and Law, 10 Cornell L.Q. 17 (1924); Horwitz, Santa Clara Revisited: The Developmentof Corpo?
rate Theory, 88 W. Va. L. Rev. 173, 175-76 (1985); Singer, The Player and the Cards: Nihilism
and Legal Theory, 94 Yale L.J. 1, 9-25 (1984).
14. Hart, supra note 10, at 608-12; see also H.L.A. Hart, supra note 1, at 121-50.
15. H.L.A. Hart, supra note 1, at 121-50; Hart, supra note 10, at 608-12.
16. See, e.g., Gordon, supra note 2; Grey, supra note 10; Peller, supra note 2.

1988] Formalism 515
B. Choice Among Norms
Implicit in Hart's conception of formalism is the view that in the core,

unlike in the penumbra, legal answers are often tolerably determinate.
Even if this is true, and I will examine this claim presently, the possibility
remains that a decisionmaker has a choice of whether or not to follow a
seemingly applicable norm even in its core of meaning. The question in
this case is not whether a bus is a vehicle, or even whether the core of the
rule excludes buses
from the park, but whether the rule excluding vehicles
must be applied in this case. At times a decisionmaker may have a choice
whether to apply the clear and specifically applicable norm. In such cases
we can imagine a decisionmaker having and making a choice but denying
that a choice was in any way part of the process. Thus, a variant on the
variety of formalism just discussed sees formalism as involving not denial
of the existence of choices within norms, but denial that there are fre?
quently choices about whether to apply even the clear norms.
As an example of this type of formalism, consider the unreported and
widely unknown case of Hunter v. Norman.17 Hunter, an incumbent
state senator in Vermont
seeking re-election, filed his nominating petition
in the Windsor County Clerk's office on July 21, 1986 at 5:03 p.m. In
doing so he missed by three minutes the petition deadline set by title 17,
section 2356, of the Laws of Vermont.18 The statute provides, in its en-
tirety, that "Primary petitions shall be filed not later than 5:00 p.m. on
the third Monday of July preceding the primary election prescribed by
section of this title, and not later than 5:00 p.m. of the forty-second
2351
day prior the day of a special
to primary election."19 The Windsor
County Clerk, Jane Norman, duly enforced the statute by refusing to ac?
cept Hunter's petition, observing that "I have no intention of breaking the
law, not for Jesus Christ himself."20 Hunter's name, consequently, was to
be withheld from appearing on the September Democratic primary elec?
tion ballot.
Hunter, not surprisingly, took his disappointment to the courthouse and
filed an action in equity against Norman for extraordinary relief.21 He
asked that the court order her to accept his petition and to ensure that his
name would appear on the primary ballot. At the hearing, Hunter alleged
that he had called the clerk's office earlier on the date in question and
been told that he was required to deliver the petition in person because of
17. No. S197-86-WrC (Vt. July 28, 1986). The following account of the case is drawn from
Judge Cheever's brief opinion, the pleadings, news accounts in the Rutland Herald of July 22, 23,
24 and 26, 1986, and a conversationwith Marilyn Signe Skoglund,Assistant Attorney General in the
Office of the Attorney General, State of Vermont.
18. Vt. Stat. Ann. tit. 17, ? 2356 (1982).
19. Id.
20. Rutland Herald, July 23, 1986, at 5, col. 4.
21. The petition is unclear as to whether Hunter was seeking the extraordinarylegal remedy of
mandamus or a mandatory injunction in equity.

the necessity of signing forms consenting to his nomination. In fact, these

consent forms were not due until a later date. Hunter claimed that had he
not been led to appear in person by receiving this erroneous advice, the
petition would
have been filed earlier in the day. He argued that in light
of the erroneous information given to Hunter by the Clerk's office, the
clerk (and the state) were estopped from relying on the statutory deadline.
In support of this proposition, Hunter offered Ryshpan v. Cashman22 in
which the Vermont Supreme Court, on similar facts, held that because
"reliance on erroneous actions on behalf of the State has put ... its citi?
zens in inescapable conflict with the literal terms of one of the time re?
quirements instituted by that same sovereignty . . . [t]he statutory time
schedule must . . . as a matter of equity . . . yield."23
Ultimately, Hunter prevailed, and it appears that Ryshpan v. Cashman

saved the day?or at least saved Hunter's day. Ryshpan therefore seems
to have operated as an escape route from the rigors of the statute. Sup?
pose, however, that everything in Hunter's case had been the same, in?
cluding the existence of Ryshpan, but that the judge had ruled against
Hunter solely on the basis of the statutory language. Had this hardly un-
realistic alternative occurred, it would seem but a small step from the
brand of formalism discussed above to a formalist characterization of this
hypothetical decision. As long as Ryshpan exists, the judge has a choice
whether to follow the letter of the statute or instead to employ the escape
route. To make this choice and merely cite the statute as indicating the
absence of choice would therefore deny the reality of the choice that was
made. The crux of the matter is that this choice was present as long as
Ryshpan existed, whether the judge followed that case or not. The charge
of formalism in such a case would be but a variation of formalism as the
concealment of choice: Instead of a choice within a norm, as with either
pervasively indeterminate language or language containing penumbras of
uncertainty surrounding a core of settled meaning, here the choice is be?
tween two different norms.
This variation on Ryshpan reveals the reasons we condemn the mask?
ing of choice. When the statute and Ryshpan coexist, neither determines
which will prevail. Thus, the choice of the escape route represented by
Ryshpan over the result indicated by the statute, or vice versa, necessarily
would be made on the basis of factors external to both. These factors
might include
the moral, political, or physical attractiveness of the parties;
the particular facts of the case; the judge's own views about deadlines; the
judge's own views about statutes; the judge's own views about the Ver?
mont Supreme Court; the judge's own views about clerks of courts; and so
on. Yet were any of these factors to cause a particular judge to decide that
22. 132 Vt. 628, 326 A.2d 169 (1974).

23. Id. at 630-31, 326 A.2d at 171.

1988] Formalism 517
the statute should prevail, mere citation of the statute as inexorably dictat-
ing the result would conceal from the litigants and from society the actual
determinative factors. Insofar as we expect the reasons for a decision to be
open (and that, after all, is usually the reason judges write
for inspection
opinions),24 failure
to acknowledge that a choice was made can be criti?
cized because knowing how the choice was made helps to make legitimate
the products of the system.
C. Is There Always a Choice?
Ryshpan v. Cashman is a trifle obscure, but it is hardly unique. Con?

sider the number of Ryshpan equivalents that allow decisionmakers to
avoid the specific mandates of a particular rule. A decisionmaker may de?
termine that the literal language of a rule does not serve that rule's origi?
nal intent, as the Supreme Court has interpreted the Civil Rights Act of
1964,25 the contracts clause of the Constitution,26 and the Eleventh
Amendment.27 Or a decisionmaker may apply the "mischief rule" or its
variants to determinethat a literal application of the rule would not serve
the rule's purpose.2* Or a decisionmaker may apply a more general rule
that denies relief to a claimant entitled to relief under the most locally
applicable rule;29 for example, she might apply the equitable principle of
24. Although it is generally accepted that judges should write opinions explaining their actual
reasons for decision, see, e.g., Shapiro, In Defense of Judicial Candor, 100 Harv. L. Rev. 731
(1987), some scholars have suggested that there may be reasons to avoid a candid explanation of the
reasoning process.See sources cited id. at 731 n.4. Thus an opinion might be equated with a statute,
whose message legitimatelymay depart from a reflectionof the processthat generatedit. While recog?
nizing that reasons going to the symbolic, guiding, and persuasive function of opinions may urge
against candid explanation of the decision process, I address here only opinions in which honesty is
deemed appropriate.
25. See, e.g., California Fed. Sav. & Loan Ass'n v. Guerra, 107 S. Ct. 683, 691 (1987); United
Steelworkersv. Weber, 443 U.S. 193, 201 (1979).
26. See, e.g., Keystone Bituminous Coal Ass'n v. DeBenedictis, 107 S. Ct. 1232, 1251 (1987).
27. See, e.g., Monaco v. Mississippi, 292 U.S. 313, 329-30 (1934); Hans v. Louisiana, 134 U.S.
1, 10-11 (1890).
28. Heydon's Case, 76 Eng. Rep. 637 (Ex. 1584), phrases the rule as deriving from original
legislative intent. See infra note 68. However, limitation of the purpose of a rule to the intent of the
legislature that passed it unnecessarilyrestrictsthe meaning of the term "purpose." Purpose gleaned
from the words of a rule itself should not be confusedwith the psychologicalintentions of the drafters.
Consider a rule which specifically excludes from a park children, radios, musical instruments, dogs
but not eats, and cars and trucks but not bicycles. One might conclude from reading this rule that its
purpose is to prevent noise. Even if the drafters of the rule intended to promote safety rather than
prevent noise, their psychologicalintentions would not negate this reading of purpose from the rule's
words themselves,any more than a person, having said "stop," could deny the import of that phrase
because she in fact meant "go."
29. See Singer, supra note 13, at 17-18. The most "locally" applicable rule (or statute) is that
which most narrowly pertains to the situation at hand. "Dogs should be leashed" is, in a case involv?
ing a dog, more locally applicable than "animals should be restrained." Similarly, "beneficiaries
named by the testator are to inherit accordingto the will" is more locally applicable than "no person
should benefit by his own wrong." The idea of local applicability distinguishes the rules in each of
these pairs, for in each pair both rules might apply to the same situation. Local applicabilitycaptures
our intuition that a more specifically applicable rule is somehow more applicable than a less specifi?
cally applicable, but still applicable, rule.

unclean hands or laches,80 the legal principle of in pari delicto,*1 or the

civil law principle of abuse of right.82 Any reader of this article could
easily add to this list.38
Yet, what if none of these established routes were available in a partic?
ular case?would a judge then be forced to apply the specifically applica?
ble rule? To answer this question, let us examine another variation on
Hunter v. Norman. Suppose that Ryshpan v. Cashman did not exist, but
that everything else about the facts and the applicable law in Hunter re-
mained the same. What choices, if any, would be open to the judge? The
judge could, of course, simply hold that the statute applied and rule
against Hunter. But must he? Could the judge instead "create" Ryshpan
by concluding that Hunter should win because he was misled by the
clerk's office?
This option of creating Ryshpan does not seem inconsistent with the
way the American legal system operates. Despite the lack of any specific
statute or case authorizing such a result, allowing Hunter to win because
he was misled would raise no eyebrows in American legal circles. No one
would call for an investigation of the judge's competence, as someone
might had the judge ruled for Hunter because Hunter was a Capricorn
and Norman a Sagittarius. If the creation of such an escape route would
be consistent with American judicial traditions, then the judge can be seen
to have had a choice between deciding for Hunter and deciding for Nor?
man even without Ryshpan. Thus a judge who ruled against Hunter on
the basis of the statute would
be denying the extent to which there was
still a choice to create Ryshpan and thereby rule for Hunter.
Of course, a judge who decided to "create" Ryshpan would probably
not simply assert that Hunter should win because he relied on erroneous
information from a state official. Rather, the judge would justify this con?
clusion by reference to general principles that lurk in various corners of
the legal system. For example, the judge might say that, as a general prin?
ciple, parties are estopped from relying on laws whose contents they have
30. See, e.g., Brenner v. Smullian, 84 So. 2d 44 (Fla. 1955) (unclean hands); Gorham v. Sayles,
23 R.I. 449, 50 A. 848 (1901) (laches).
31. See, e.g., Rozell v. Vansyckle, 11 Wash. 79, 39 Pac. 270 (1895). Riggs v. Palmer, 115 N.Y.
506, 22 N.E. 188 (1889), made famous in R. Dworkin, Law's Empire 15-20 (1986) [hereinafter
Law's Empire] and R. Dworkin, Taking Rights Seriously 23 (1977) [hereinafter Taking
Rights Seriously], presents a similar issue. Riggs is significant because the most locally applicable
legal rule, the relevantstatute of wills, would allow the murderingheir to inherit. Only the imposition
of the less locally applicable general principle that no person should profit from his own wrong al?
lowed the court to avoid the result indicated by the most directly applicable legal norm. From the
perspectiveof the result dictated by the most immediately applicable legal rule, Riggs is not a hard
case, but an easy one. UnderstandingDworkin's enterprise requires an understandingof his attempt
to explain the ways in which the result "easily" dictatedby the most locally applicable rule frequently
yields to less locally applicable legal and nonlegal norms. See Schauer, TheJurisprudence of Reasons
(Book Review), 85 Mich. L. Rev. 847 (1987).
32. See generally Gutteridge,Abuse of Rights, 5 Cambridge L.J. 22 (1935) (discussing possibil?
ity of incorporatingprinciple forbiddingexercise of legal rights for purposes of malevolence).
33. See Singer, supra note 13, at 17-18.

1988] Formalism 519
misstated to the disadvantage of another; a decision against the clerk

would be merely a specific instance of the application of that general prin?
ciple. Or, the judge might cite other particular principles, such as the
principle of reliance in securities law, and analogize this case to those.34
Under either analysis the judge would attempt to ground the new princi?
ple in some already existing principle.
On the basis of these variations, we can distinguish three possible mod?
els of escape route availability. Under one model, the existing escape
routes in the system represent an incomplete list of principles to amelio-
rate the rigidity of rules, and the judge may add to this list where amelio-
ration is indicated but no applicable ameliorative principle exists. In such
instances, the judge might discuss justice or fairness or some other general
value and explain why this value supports the creation of a principle like
that in Ryshpan v. Cashman. The implicit ideal of this system is the
availability of an ameliorative principle whenever the circumstances de?
mand it. Thus the judge who creates a new ameliorative principle on an
appropriate occasion furthers the goals of this system.
Alternatively, we could develop a model of a system in which there is

already a more or less complete stock of ameliorative principles. In such a
system, a judge would always have
escape some route available if all the
circumstances indicated that the applicable norm was not the best result to
be reached in that case. If Ryshpan itself did not exist, the judge would be
able to pick other extant ameliorative principles that would get Hunter's
name on the ballot.
Both the first model, which resembles Dworkin's account of the law,35
and the second, which borrows from Llewellyn's,36 acknowledge the per-
vasiveness of judicial choice in their recognition of the judge's opportunity
(or perhaps even obligation) to avoid the arguably unjust consequences of
mechanical application of the most directly applicable legal rule. If either
of these models is an accurate rendition of some legal system, then a deci?
sionmaker within such a system who simply applies the most
directly ap?
plicable legal rule without further thought or explanation either denies
herself a choice that the system permitted or required, or denies to others
an explanation of why she chose not to use the escape routes permitted by
the system. This failure to explain the choice to apply the most locally
applicable rule is simply a variation on the more egregious forms of for?
malism as denial of choice.37
34. Use of precedentis not as simple as I make it out to be here, but these subtleties of preceden?
tial reasoning need not detain us here. For a discussion of precedent, see Schauer, Precedent, 39
Stan. L. Rev. 571 (1987).
35. See Law's Empire, supra note 31; Taking Rights Seriously, supra note 31. See also
discussion of Dworkin, supra note 31.
36. See especially Llewellyn, Remarks on the Theory of Appellate Decision and the Rules or
Canons About How Statutes Are To Be Construed, 3 Vand. L. Rev. 395 (1950).
37. Whether a system allows judges to create norms of rule avoidancewhere none exist, whether

These two models?one allowing the creation of rule-avoiding norms,

and the other presenting a complete list of such norms for use38?must be
contrasted with a third model. Under
this third model, the stock of extant
rule-avoiding norms
is not temporarily incomplete but completable, as in
the first model, nor is it complete, as in the second. Instead, it is both
incomplete and closed. A decisionmaker will therefore be confronted with
situations in which the immediately applicable rule generates a result the
decisionmaker wishes to avoid but for which the system neither contains
an escape route nor permits one to be created. Under this model, a judge
who followed the rule?rather than the course she otherwise would have
taken on the basis of all relevant factors?would not have acted formalis-
tically in the sense now under discussion. Where there was no choice, a
decisionmaker following the mandates of
directly the most
applicable
norm could not be accused of having a choice but denying its existence.
If we can imagine a model in which a rule-avoiding norm is both non-
existent and precluded in some instances, then we can also imagine a
model in which no rule-avoiding norms exist at all. In such a system, a
decisionmaker would be expected simply to decide according to the rule
when there was a rule dealing specifically with the situation. Because
there was no choice to be made, the decisionmaker could not be charged
with masking a choice.
This third model presents the conceptual possibility of a different type
of formalism than that which has been the focus of this section. In this
third model, the charge of "formalism" would possess a different signifi?
cance than in the other two models, for the decisionmaker accused of be?
ing formalistic might not be denying a choice made in the decisionmaking
process, but might never have had a choice at all. To investigate the possi?
bility of this type of formalism we must determine whether a system can
truly foreclose choices from the decisionmaker. It is to this issue that I
now turn.
II. Formalism as the Limitation of Choice
A. Can Language Constrain?
Each of my variations on Hunter v. Norman presupposed that the

judge reached a conclusion that was not influenced by the language of the
judges in fact create such norms, and whether a sufficient stock of rule-avoidingnorms exists such that
judges need only apply them are all unavoidablyempirical questions. See Kennedy, Toward a Critical
PhenomenologyofJudging, 36 J. Legal Educ. 518, 547-48, 562 (1986); see also Trubek, Where the
Action Is: Critical Legal Studies and Empiricism, 36 Stan. L. Rev. 575 (1984). There is no reason,
of course, to presume that the answers to these empirical questions will remain consistent across all
decisional domains within a legal system. For example, no logical necessity dictates that the stock of
rule avoidancenorms applicable to administrativedeterminationof individual social security claims be
identical to that applicable to Supreme Court adjudicationof constitutionalquestions.
38. There need not be any conceptual inconsistencybetween the two models. The second can be
conceivedof as the end product of the first.

1988] Formalism 521
rule.39 This rule-independent conclusion presents the possibility that the

results required by the most
locally applicable may diverge statute from
the result the judge considers to be the optimal result for this case in light
of a range of factors wider than that specifically mandated by the statute.
In cases of such divergence between a judge's unconstrained judgment and
the result indicated by the most locally applicable statute, a rigid require?
ment that the decision follow the statutory language would limit the
choices open to the judge.40
Insofar as rigid adherence to the most locally applicable statute is re?
quired, either by the norms governing a decisional domain or by a judge's
understanding of her role, a judge following that requirement would not
be formalistic in the sense discussed in the previous section. Nevertheless,
legal theorists condemn this type of decisionmaking as formalistic because
it requires that a decisionmaker allow her best judgment about what
should be done in this situation to yield to the dictates of a mere rule. In
particular, it is the language of the rule that is perceived as binding the
decisionmaker;41 critics therefore condemn this decisionmaking process as
formalistic because it appears to be a commitment to constraint by mere
words on a printed page, words chosen and perpetuated without consider?
ation of the exact situation now at hand. Formalism in this sense is not
the denial by the judge, as above, but the denial of choice to the
of choice
judge. To be formalistic, it is said, is to be enslaved by mere marks on a
printed page.42
39. I need not consider here which factors the judge actually used to reach a conclusion, for I am
not trying to catalog the considerationscomprisingan ideal decisionmaker'stotally particularizeddeci?
sionmaking process. Instead, I seek merely to distinguish the concept of a complete array of factors
that any particularizingdecisionmakerwould take into account, regardlessof the source of the partic-
ularizing norms, from the more limited array of factors available to a decisionmakerinhibited by
rules.
40. I assume here a distinctionbetween internal and external constraint.A host of factorsdefining
what I am and how I got that way constrain me from appearing unclothed in a football game at
Michigan Stadium. Some of these are internal constraints?the factors that shape my very existence.
These internal constraints may be psychological,ideological, or economic, but all shape what I am
internally up to the moment of decision to appear clothed rather than naked at the football game.
Even if I could overcomethese internal constraints,however, external ones, such as social disapproval
and a formal rule against such behavior, still might deter me from that action. Similarly, all sorts of
internal factors influence the decision a judge might reach about the optimal result in this case. But
these influences are distinguishablefrom external constraints, such as rules, that come from outside
the judge's personal determinationof what should be done.
Rules are only one possible example of external constraints. A decisionmakeralso may believe
herself to be externally constrainedby statutorypurpose. As I will demonstratebelow, however, see
infra text accompanyingnotes 77-79, statutory purpose is an external constraint when, and only
when, it operates as a rule in the sense central to my argument. That is, purpose is an external
constraintonly when someformulation of that purpose, on paper or in the mind, operates in substan?
tially the same way that a canonically formulated rule operates.
41. I explore this issue in depth below. See infra Section II-C.
42. For an example of this common use of the term "formalism,"see Levinson, What Do Lawyers
Know (and What Do TheyDo with Their Knowledge)?Commentson Schauer and Moore, 58 S. Cal.
L. Rev. 441, 445 (1985) (erroneouslyconcludingthat Schauer "is much too sophisticateda theorist to
endorse . . . linguistic formalism").This usage of the term "formalism"parallels that of other disci?
plines. See, e.g., Michaels, Against Formalism: The Autonomous Text in Legal and Literary Inter-

Formalism as the linguistic limitation of choice can be illustrated in a

number of ways. Think of the judge who evicts the destitute widow and
her family on Christmas Eve because "the law" permits no other result.
Consider the classic, fictional case of i?. v. 0jibway,4S in which the judge
determines that a pony with a down pillow on its back is a small bird
because it literally fits a statutory definition of a small bird as a two-
legged animal covered with feathers.44 And recall Justice White's dissent
in Bowsher v. Synar,4* in which he accuses the majority of being "formal?
istic" for taking its narrow reading of article II to be more important than
the practical consequences of striking down attempts to deal with the defi?
cit problem.46
These cases exemplify a decisionmaking process that, by distinguishing
the literal mandates of the most locally applicable legal norm from some
arguably better result reachable by considering a wider range of factors,
reinforces the systemic isolation, or closure, of the legal system.47 Those
who condemn such an outlook as formalistic the perception
criticize of law
as a closed system, within which judgments are mechanically deducible
from the language of legal rules.48 Note that this description of formalism
pretation, 1 Poetics Today 23 (1979).

43. Pomerantz & Breslin, Judicial Humour?Construction of a Statute, 8 Crim. L.Q. 137
(1966).
44. Note, of course, that four-legged animals have two legs?and more. Id. at 138.
45. 478 U.S. 714, 106 S. Ct. 3181, 3205 (1986) (White, J., dissenting).
46. The plausibility of Justice White's dissent indicates that the majority opinion may also have
been formalisticin the first sense consideredin this article, see supra Section I, for it suggests that the
majority had a choice. Yet by phrasing the opinion largely in terms of the clear mandate of the
Constitution,the majoritydenied the existence of that choice and thus denied its audience the benefit
of knowing how that choice was made. See Strauss, supra note 1.
47. On the relationshipbetween the idea of systemic isolation and the more familiar terminology
of legal positivism and its opponents, see infra note 81.
48. See, e.g., M. Horwitz, supra note 1, at 250-51. After defining formalism as syllogistic de?
duction,Judge Posner concludesthat formalismis inapplicableto statutoryrules (rules with a canoni?
cal embodiment)because the decision to take these rules literally is itself a choice. Posner, supra note
1. Posner stumbles, however, in taking this preliminarychoice to distinguish rule interpretationfrom
common law adjudication.He offers the following as an instance of a common law deduction:"So if
an enforceablecontractis a promise supported by consideration,and A's promise to B was supported
by consideration,the promise is a contract."Id. at 182. He then contrasts that example with the
following requirement:"[OJnemust be at least thirty-five to be eligible [to be President], X is not
thirty-five, therefore X is not eligible." Id. at 188. The latter case, says Posner, is not deductive,
superficial appearancesto the contrary, because interpreting the text to produce the premise is not
deductive.Accordingto Posner, the text could have been interpretednonliterally;thus interpretation
of the text to require that a President actually be at least thirty-five years old, rather than some less
determinatemeasure of maturity, involves a nondeductivechoice.
Posner's conclusion is correct,but only because of Posner's sleight of hand in drawing the prelimi?
nary distinctionbetween statutory rule application and common law adjudication.Posner builds into
the common law case a hardly noticeable "if." Thus, although the statutory case is not deductive
becausethe generationof the majorpremise involves an interpretivechoice, neither is the common law
case deductive, because its major premise is also a choice. Note that although both are equally
nondeductive,both can be equally deductive once the major premise is generated. We therefore can
reformulatethe issue this way: If we make the original determinationthat the language of a rule is to
be interpretedliterally, then the processof rule application is indeed deductivein any case in which a
putative application is definitionally incorporatedwithin the scope of the rule as set forth in its major
premise.

1988] Formalism 523
conjoins two different elements: mechanical deducibility and the existence

of a closed system. Neither element on its own necessarily implies the
other, however. Mechanical deducibility need not entail closure. If we had
a legal rule prohibiting all actions specifically condemned by the United
Nations, for example, the coverage of the rule would be readily determi-
nable, even if the answers were found outside the legal system (narrowly
construed). Conversely, nonmechanical judgments can be made within the
boundaries of a single
system.49 Consider the questions of whether there
should be a three-point shot in basketball or a designated hitter in base?
ball. These are not easy questions (nor are they important ones), but their
answers are internal to the games at issue; they involve a determination of
whether the proposed
change serves the goals of the game.50 Although
mechanical deducibility is thus analytically severable from systemic isola?
tion, the two are commonly conjoined when critics deride legal decisions
or theories as "formalistic," because both limit the domain of choices
available to a decisionmaker.51
Having posited a model in which the decisionmaker's choice is limited
by rules, we now must determine whether this model is descriptively accu?

rate and normatively sound. The descriptive question, which I will take
up first, is whether such limitation of choice by the words on a printed

page is possible. To put it differently, and to distinguish this version of
formalism from that considered in the previous section, the question is
whether choice can be constricted by a canonical set of words on a printed
page, or whether the choices open to a seemingly constrained deci?
sionmaker are in fact virtually the same as those available to an uncon-
strained decisionmaker. This descriptive question in turn has both concep?
tual and psychological aspects. if it
Even may be conceptually possible for
language to constrain choice, it may still be beyond the psychological ca-
49. For an important defense of this variety of formalism, see Weinrib, Legal Formalism, 97
Yale L.J. (forthcoming 1988).
50. My point here parallels Dworkin's notion of "fit." As Dworkin illustrates the point, the de?
terminationwhether the existence of a homosexual relationshipbetween David and Steerforthbest fits
David Copperfieldis by no means mechanical,but its resolutiontakes place largely within the bound?
aries of the novel. Dworkin, No Right Answer?, in Law, Morality and Society: Essays in Hon?
our of H.L.A. Hart 58 (P. Hacker & J. Raz eds. 1977). A slightly different version of this article
appears under the same title in 53 N.Y.U. L. Rev. 1 (1978) and as Is There Really No Right Answer
in Hard Cases?, in R. Dworkin, A Matter of Principle 119 (1985). In a later work, Law's
Empire, supra note 31, Dworkin broadens the systemic boundaries with which he is concerned to
encompassthose norms commonly understoodas legal, political, and moral. The expansion of these
boundariesis a separateissue, however;one could agree with Dworkin that it is possible to look for fit
within a domain while disputing the size of the relevant domain.
51. The mechanical aspects of formalism are stressed in the important discussion in Kennedy,
supra note 1. The concept of formalismas not necessarilymechanicalbut involving significant limita?
tions on otherwise eligible results is the focus of Tushnet, supra note 1. See also, Grey, The Constitu?
tion as Scripture, 37 Stan. L. Rev. 1, 4 n.8 (1984) (distinguishing between "operative" textual
norms that guide decisions themselves and "non-operative"textual norms that tell decisionmakersto
use decisive norms outside text).

pacity of those who make decisions to abide by these constraints. But let
us turn first to the conceptual question.
Is it possible for written norms to limit the factors that a decisionmaker
considers? At first glance, the answer to this
question seems to be "no."
Language is both artificial and contingent and therefore appears insuffi-
ciently rigid to limit the choices of the human actors who have created it.
The word ucat," for example, could have been used to refer to canines,
and the English language could have followed the language of the Es?
kimos in having several different words to describe the varieties of snow.
Yet this answer confuses the
long-term mobility language of with its
short-term plasticity, and is a conclusion comparable to taking the ponder-
ous progress of a glacier as indicating that it will move if we put our
shoulders against it and push. Of course language is a human
creation,
and of course the rules
of language are contingent, in the sense that they
could have been different. It is also beyond controversy that the rules of
language reflect a range of political, social, and cultural factors that are
hardly a priori. But this artificiality and does not deny the
contingency
short-term, or even intermediate-term, noncontingency of meaning. If I go
to a hardware store and
request a hammer, the clerk who hands me a
screwdriver has made
a mistake, even though it is artificial, contingent,
and possibly temporary that the word "hammer" represents hammers and
not screwdrivers. Similarly, a rule requiring candidates to file nominating
petitions at a certain place by a certain time on a certain day is violated by
filing in the wrong place or after the specified time. Whatever the real
judge did say in Hunter v. Norman, and whatever some judge might have
said in any of my hypothetical variants, none of them would be that
Hunter, in filing at 5:03 p.m., had filed at or before
p.m. 5:00
The questions about
the possibility of linguistic constraint can be clari-
fied by considering again the rule prohibiting vehicles in the park. But
now let us turn from its peripheral applications to the central applica?
tions?whether cars and
trucks are excluded. Hart assumed that,
whatever else the rule did, it excluded cars and trucks. This was the rule's
ucore" of settled meaning and application.52 Against this, Fuller offered
the example of a statue of a truck erected as a war memorial by a group
of patriotic citizens. According to Fuller, the example challenges the idea
that a rule will have a settled core of meaning which can be applied with?
out looking at the rule's purpose. Fuller argues that it cannot be deter?
mined whether the truck, which is a perfectly functional vehicle, fails into
the rule's periphery or core unless one considers the purpose of the rule.53
Fuller's challenge is ambiguous, however; there are three variant interpre-
52. Hart, supra note 10, at 607.

53. Fuller, Positivism and Fidelity to Law ? A Reply to Professor Hart, 71 Harv. L. Rev. 630,
663 (1958).

1988] Formalism 525
tations of his challenge to the theory of linguistic constraint.54 One inter?

pretation of Fuller's challenge is that legal systems necessarily incorporate
rule-avoiding norms such as those discussed earlier.55 Legal systems must
provide some escape route from the occasional absurdity generated by lit?
eral application because
applying the literal meaning of a rule can at
times produce a result
which is plainly silly, clearly at odds with the pur?
pose behind the regulation, or clearly inconsistent with any conception of
wise policy. Insofar as a legal system offers its decisionmakers no legiti?
mate escape from unreasonable consequences literally indicated by the
system's the system is much less a legal system, or is at least not a
norms,
legal system worthy of that name. This argument, however, asserts a nor?
mative point about how legal systems should operate, rather than any nec?
essary truth about how the norms themselves operate. Moreover, the ar?
gument itself admits
potential the
binding authority of rules: If rules
require an escape route to avoid the consequences of literal application,
then it must be that literal application can generate answers different
from those which a decisionmaker would otherwisechoose. Thus, this in?
terpretation fails to challenge the possibility of linguistic constraint; it
merely points out the undesirability of employing it too rigorously in cer?
tain domains.
Alternatively, Fuller might be arguing that legal systems necessarily re?
quire the interpretation of regulatory language in light of the purpose of
the regulation. As with the first interpretation of the challenge, however,
this interpretation focuses on whether a rule should bind, and it leaves the
claims of linguistic determinacy untouched. We still can imagine a system
in which decisionmakers do not interpret clear regulatory language ac?
cording to its purpose if its purpose from the regulatory
diverges lan?
guage. The outcome in some instances might seem absurd, but it is ques-
tion-begging to use the existence of the absurd result as an attack on the
possibility of a core of literal meaning.
Finally, Fuller might be interpreted as making a point about language
itself: He might be arguing that meaning cannot be severed from the
speaker's purpose and thatmeaning must be a function of the specific
context in which words are used. Fuller's argument that the idea of literal
meaning is incoherent, an argument also made by other critics,56 reveals a
54. Fuller's example and other illustrations of seemingly absurd results generated by applying a
rule without attention to the circumstancesof its creationfigure prominentlyin criticismof formalism.
See, e.g., Moore, A Natural Law Theory of Interpretation, 58 S. Cal. L. Rev. 277, 386-88 (1985);
H. Hart & A. Sacks, The Legal Process: Basic Problems in the Making and Application of Law
1148-78 (tentative ed. 1958) (unpublished manuscript);see also Dworkin's use of Riggs v. Palmer,
supra note 31.
55. This use of the term "necessarily"to describe the essential features of anything properly
called a legal system would be consistent with the general tenor of Fuller's jurisprudence. See L.
Fuller, The Morality of Law (1964); R. Summers, Lon L. Fuller 27-31, 36-40 (1984).
56. See, e.g., Boyle, The Politics of Reason: Critical Legal Theoryand Local Social Thought, 133
U. Pa. L. Rev. 685, 708-13 (1985) [hereinafterThe Politics of Reason] (arguing that words do not

mistaken view of the nature of language. Fuller and his followers fail to
distinguish the possibility and existence of meaning from the best or ful-
lest meaning that might be gleaned from a given communicative context.
In conversation, I am assisted in determining what a speaker intends for
me to understand by a number of contextual cues, including inflection,
pitch, modulation, and body language, as well as by the circumstances
surrounding the conversation. That such contextual cues assist my under?
standing, however, does not imply that the words, sentences, and
paragraphs used by the speaker have no meaning without those cues. The
uno vehicles in the park" rule clearly points to the exclusion of the statue
from the park even if we believe that the exclusion is unnecessary from
the point of view of the statute's purpose.
If I come across an Australian newspaper from 1827, I can read it
because I understand, acontextually, the meaning of most of the words
and sentences in that newspaper, even though with better historical un?
derstanding I might understand more of what was written by a colony of
transported English convicts. This example does not demonstrate that lan?
guage is unchanging, nor that language can be perfectly understood with?
out attention to context, but rather that some number of linguistic conven?
tions, or rules of language, are known and shared by all people having
competence in the English language. Linguistic competence in a given lan?
guage involves understanding some number of rules also understood by
others who are linguistically competent in the same
language. When indi?
viduals understand the same rules, they convey meaning by language con-
forming to those rules.57 Members of the community of English speakers,
have essences or core meanings); Boyle, Thomas Hobbes and the Invented Tradition of Positivism:
Reflections on Language, Power, and Essentialism, 135 U. Pa. L. Rev. 383, 408-19 (1987) (dis?
cussing Hobbes' rejectionof notion of linguistic essences). In passing, I note my disagreementwith
those who describe as "post-Wittgensteinian"the view that meaning cannot be separated from the
particular context of a particular utterance. E.g., Boyle, The Politics of Reason, supra, at 708. A
footnotein a law review article is hardly the place to debate interpretationsof Wittgenstein, including
whether Wittgensteincan even plausibly be interpretedto support a pragmatist/particularisttheory of
meaning. Yet I would briefly note that a fair reading of Wittgenstein reveals that he argued that the
meaning of a word is a function of how that word is contingently used in an existing linguistic
community, but emphatically not a function of how the word is used on a particular occasion by a
particular member of that community.
It is crucial to recognizethe seductivequality of phrases like "post-Wittgensteinian,"which suggest
that if the reader acknowledgesWittgenstein'sgenius, then she must agree with the point describedin
those terms. It is better to discuss the point at issue without attempting to lean on the argumentative
props of associationswith philosopherswhose names are currentlyfashionablein legal circles. In light
of the still-raging disputes about the most foundational questions in the philosophy of language, to
substitute Wittgenstein'sname for an argument is unwarrantedeven if the use of his name is accu?
rate. When that use is mistaken or at the very least contested, the dangers of facile borrowing from
other disciplines are compounded.
This criticism of the presentationof Boyle's argument has no bearing on its underlying validity,
however. Although I disagree with much of what he and Fuller argue, those arguments raise central
questions about the nature of law which I believe should be confronteddirectly. Boyle's useful per?
spectives are ill-served by clothing them in what appears to me to be an idiosyncraticmisreading of
Wittgenstein.
57. For a particularlyinsightful and influential articulationof the view that meaning exists inde-

1988] Formalism 527
for example, possess shared understandings that enable them to talk to all
other members of the community.
Among the most remarkable features of language is its compositional
nature, i.e., the way in which we comprehend sentences we have never
heard before. We can do this because rules, unspecified and perhaps un-
specifiable, allow us to give meaning to certain marks and certain noises
without having to inspect the thought processes of the speaker or the full
context in which words appear. Words communicate meaning at least
partially independently of the speaker's intention. When the shells wash
up on the beach in the shape of C-A-T, I think of small house pets and
pendent of speaker's purpose or other related aspects of context, see J. Searle, Speech Acts: An
Essay in the Philosophy of Language 42-50 (1969). This view also seems to be the import of
paragraphs 489-512 of L. Wittgenstein, Philosophical Investigations (G.E.M. Anscombe
trans. 3d ed. 1953). A similar interpretationof Wittgenstein, relying on different passages, is G.
Baker & P. Hacker, Wittgenstein: Rules, Grammar and Necessity 329-38 (1985). Indeed,
even those who are rightly concerned with the foundational rule-following questions posed by
Wittgenstein would not dispute that "communal language constitutes a network of determinate pat?
terns." Wright, Rule-Following, Objectivityand the Theory of Meaning, in Wittgenstein: To
Follow a Rule 99, 105 (S. Holtzman & C. Leich eds. 1981).
Interpretationsof Wittgenstein apart, acceptance of the possibility of literal meaning has passed
into the commonplaceof contemporaryanalytical philosophy of language, even while philosophers
hotly dispute the source or explanation of that phenomenon.See, e.g., W. Alston, Philosophy of
Language 74-75 (1964); M. Black, Meaning and Intention, in Caveats and Critiques: Philo?
sophical Essays in Language, Logic, and Art 109 (1975); S. Cavell, Aesthetic Problems of
Modern Philosophy, in Must We Mean What We Say? 73, 80-82 (1969); S. Cavell, Knowing
and Acknowledging,id. at 238, 248-49; D. Davidson, Inquiries into Truth and Interpreta?
tion xix, 243-64 (1984); D. Holdcroft, Words and Deeds: Problems in the Theory of
Speech Acts 122-23 (1978); R. Martin, The Meaning of Language 217 (1987); M. Platts,
Ways of Meaning: An Introduction to a Philosophy of Language 130-32 (1979); I.
scheffler, beyond the letter: a philosophical inquiry into ambiguity, vagueness and
Metaphor in Language 81 (1979).
In one philosopher'swords:
It is a platitude?something only a philosopherwould dream of denying?that there are con?
ventions of language, although we do not find it easy to say what those conventionsare. If we
look for the fundamentaldifferencein verbal behaviorbetween membersof two linguistic com?
munities, we can be sure of finding something which is arbitrarybut perpetuatesitself because
of a common interest in coordination. In the case of conventions of language, that common
interest derives from our commoninterest in taking advantageof, and in preserving,our ability
to control others' beliefs and actions to some extent by means of sounds and marks. That
interest in turn derives from many miscellaneous desires we have; to list them, list the ways
you would be worse off in Babel.
D. Lewis, Languages and Language, in 1 Philosophical Papers 163, 166 (1983).
Obviously some tension exists between the way that language is discussed in analytic philosophy of
language and the way that it is discussedin other circles, including literary theory. Part of the differ?
ence between the terms of the debate in these two circles can be explained by the different extent to
which the two disciplines focus on "difficult" interpretations.This may also explain the extent to
which some branchesof legal theory, with their focus on difficult interpretationsin linguistically hard
cases, have been drawn to literary theory. Moreover, insofar as literature exists primarily to illumi-
nate, inspire, and transform, its very existence encourages attempts to pierce literal meaning. The
relationship between the enterprise at issue and the view of literal meaning adopted suggests an im?
portant question: Might the purposes of the legal enterprise be so different from those of interpreting
literature that literal meaning is no longer an obstacle but instead a tool? I have no answer to this
question, nor do I intend to offer a few easy citations to suggest a familiarity I do not possess. Never?
theless, the very differencesin focus between analytic philosophy of language and literary theory may
suggest that it is a bit too easy, for me or for those who draw on literary theory or other perspectives
on language, to assume that the applications of these perspectivesto law cannot take place without
some theoretical slippage.

not of frogs or Oldsmobiles precisely because those marks, themselves,

convey meaning independently of what might have been meant by any
speaker. Of course there can never be totally acontextual meaning.58 The
community of speakers of the English language is itself a context. Yet

meaning can be "acontextual" in the sense that that meaning draws on no
other context besides those understandings shared among virtually all
speakers of English.
Given that the meaning of words may be acontextually derived from
our understandings of language, the central question becomes whether
enough of these understandings exist to create the possibility of literal lan?
guage. In other words, we must ask whether words have sufficient acon?
textual import so that communication can take
place among speakers of
English a
in suchway that at least a certain limited range of meaning, if
not one and only one meaning, will be shared by all or almost all speakers
of English. The answer to this question is clearly "yes." As with the shells
that washed up on the beach in the shape of C-A-T, words strung to?
gether in sentences point us toward certain meanings on the basis of our

shared understandings. At times these sentences may be descriptive, but at
other times these comprehensible sentences may be general prescrip?
tions?rules. Because we understand the rules of language, we understand
the language of rules. Contextual understanding might be necessary to
determine whether a given application does or does not serve the purposes
of a rule's framers. Yet the rule itself communicates meaning as well,
although that meaning might depart from the purposes behind the rule or
from the richer understanding to be harvested from considering a wider
range of factors than the rule's words. That we might learn more from
considering additional factors or from more fully understanding a
speaker's intentions does not mean that we learn nothing by consulting the
language of rules themselves.
Of course, certain obvious, accessible, and by and large undisputable
features of rules distinguish the meaning we cull from them from our
interpretation of other types of communications. For example, when we
interpret a rule of law, we understand that it is a law and that it is to be
interpreted in light of surrounding language in the same law. In addition,
ordinary "lawspeak" (habeas corpus, certiorari, party, appeal) can be
viewed as a language for a subcommunity in the community of English
speakers, capable of doing within the subcommunity what ordinary lan?
guage does within the larger community of English speakers. Thus, al?
though all those reading a statute come to that task with certain shared
58. See J. Searle, Literal Meaning, in Expression and Meaning 117 (1979) (literal meaning
exists albeit only against a set of backgroundassumptions about contexts in which sentence could
appropriatelybe uttered);Moore, supra note 54, at 304-07 (arguing that minimal context allows and
is required for fixing referencesto singular terms).

1988] Formalism 529
assumptions, it is probable that almost all lawyers add to these an addi?

tional set of assumptions which are shared mainly by lawyers.69
Both those within the legal community and those withinthe larger lin?
guistic community are capable of deriving the literal import of rules, even
though the literal lawyer's meaning may occasionally diverge from the
literal lay meaning of the same term. A law that limits membership in
Parliament to those who take an oath "on the true faith of a Christian"
literally excludes Jews by its language.60 A statute requiring that the
master of a vessel shall record in the log book "[e]very birth happening on
board, with the sex of the infant, and the names of the parents,"61 can be
understood by virtually any speaker of English as requiring the master to
take certain actions.62 In these and countless other cases, statutes can be
wrenched from most of the context of their enactment and application and
still be read and understood.
59. Note that I am talking about language and about two different embellishmentson the main
theme of literal meaning. First, ordinary people within a given linguistic culture might share, as
linguistic conventions,knowledge about how to interpret the language of rules, including conventions
relating to the differencebetween normativeand descriptivelanguage and conventionstelling them to
interpretwords in light of surroundinglanguage in the same rule or statute. This suggests only that
all competent speakersof the language in which the text is written have access to a certain minimal
amount of noncontroversialinformation about what kind of text it is.
Second, literal meaning is not necessarily ordinary meaning, because linguistic conventions may
exist within a technicalor professionalsubcommunityof a larger community.For example, photogra-
phers may have a literal sense of the meaning of the term "burningin," physicians may have a literal
sense of the meaning of the term "Cushing's Syndrome,"and lawyers may have a literal sense of the
meaning of the term "assumpsit,"even though none of these are terms used at all or in the same way
by ordinary English speakers. This second embellishment, however, must be sharply distinguished
from other notions of "conventionalism"that build in much more than linguistic meaning. See, e.g., S.
Burton, An Introduction to Law and Legal Reasoning (1985); Fiss, Conventionalism, 58 S.
Cal. L. Rev. 177 (1985); Fiss, Objectivityand Interpretation, 34 Stan. L. Rev. 739 (1982). The
conventionalistlegal literature talks merely about the conventionsof permissible legal argument and
does not confront the question of the relationship between the conventionsof permissible legal argu?
ment and the conventionsof literal meaning, whether ordinaryor technical. Thus, legal conventional?
ists such as Fiss avoid questions regardingwhether and why certain literal readings of legal rules are
or are not permissible arguments within the legal interpretive community. It is these questions, in
some sense more foundational,that concern me here, because my aim is to locate the particular per?
missible arguments in the legal interpretivecommunity, rather than merely to assert the existence of
permissible arguments.
60. Salomons v. Miller, 8 Ex. 778, 155 Eng. Rep. 1567 (1853); Miller v. Salomons, 7 Ex. 475,
155 Eng. Rep. 1036 (1852).
61. 46 U.S.C. ? 201 (1958), repealed by Pub. L. No. 98-89, 97 Stat. 600 (1983).
62. "When you come tomorrow, bring my football boots. Also, if humanly possible, Irish water
spaniel. Urgent. Regards. Tuppy."
"What do you make of that, Jeeves?"
"As I interpret the document,sir, Mr. Glossop wishes you, when you come tomorrow,to bring his
football boots. Also, if humanly possible, an Irish water spaniel. He hints that the matter is urgent,
and sends his regards."
"Yes, that's how I read it, too . . . ."
P.G. Woodehouse, The Ordeal of Young Tuppy, quoted in S. Blackburn, Spreading the Word:
Groundings in the Philosophy of Language 3 (1984).

B. Does Language Constraint
The conceptual question of whether literal meaning is possible can

therefore be answered affirmatively. Rules may point to results that di-
verge from those that a decisionmaker would have reached apart from the
literal meaning of the rule. When there is such divergence, however, the
psychological question remains: Is it possible in such cases for deci?
sionmakers to follow the literal meaning of the rule rather than their own
judgment regarding how the case should be resolved?
The psychological challenge to formalism involves the claim that deci?
sionmakers will usually take all factors they believe to be relevant into
account, or at least that they will usually feel compelled to reach "reason?
able" results whether or not the language of the rule points in that direc?
tion.63 When expressed this way, it is obvious that the psychological ques?
tion is an empirical one. Accordingly, it cannot be answered by mere
argument. Yet despite legal
scholarship's sorry failure to take the psycho?
logical challenge seriously,64 the possibility that judges usually obey their
own rule-independent judgment is, on its face, quite plausible. We can
easily imagine a world in which decisionmakers consider everything that
they feel relevant and ignore, or at least slight, any inconsistent external
instructions in making their decisions. The question is whether that is the
world of the law.
Certainly some legal decisionmakers conform to this model. Although
one may dispute as excessive the generalizations of the more extreme
Realists, it is difficult to deny the existence of decisionmakers who consult
the rules only to create post hoc rationalizations. Indeed, to the extent that
legal systems resemble the model in which a rule-avoiding norm is always
available, such behavior is encouraged. Insofar as the view ever prevailed
that there were few decisionmakers who rejected, ignored, or bent the
plausibly determinate mandates of governing rule, it is important that that
view be shown for the optimistic fantasy it is. Yet to accept that some
judges arrive at decisions without considering rules does not imply that all
or most decisionmakers act in such a way as either an inevitable feature of
human nature or even as a contingent feature of judicial behavior.
Just as it is a mistake to assume that because some judges ignore rules
most judges do so, it is also a mistake to assume that because rules some?
times constrain, they usually constrain. The truth, an empirical rather
than a logical one, plainly lies between the extremes of always and never,
or even between the lesser extremes of rarely and usually. Although this is
not the place to examine the rudimentary empirical work that has been
63. See Aleinikoff, Constitutional Law in the Age of Balancing, 96 Yale L.J. 943, 985, 1004
(1987).
64. Noteworthy exceptions are J. Frank, Law and the Modern Mind (1930), and Kennedy,
supra note 37. However, legal scholarship still must systematically investigate the important issues
that Frank, Kennedy, and others have raised in an impressionisticway.

1988] Formalism 531
done on the question, it is sufficient for my purposes to note that this

research has, not surprisingly, yielded the result of "sometimes."66 In
some settings, decisionmakers sometimes apply instructions external to
their own decisional process even if those instructions diverge in outcome
from the results the decisionmakers otherwise would have reached. This
conclusion should cause no surprise as long as we recognize that people
often do what others think best. If privates in the army often follow orders
instead of making autonomous choices, and if privates might behave in
this way with respect to general orders in addition to particularized com-
mands, we can imagine judges doing the same with respect to rules.66
We have seen that, as a descriptive and conceptual matter, rules can
generate determinate outcomes; that those outcomes may diverge from
what some decisionmakers think ought to be done; and that some deci?
sionmakers will follow such external mandates rather than their own best
particularistic judgment. The normative question of formalism now re?
mains: To what extent should a system legitimate the avoidance of literal
meaning when avoidance seems to be the optimal outcome to the deci?
sionmaker? To put it simply, now that we have established that formal?
ism?in the sense of following the literal mandate of the canonical formu?
lation of a rule?is conceptually and psychologically possible, we must ask
whether it is desirable. Before turning to that question, however, I want
65. See D. Black, The Behavior of Law (1976); L. Friedman, The Legal System: A
Social Science Perspective (1975); Hogan & Henley, Nomotics: The Science of Human Rule
Systems,5 Law & Soc'y Rev. 135 (1970); Johnson, Law, Politics, and Judicial Decision Making:
Lower Federal Court Uses of Supreme Court Decisions, 21 Law & Soc'y Rev. 325 (1987);
Scandura,New Directionsfor Theoryand Research on Rule Learning, 28 Acta Psychologica 301
(1968).
66. There is somethingunrealistic about all of this, because it erroneouslyassumes that my para?
digm "easy" cases are representativeof the kinds of decisions that come before decisionmakers.They
are not, at least when we take "decisionmaker"in a somewhat narrow sense to refer to formal deci?
sionmakerssuch as judges sitting in courts of law. In most legal systems, various screening devices
ensure that cases at the center of decisionaldeterminacywill not enter the formal adjudicativeprocess.
The time and expense of litigation and the widespreadinclination to avoid futile battles are such that
decisionsat the core of settled meaning seldom confrontany formal decisionmakingprocess.See Priest,
Reexamining the Selection Hypothesis:Learning from Wittman's Mistakes, 14 J. Legal Stud. 215
(1985) (develops selection hypothesis to determine which cases are settled and which are litigated);
Priest and Klein, The Selection of Disputes for Litigation, 13 J. Legal Stud. 1 (1984) (presents
model to predict whether litigation will be resolved by suit or settlement).
This, however, is but a contingentfeature of modern legal systems. It is possible to imagine a legal
system closer to a sporting event, where umpires call "safe" or "out" on every play, or where officials
with red penalty flags in their back pockets patrol the social landscape, ready to throw the flag and
call "tort,"or "crime,"or "breachof etiquette" whenever there is a transgressionof the rules. Obvi?
ous logistical problemsprevent such a system from being a reality, but it is a useful Gedankenexperi-
ment for thinking about the innumerable instances in which rules are followed or clearly broken
without coming to the attention of the judicial system.
Many legal systems, unlike those with "roving umpires," operate largely in the area of linguistic
indeterminacy,generated either by vagueness of the governing norm or by open texture when previ?
ously clear norms confrontthe unexpected.And in some systems, such as that of the United States, the
likelihood of success is sufficient to make it worth litigating cases in which linguistic determinacy
produces a politically or morally uncomfortableresult. But that is exactly our question, because the
weight the system gives to literal meaning will determine the extent to which it is worth litigating
against literal meaning.

to answer an important counterargument to the possibility of the type of

formalism defined in this section.
C. Language and Rules
Until this point my argument may appear to create a false dichotomy. I

have counterpoised the vision of decisionmakers who follow the literal lan?
guage of a rule with that of decisionmakers who follow the dictates of
their own externally unguided opinions. But are these the only alterna?
tives? Although rules can and do constrain, is it not possible that these
rules not be equated
need with the literal meaning of the language in
which they happen to be articulated? In other words, is portraying the
vice of formalism as the vice of literalism actually confusing rules with the
literal meanings of their explicit formulations?
The argument that can be distinguished
rules from the language in
which they are written
has a distinguished lineage in Anglo-American le?
gal thought. We see, for example, Ronald Dworkin urging interpreters to
search for or to construct the "real" rule lying behind the mere words on
a printed page.67 Somewhat less explicitly, the "mischief rule" of the com?
mon law compels the literal language of a rule to yield to the purpose
behind a rule when application of the literal language would frustrate the
rule's purpose.68 Indeed, the mischief rule and related principles urging
the primacy of purpose over text are features of the thinking of Fuller, of
Hart and Sacks, and also of Llewellyn's "Grand Style" of judging.69 As a
matter of fact, Llewellyn distinguished the Grand Style from the Formal
Style he believed
because that formalism, as the obeisance to the literal
language of a rule, could frustrate the rule's purpose and lead to difficul?
ties where the practical consequences of the decision would indicate a dif?
ferent result.
The language in which a rule is written and the purpose behind that
rule can divergeprecisely because that purpose is plastic in a way that
literal language is not. Purpose cannot be reduced to any one canonical
formulation, for when purpose is set down canonically, that canonical for?
mulation of purpose may frustrate the purpose itself. It is because purpose
is not reduced to a concrete set of words that it retains its sensitivity to
novel cases, to bizarre applications, and to the complex unfolding of the
human experience. Thus, for the recourse to purpose to "solve" the prob?
lem of formalism, the purpose must not be imprisoned in the rigidity of
67. Law's Empire, supra note 31, at 16-17.

68. The standard referencesfor this rule are Heydon's Case, 76 Eng. Rep. 637 (Ex. 1584), and
Church of the Holy Trinity v. United States, 143 U.S. 457 (1892); see discussionsupra note 28. See
also J.G. Sutherland, Statutes and Statutory Construction ?? 45.05, 45.09 (N. Singer
rev. ed. 1984).
69. K. Llewellyn, supra note 1, passim; see W. Twining, Karl Llewellyn and the Real?
ist Movement 210-11 (1973).

1988] Formalism 533
words. This unrigidified can be explained,

purpose clarified, and enriched
as new examples and
applications come to our attention. The purpose
behind the "No vehicles in the park" regulation is not embarrassed by the
statue of the truck exactly because purpose can bend to the circumstances
of the moment in a way that language, with its acontextual autonomy of
meaning, cannot. In contrast, the term "vehicles," at least at the core,
literally refers to vehicles;70 if it turns out that the prohibition of some
vehicles does not serve the purpose of the regulation, then the embarrass?
ment is unavoidable.71
If adhering to concretized language causes this embarrassment, then
why not adhere to the purpose of the rule rather than the words of the
rule? To do so would conform with the models advocated by Fuller,72
Hart and
Sacks,78 Dworkin,74 and Llewellyn75 and would match what
can be called, with only negligible exaggeration, the current paradigm of
American statutory interpretation.76 Yet locating the idea and the force of
a rule in its purpose rather than in its formulation poses the same prob?
lem posed by concretized rules, except at one remove. To illustrate the
point, suppose the purpose of the "no vehicles in the park" regulation is
the preservation of peace and quiet in the park. Suppose, as well, that this
purpose derives from an even deeper purpose of maximizing the pleasure
of the residents of the town. Now imagine that a town native who has just
70. Note that the "No vehicles in the park" example may be a flawed illustration of the problem
Hart, Fuller, and I explore, because locomotive capacity may now be definitional of a "vehicle."
Insofar as this is true, the statue is not a vehicle, and no conflict arises between literal meaning and
purpose. This is a defect only in the example, however, rather than in the general formulationof the
issue. I will therefore stipulate, for the purposes of this argument, that a statue of a vehicle is a
vehicle, just as a lion in a cage is still a lion. Consider a rule prohibiting "live animals on the bus"
and whether it would prohibit carrying on the bus three live goldfish in a sealed plastic bag.
71. My point about the plasticity of purpose should not be confused with claims, often correct,
about the indeterminacyof purpose. See, e.g., Easterbrook,Foreword: The Court and the Economic
System, 98 Harv. L. Rev. 4, 15-18 (1984); Easterbrook,Statutes' Domains, 50 U. Chi. L. Rev.
533, 537-38 (1983); Kennedy, supra note 1; Posner, Statutory Interpretation?In the Classroom
and in the Courtroom, 50 U. Chi. L. Rev. 800, 819-820 (1983). Insofar as purpose becomes both
concreteand determinate,as when everyone agrees what the purpose is, the argumentthat "ruleness"
resides in purpose becomes more plausible. But when some conception of purpose is determinate,
noncanonicalpurpose itself can operate formalistically.Conversely, if purpose is comparativelyinde?
terminate, then it looks especially odd to say that the rule exists not in the specific rule-formulation,
but in the quite different and nonspecific purpose. Thus, those who argue for the indeterminacyof
purpose make claims consistent with mine.
72. L. Fuller, supra note 55, at 81-91; Fuller, supra note 53; Fuller, The Speluncean Explor-
ers, 62 Harv. L. Rev. 616, 620-26 (1949) ("opinion" of Foster, J.).
73. H. Hart & A. Sacks, supra note 54.
74. Law's Empire, supra note 31.
75. K. Llewellyn, supra note 1.
76. See Wellman, Dworkin and the Legal Process Tradition: The Legacy of Hart & Sacks, 29
Ariz. L. Rev. 413 (1987); Note, Intent, Clear Statements, and the Common Law: Statutory Con?
struction in the Supreme Court, 95 Harv. L. Rev. 892 (1982). Recent manifestationsof this para?
digm include G. Calabresi, A Common Law for the Age of Statutes (1982); Eskridge, Dy?
namic StatutoryInterpretation, 135 U. Pa. L. Rev. 1479 (1987); Langevoort,Statutory Obsolescence
and the Judicial Process: The Revisionist Role of the Courts in Federal Banking Regulation, 85
Mich. L. Rev. 672 (1987). But see, e.g., United States v. Locke, 471 U.S. 84 (1985) (failure to file
timely claim deprives petitioner of right, irrespectiveof statutory purpose).

won six gold medals in the Olympic Games is returning to this park, the
scene of her youth, along with a widely popular President of the United
States. Suppose as well that the park with the "no vehicles in the park"
is the only suitable place for the motorcade, which must be a
regulation
motorcade because the President is disabled and cannot walk. Under these
circumstances, the purpose behind the "no vehicles in the park" rule
would be served by excluding the motorcade, but the purpose behind that
Thus the same logic that requires the for?
purpose would be frustrated.
mulation of a rule to be defeasible in the service of its purpose would also
that purpose to be defeasible in the service of the purpose behind
require
it.
As the example reveals, the potential tension between the general goal
and its concretized instantiation exists at every level. At one level, the ten?
sion is between language and purpose; at the next, it is between that pur?
behind at the between the deep
pose and the deep purpose lying it; next,
and an even deeper purpose; and so on. When we decide that
purpose
must not be frustrated by its instantiation, we embark upon a
purpose
infinite in which all forms of concretization are
potentially regress
defeasible.
The view that rules should be interpreted to allow their purposes to
in fact collapses the distinction between a rule and a
trump their language
reason, and thus loses the very concept of a rule.77 Rules are by definition
known and unknown under
general. They gather numerous particulars
headings such as "vehicles," "punishment," "dogs," and "every person
who is directly or indirectly the beneficial owner of more than 10 per
centum of any class of any [registered] equity security (other than an ex-
empted security)." After identifying a category of items or events to which
the rule applies, in the protasis, rules then prescribe what shall be done
with these particulars in the apodosis7* Occasionally, however, not all of
the particulars comprising the rule's category of coverage are suitable for
the prescribed treatment; the generalizations that are a necessary part of
in a manner that be
any rule treat all members of the class may appropri?
ate only for most members of the class. What, then, is to happen when a
case arises in which the generalization does not apply to this particular?
When a rule's prescribed treatment is unsuitable, if the decisionmaker
77. Note, however, that this claim is not inconsistent with the view that rules should be inter?
preted to further their purposes when several interpretationsof the rule are possible and all are
supported by the language ofthe rule. In such cases, it is not only possible but positively desirableto
choose the interpretationthat will serve the rule's purpose. See H.L.A. Hart, Introduction, in Es?
says in Jurisprudence and Philosophy 1, 8 (1983).
78. On this terminologyfor the structureof rules, which distinguishesthe part of the rule specify-
ing its operative facts from the part describing the consequencesflowing from the existence of those
facts, see W. Twining & D. Miers, How To Do Things With Rules 136-40 (2d ed. 1982). See
also Friedman,Legal Rules and the Process of Social Change, 19 Stan. L. Rev. 786, 786-87 (1967)
(same distinction with different labels); Schlag, Rules and Standards, 33 UCLA L. Rev. 379,
381-83 (1985) (same).

1988] Formalism 535
were to ignore the rule, the rule would not

rule providing
be a real a
reason for decision but would be a mere rule
of thumb, defeasible when
the purposes behind the rule would not be served. If every application
that would not serve the reason behind the rule were jettisoned from the
coverage of the rule, then the decision procedure would be identical to one
applying reasons directly to individual cases, without the mediation of
rules. Under such a model, rules are superfluous except as predictive
guides, for they lack any normative power of their own. By contrast, if in
cases in which the particular application would not serve the reasons be?
hind the rule, the rule nevertheless provides its own reason for deciding
the case according to the rule, the rule itself has a normative force that
provides a reason for action or decision.
In summary, it is exactly a rule's rigidity, even in the face of applica?
tions that would ill serve its purpose, that renders it a rule.79 This rigidity
derives from the language of the rule's formulation, which prevents the
contemplation of every fact and principle relevant to a particular applica?
tion of the rule. To be formalistic in Llewellyn's sense is to be governed
by the rigidity of a rule's formulation; yet, this governance by rigidity is
central to the constraint of regulative rules. Formalism in this sense is
therefore indistinguishable from "rulism," for what makes a regulative
rule a rule, and what distinguishes it from a reason, is precisely the un?
willingness to pierce the generalization even in cases in which the general?
ization appears to the decisionmaker to be inapposite. A rule's acontextual
rigidity is what makes it a rule.
D. The Idea of a Closed System
We now are in a position to reconsider the charge that formalism em-

bodies the erroneous view that law (or any other decisional domain) is a
closed system. We have seen that rules can generate answers or exclude
otherwise eligible answers from consideration. We also have seen that
there are rules, such as one prohibiting the shooting of pelicans, whose
application throughout much of their range requires recourse only to the
rule and to uncontroversial judgments of meaning and identification of
discrete particulars. There can therefore be systems whose operations re?
quire recourse only to the norms of the system and to accepted linguistic
and observational skills.
Such would be closed, but it would not necessarily
a system be com?
plete. Closedness
and completeness are different properties. Closedness re?
fers to the capacity of a system to decide cases within the confines of that
79. This is not to say that rules are always or even ever good things to have. My aim now is to
distinguish a form of decisionmakingin which generalizations have independent normative power
from a form of decisionmakingin which the full richness of the particular event always is open to
consideration.The questions of whether rules should be employed, in which domains, and to what
extent, are addressedbelow. See infra Section III.

system, while completeness refers to the extent to which a system deals

with those cases at all. A mathematical system is closed insofar as the
rules of mathematics provide an answer to the question "What is the sum
of 97 and 53?" But that same system is incomplete insofar as it provides
no answer to the question "What should the United States do about the
problem of poverty?" The dimension of completeness, although perhaps
unimportant for mathematics, is important for law precisely because most
modern legal systems claim
the ability to deal with a wide range of issues.
Insofar as the human experience is especially complex and fluid, the legal
system is likely to be frustrated by its incompleteness, its frequent inabil?
ity to answer the questions it wants to answer. Commonly, we plan for
these frustrations by rendering the norms of the law less determinate and
thereby using vagueness as the tool by which we plan for the open texture
of experience.80 As a result, legal systems, to avoid the consequences of
widespread incompleteness, often employ norms sufficiently indeterminate
to accommodate much that is important in the world at large, and in do?
ing so sacrifice the occasional virtues of closedness. Such systems are more
open even at the expense of being less predictable and less constraining of
their decisionmakers.
Thus, legal systems often reject closedness because they must deal with
a large array of problems presented by a complex and fluid world. But
this is to say that comparatively closed systems may sometimes be undesir?
able, not that they are not possible. The importance of drawing this dis?
tinction is to stress that the degree of closedness may vary, and that closed?
ness is a tool that might be usable in some domains even if, in untempered
form, it is not the only tool we would want to use for an entire system of
social control.
When applied to individual norms rather than to entire systems, closed?
ness is merely another word for ruleness. By limiting access to the reasons
behind the rule, rules truncate
the array of considerations available to a
decisionmaker.81 Rules get in the way. They exclude from consideration
factors that a decisionmaker unconstrained by those rules would take into
80. See Schauer, Authority and Indeterminacy, in Authority Revisited: NOMOS XXIX 28
(1987).
81. The closed system/open system dispute merely recasts the debate about legal positivism in
differentterminology.Any version of legal positivism is premised on what Ronald Dworkin, no posi?
tivist, has felicitously referred to as "pedigree."Taking Rights Seriously, supra note 31, at 17.
Positivism posits that legal norms are identified by reference to some other norm, rule, or standard
that distinguishes legal from non-legal norms. Hart's "rule of recognition" serves this purpose,
H.L.A. Hart, supra note 1, as does the "next higher norm" for Kelsen. H. Kelsen, The Pure
Theory of Law 193-278 (M. Knight trans. 1967). The positivist conceives of the set of norms so
pedigreedas constitutingsome sort of a closed system, although that system will not necessarilydecide
all or even most cases that come before the courts. Kelsen, for example, sees every law-applying act as
only partially determined by law. Id. at 233-36, 244-45. By contrast, the opponents of positivism,
most notably Dworkin, attack the pedigreeabilitythesis by arguing that in no case is the distinction
between pedigreed and nonpedigreednorms dispositive and consequently the characterizationof law
in terms of pedigreed norms is descriptivelyinaccurate.

1988] Formalism 537
account. Understanding the way in which rules truncate the range of rea?
sons available to a decisionmaker helps us to appreciate the distinction
between formalism and functionalism, or instrumentalism.82 Functional-
ism focuses on outcomes and particularly on the outcome the deci?
sionmaker deems optimal. Rules get in the way of this process, and thus
functionalism can be perceived as a view of decisionmaking that seeks to
minimize the space between what
a particular decisionmaker concludes,
all things considered, should be done and what some rule says should be
done. Rules block consideration of the full array of reasons that bear upon
a particular decision in two different ways. First, they exclude from con?
sideration reasons that might have been available had the decisionmaker
not been constrained by a rule. Second, the rule itself becomes a reason for
action, or a reason for decision.
The notion of a rule as a reason for decision requires further explora?
tion. What makes formalism formal is this very feature: the fact that tak?
ing rules seriously involves taking their mandates as reasons for decision
independent of the reasons for decision lying behind the rule.88 If it were
otherwise, the set of reasons considered by a decisionmaker would be con-
gruent with the set of reasons behind the rule, and the rule would add
nothing to the calculus. Rules therefore supply reasons for action qua
rules. When the reason supplied by a rule tracks the reasons behind the
rule, then the rule is in a different way superfluous in the particular case.
Rules become interesting when they point toward a different result than
do the reasons behind the rule?when they indicate, for example, that
statues of vehicles ought to be excluded eventhough the reasons behind
the rule indicate that the statues ought not to be excluded. To take these
occasionally perverse reasons as always relevant and therefore sometimes
dispositive is condemned as "formalistic" because it abstracts the mandates
of a rule from the reasons behind it. Yet that is what rules do. Refusal to
abstract the rule from its reasons is not to have rules. This refusal reduces
rules to rules
of thumb, useful but intrinsically unweighty indicators of
the results likely to be reached by direct application of reasons.
Thus, the essential equivalency of formalism and "ruleness" is before
us. Viewing formalism as merely rule-governed decisionmaking does not
make it desirable. Yet recognizing the way in which formalism is merely a
way of describing the process of taking rules seriously allows us to escape
82. See, e.g., R. Summers, Instrumentalism and American Legal Theory 136-75 (1982);
Aleinikoff, supra note 63, at 985; Summers, Professor Fuller's Jurisprudence and America's Domi?
nant Philosophy ofLaw, 92 Harv. L. Rev. 433 (1978).
83. Insofar as a system permits recourse to the purpose behind a rule's formulation but does not
permit departure from that purpose when adhering to it will produce unfortunate results or will
frustrate the even deeper purpose behind it, that system will still be formal in the sense I am now
using that term. It will also be rule-bound, because the less than totally plastic purpose (although
more plastic than the rule-formulation)will operate as a rule vis-a-vis the higher order reasons that
generated that particular purpose.

the epithetical mode and to confront the critical question of formalism:

What, if anything, is good about the unwillingness to go beneath the rule
and apply its purpose, or the purposes behind that purpose, directly to the
case before the decisionmaker?
III. Should Choice Be Restricted?
Let me recapitulate. One conception takes the vice of formalism to con-

sist of a decisionmaker's denial, couched in the language of obedience to
clear rules, of having made any choice at all. Yet rules, if followed, may
not leave a decisionmaker free choice. Rules can limit decisional choice,
and decisionmakers can abide by those limitations. Those limitations come
in most cases from the literal
language of a rule's
formulation, for to take
a rule as anything other than the rule's formulation, or at least the mean?
ing of the rule's formulation,84 is ultimately to deny the idea of a rule.
Thus, formalism merges into ruleness, and both are inextricably inter-
twined withliteralism,86 i.e., the willingness to make decisions according
to the literalmeaning of words or phrases or sentences or paragraphs on a
printed page, even if the consequences of that decision seem either to frus-
trate the purpose behind those words or to diverge significantly from what
the decisionmaker thinks?the rule aside?should be done. But does dem-
84. In a trivial sense, rules differ from their formulations.See, e.g., G. Baker & P. Hacker,
supra note 57, at 41-52; M. Black, The Analysis of Rules, in Models and Metaphors: Studies
in Language and Philosophy 95 (1962); D. Shwayder, The Stratification of Behaviour
241 (1965); G. von Wright, Practical Reason 68 (1983). "Do not walk on the grass," "Walking
on the grass is prohibited,"and "No walking on the grass" constituteone and not three rules. Refer?
ring to these three formulationsas formulationsof only one rule, however, presupposesthat all have
the same meaning, that the differencesare syntacticaland not semantic. Thus, the distinctionbetween
a rule and its formulationis like the distinctionbetween a propositionand a sentence. When I discuss
a rule and equate it with its formulation, I therefore mean that a rule is that set of semantically
equivalent rule formulations.
85. My referencesto "literalism"are slightly metaphorical.As noted above, see supra note 59,
literalismincludesthose aspects of context, such as the appearanceof words in a statute rather than in
a poem, that are accessible to all or most readers. Moreover, although I often use single words as
examples, statutes are not read word by word, but instead by sentences, paragraphs,and larger units
of text. This is not to deny, however, that the ability to assign meanings to individual words is what
enables us to understanda sentence we have never seen before. See D. Davidson, Truth and Mean?
ing, in Inquiries into Truth and Interpretation 17 (1984). This assertion, however, superfi-
cially in conflict with Frege's assertion that only in the context of a sentence does a word have any
meaning, see G. Frege, The Foundations of Arithmetic (J.L. Austin trans. 1959), is not with?
out its detractorsand complexities. See, e.g., Wallace, Only in the Context of a Sentence Do Words
Have Any Meaning, in Contemporary Perspectives in the Philosophy of Language 305 (P.
French, T. Uehling, Jr., & H. Wettstein eds. 1979).
Still, pace Fuller, supra note 53, at 662-63, sentences and paragraphs can have literal and even
acontextualmeaning insofar as an entire sentence or paragraphmay supply enough context to make
its meaning comparativelyclear. As texts become lengthier and richer it is often more possible to
understandthose texts without departingfrom them and thus more possible for them to have acontex?
tual meaning.
Moreover, literal meaning need not always be ordinary meaning. Where some aspect of the mini?
mal and uncontestedcontext makes it plain that a settled specializedor technical meaning of a term or
phrase applies, that technical meaning, rather than the ordinary usage of the man on the Clapham
omnibus, is controlling.

1988] Formalism 539
onstrating that formalism is ruleness rescue formalism? Restated, what is

so good about decision according to rules?
The simple answer to this question, and perhaps also the correct one, is
"nothing." Little about decision constrained by the rigidity of rules seems
intrinsically valuable. Once get in the way, that
we understand that rules
they gain their ruleness by cutting off access to factors that might lead to
the best resolution in a particular case, we see that rules function as im-
pediments to optimally sensitive decisionmaking. Rules doom decision?
making to mediocrity by mandating the inaccessibility of excellence.
Nor is there anything essentially just about a system of rules. We have
scarce reason to believe that rule-based adjudication is more likely to be
just thanare systems in which rules do not block a decisionmaker, espe?
cially a just decisionmaker, from considering every reason that would as?
sist her in reaching the best decision. Insofar as factors screened from con?
sideration by a rule might in a particular case turn out to be those
necessary to reach a just result, rules stand in the way of justice in those
cases and thus impede optimal justice in the long term. We equate Solo-
mon's wisdom with justice not because Solomon followed the rules in solv-
ing the dispute over the baby but because Solomon came up with exactly
the right solution for that case. We frequently laud not history's rule fol-
lowers, but those whose abilities at particularized decisionmaking tran?
scend the inherent limitations of rules.
Still, that rules
may be in one sense unjust, or even that they may be
inappropriate in much of what we call a legal system, does not mean there
is nothing to be said for rules. One of the things that can be said for rules
is the value variously expressed as predictability or certainty. But if we
pursue the predictability theme, we see that what most arguments for
ruleness share is a focus on disabling certain classes of decisionmakers
from making certain kinds of decisions.86 Predictability follows from the
decision to treat all instances failing within some accessible category in the
same way. It is a function of the way in which rules decide ahead of time
how all cases within a class will be determined.
Predictability is fostered to the extent that four different requirements

are satisfied. The first of the factors contributing to predictability is the
capacity on the part of those relying on a rule to identify certain particu?
lars as instances of a given category (for example, that pelicans are birds).
When there is a more or less uniform and uncontroversial ability to say
86. See generally A. Kocourek, An Introduction to the Science of Law 165-85 (1930);
R. Wasserstrom, The Judicial Decision: Toward a Theory of Legal Justification 60-66
(1961); Marsh, Principle and Discretion in theJudicial Process, 68 Law Q. Rev. 226 (1952). Along
with Wasserstrom,supra, at 61, I object to the use of the word "certainty"in this context because,
unlike the term "predictability,"it suggests that no doubt is involved.I can predictthat it will snow in
Vermont this winter and rely on that predictionin making winter plans, yet still not be certain that it
will snow. Although one usage of "certain"does recognize variability, I prefer the term "predictabil?
ity" because its common usage implies such variability.

that some item is a member of some category, little in the way of poten?
tially variable judgment clouds the prediction of whether the rule will ap?
ply to this particular item. This relates to the second factor: that the deci?
sionmakers in the system will perceive those particulars as being members
of the same category perceived by the addressees and will be seen as so
perceiving by those affected. That is, people perceive pelicans as birds;
decisionmakers perceive pelicans as birds; and people know that deci?
sionmakers will perceive pelicans as birds. Third, the rule must speak in
terms of an accessible category. Predictability requires that a rule cover a
category whose
denotation is substantially noncontroversial among the
class of addressees of the rule and common to the addressees of the rule
and thosewho apply it. Finally, the rule must treat all members of a
category in the same way. Only if the consequences specified in the apod-
osis of the rule are as accessible and noncontroversial as the coverage spec?
ified in the protasis can a rule produce significant predictability of appli?
cation. Thus, predictability comesfrom the knowledge that if this is a bird
a certain result will follow, and from the confidence that what I now per?
ceive to be a bird will be considered a bird by the ultimate decisionmaker.
This predictability comes at a price.87 Situations
only may arise in
which putting this particular that into
category seems just too
crude?something about this particular makes us desire to treat it spe-
cially. This vehicle is merely a statue, emits no fumes, makes no noise,
and endangers no lives; it ought to be treated differently from those vehi?
cles whose characteristics mesh with the purpose behind the rule. Serving
the goal of predictability, however, requires that we ignore this difference,
because to acknowledge this difference is also to create thepower?the
jurisdiction?to determine whether this vehicle or that vehicle actually
serves the purpose of the "no vehicles in the park" rule. It is the jurisdic?
tion to determine that only some vehicles fit the purpose of the rule that
undermines the confidence that all vehicles will be prohibited. No longer
is it the case that anything that is a vehicle, a moderately accessible cate?
gory, is excluded. Instead, the category is now that of vehicles whose pro?
hibition will serve the purposes of the "no vehicles in the park" rule, a
potentially far more controversial category.
Thus, the key to understanding the relationship of ruleness to predict?
ability is the idea of decisional jurisdiction.88 The issue is not whether the
statue serves the purpose of the "no vehicles in the park" rule. It is
whether giving some decisionmaker jurisdiction to determine what the
87. See, e.g., H.L.A. Hart, supra note 1, at 121-32.

88. On jurisdiction in this sense, see Schauer, Slippery Slopes, 99 Harv. L. Rev. 361, 367-68
(1985). See also Fried, Two Conceptsof Interests: Some Reflections on the Supreme Court's Balanc?
ing Test, 76 Harv. L. Rev. 755, 759-65, 771 (1963) (noting that courts define their competenciesin
the process of making substantive decisions); Rawls, Two Concepts of Rules, 64 Phil. Rev. 3, 10
(1955).

1988] Formalism 541
rule's purpose is (as well as jurisdiction to determine whether some item

fits that purpose) injects a possibility of variance substantially greater than
that involved in giving a decisionmaker jurisdiction solely to determine
whether some particular is or is not a vehicle. Note also that the jurisdic?
tional question has a double aspect. When we grant jurisdiction we are
first concerned with the range of equally correct decisions that might be
made in the exercise of that jurisdiction. If there is no authoritative state?
ment of the purpose behind the uno vehicles in the park" rule, granting
jurisdiction to determine that purpose would allow a decisionmaker to de?
cide whether the purpose is to preserve quiet, to prevent air pollution, or
to prevent accidents, and each of these determinations would be equally
correct. In addition to increasing the range of correct decisions, however,
certain grants of jurisdiction increase the likelihood of erroneous determi?
nations. Compare "No vehicles in the park" with "The park is closed to
vehicles whose greatest horizontal perimeter dimension, when added to
their greatest vertical
perimeter dimension, exceeds the lesser of (a) sixty-
eight feet, six inches and (b) the greatest horizontal perimeter dimension,
added to the greatest vertical perimeter dimension, of the average of the
largest passenger automobile manufactured in the United States by the
three largest automobile manufacturers in the preceding year." The sec?
ond adds no inherent variability, but it certainly compounds the possibility
of decisionmaker error. Creating the jurisdiction to determine whether the
purposes of a rule are served undermines predictability by allowing the
determination of any of several possible purposes; in addition, the creation
of that jurisdiction engenders the possibility that those who exercise it
might just get it wrong.
Grants of decisional jurisdiction not only increase permissible variance
and the possibility of "computational" error, they also involve deci?
sionmakers in determinations that a system may prefer to have made by
someone else.
We may believe that courts are less competent to make cer?
tain than other bodies; for example,
decisions we may feel that certain
kinds of fact-finding are better done by legislatures. There may also be
moral or political reasons to restrict the judge's discretion, for decision?
making implicatesprofound questions of just who in a given domain may
legitimately makecertain decisions. It is, for example, a plausible position
that the public rather than the University of Michigan philosophy depart?
ment should make the moral determinations involved in governing the
United States, even if the University of Michigan philosophy department
would make better choices.
Although decreasing the possibility of variance and error by the deci?

sionmaker contributes to the ability of addressees of rules to predict the
consequences of application of those rules, limited variance can serve other
values as well. If decisionmakers are denied jurisdiction to determine
whether a particular instance actually justifies its inclusion in a larger

generalization or are denied jurisdiction to determine the best result on the

basis of all germane factors, the part of the system inhabited by those
decisionmakers becomes more stable. Treating a large group of different
particulars in the same way?the inevitable byproduct of the generaliza?
tion of rules?dampens the range of variance in result by suppressing con?
sideration of a wide range of potentially relevant differences. Thus, stabil-
ity, not as a necessary condition for predictability but as a value in its own
right, is fostered by truncating the decisionmaking authority.
Because rule-bound decisionmaking is inherently stabilizing, it is inher?
ently conservative, in the nonpolitical sense of the word.89 By limiting the
ability of decisionmakers to consider every factor relevant to an event,
rules make it more difficult to adapt to a changing future. Rules force the
future into the categories of the past. Note the important asymmetry here,
the way in which rules operate not to enable but only to disable. A deci?
sionmaker can never exceed the optimal result based on all relevant fac?
tors. Thus, a rule-bound decisionmaker, precluded from taking into ac?
count certain features of the present case, can never do better but can do
worse than a decisionmaker seeking the optimal result for a case through
a rule-free decision.
Yet this conservatism, suboptimization, and inflexibility in the face of a
changing future need not be universally condemned. Rules stabilize by
inflating the importance of the classifications of yesterday. We achieve sta-
bility, valuable in its place, by relinquishing some part of our ability to
improve on yesterday. Again the issue is jurisdiction, for those who have
jurisdiction to improve on yesterday also have jurisdiction to make things
worse.90 To stabilize, to operate in an inherently conservative mode, is to
give up some of the possibility of improvement in exchange for guarding
against some of the possibility of disaster. Whether, when, and where the
game is worth the candle, however, cannot be determined acontextually.91
89. See Horwitz, The Rule ofLaw: An Unqualified Human Good? (Book Review), 86 Yale L.J.
561 (1977). I use the term "conservatism"to refer to the desire to hold onto the past or present in the
face of pressures to change. This usage bears only a contingent connection to the range of political
views now labeled "conservative."Left-wing conservatismis not oxymoronic,becauseone can imagine
left-wing systems adopting preservational(conservative)strategies or systems to prevent movement
away to the right.
90. This is not a logical truth. Grants of jurisdiction can incorporatesubstantive requirements.
Dworkin, Non-Neutral Principles, in Reading Rawls: Critical Studies of A Theory of Jus?
tice 124 (N. Daniels ed. 1975). Insofar as some grants of jurisdiction aim to increase the ability of
decisionmakersto adapt to an unknown future, however, they will be comparativelyopen-ended. It is
this open-endedness,whether couched in substantive(do good) or less substantive(determinethe pur?
pose) terms, that creates the possibility of unintended and uncontrollablevariance.
91. I therefore disagree with Kennedy, Form and Substance in Private Law Adjudication, 89
Harv. L. Rev. 1685 (1976), insofar as he argues that ruleness is acontextually individualistic and
particularizationis acontextuallyaltruistic. Even if there is truth in Kennedy's acontextuality,it still
is not clear that his analysis of the acontextual tendencies promotedby ruleness is correct. It is quite
plausible that the inherently stabilizing tendenciesof rule-bound adjudicationwill dampen individual
differences,stifle claims to special treatmentas an individual, and encouragedecisionalmodesty rather
than decisionalarrogance.It could be argued quite sensibly that all these tendenciesfoster rather than
impede altruism.

1988] Formalism 543
In sum, it is clearly true that rules get in the way, but this need not
always be considered a bad thing. It may be a liability to get in the way of
wise decisionmakers who sensitively consider all of the relevant factors as
they accurately pursue the good.
However, it may be an asset to restrict
misguided, incompetent, wicked, power-hungry, or simply mistaken deci?
sionmakers whose own sense of the good might diverge from that of the
system they serve. The problem, of course, is the difficulty in determining
which characterization will fit decisionmakers; we must therefore decide
the extent to which we are willing to disable good decisionmakers in order
simultaneously to disable bad ones.
With these considerations in mind, let us approach formalism in a new
light. Consider some of the famous marchers in formalism's parade of
horribles,examples such as R. v. Ojibway, Fuller's statue of the truck in
the park, and the poor Bolognese surgeon who, having opened the vein of
a patient in the course of performing an emergency operation outdoors,
was prosecuted for violating the law prohibiting "drawing blood in the
streets."92 Each of these
examples reminds us that cases may arise in
which application of the literal meaning of words produces an absurd re?
sult. But now we can recast the question, for we must consider not only
whether the result was absurd
in these cases but also whether a particular
decisionmaker should
be empowered to determine absurdity. Even in cases
as extreme as these, formalism is only superficially about rigidity and ab?
surdity. More fundamentally, it is about power and its allocation.
Formalism is about
power, but is also about its converse?modesty. To
be formalistic as a decisionmaker is to say that something is not my con?
cern, no matter how compelling it may seem. When this attitude is ap?
plied to the budget crisis or to eviction of the starving, it seems objectiona?
ble. But when the same attitude of formalism requires judges to ignore the
moral squalor of the Nazis or the Ku Klux Klan in First Amendment
cases, or the guilt of the defendant in Fourth Amendment cases, or the
wealth of the plaintiff who seeks to recover for medical expenses occa?
sioned by the defendant's negligence, it is no longer clear that refusal to
take all factors into account is condemnable.
Modesty, of course, has its darker side. To be modest is at times good,

but avoiding authority is also avoiding responsibility. In some circum?
stances we want our decisionmakers to take charge and accept the conse?
quences of their actions.98 But it is by no means clear that just because it
is good for some people to take charge some of the time, that taking
charge, even accompanied by acceptance of responsibility, is a universal
92. The last example, from 1 S. Pufendorf, De Jure Naturae et Gentium Libri Octo
(1672), comes to us through United States v. Kirby, 74 U.S. (7 Wall.) 482, 487 (1868).
93. For a recent articulationof this view, see Michelman, Foreword: Traces of Self-Government,
100 Harv L. Rev. 4 (1986). A useful contrastis Christie, An Essay on Discretion, 1986 Duke L.J.
747.

good. "Fm in charge here" has a long but not always distinguished his?
tory. Part of what formalism is about is its inculcation of the view that
sometimes it is appropriate for decisionmakers to recognize their lack of
jurisdiction and to defer even when they are convinced that their own
judgment is best. The opposite of modesty is arrogance, not just responsi?
bility. True, modesty itself carries responsibility, because an actor behav-
ing modestly is participating and thus assisting
in the legitimacy of the
grant of authority to someone else. But this is a responsibility of a differ?
ent and limited kind. That one accepts partial responsibility for the deci?
sions of others does not entail the obligation to substitute one's judgment
for that of others.
The distinctive feature of rules, therefore, lies in their ability to be for?
mal, to exclude from consideration in the particular case factors whose
exclusion was determined without reference to the particular case at hand.
This formalism of rules is not only conceptually sound and psychologi-
cally possible, but it also, as I have tried to show, is on occasion norma-
tively desirable. Insofar as formalism disables some decisionmakers from
considering some factors that may appear important to them, it allocates
power to some decisionmakers and away from others. Formalism therefore
achieves its value when it is thought desirable to narrow the decisional
opportunities and the decisional
range of a certain class of decisionmakers.
I stress that all of this is compatible with agnosticism about how rule-
bound decisionmaking applies to legal systems in general, to particular
legal systems, or to particular parts of legal systems. It is far from a neces?
sary truth that legal systems must be exclusively or even largely operated
as rule-governed institutions. Judgments about when to employ formalism
are contextual and not inexorable, political and not logical, psychological
and economic rather than conceptual. It would blunt my point about the
simultaneously plausible and contingent nature of decision according to
rule to offer in this acontextual
setting my recommendations about what if
any parts of the orAmerican
any other legal system should operate in
such a fashion. My goal is only to rescue formalism from conceptual ban-
ishment. But having been readmitted to the community of respectable
ideas, formalism, or decisionmaking according to rule in any strong sense,
still has the burden of showing that it is appropriately used in a particular
decisional domain.
IV. The Degrees of Restriction
I have thus far presented formalism and maximally contextual particu-

larism as mutually exclusive opposites, incapable of coexisting within the
same decisional domain. It may therefore appear that the advantages of
formalism can be attained only within a system willing to accept some
proportion of preposterous results and only within a system willing to

1988] Formalism 545
have its decisionmakers ignore the novelty of the situations that come
before them. Accommodation between these two forms of decisionmaking
might be possible, however.
Let us contrast two cases, both arising out of the "No vehicles in the
park" rule. The first involves the statue of the truck erected by the veter?
ans' organization. The second golf cart, as quiet as a
involves an electric
bicycle, incapable of proceeding at greater than ten miles an hour, and
emitting no noxious fumes. Can these cases be distinguished? In both
cases, exclusion of the object under consideration would not seem to serve
any of the purposes behind the

rule, regardless of whether the purpose
was the suppression of noise, the reduction of noxious odors, the limitation
of high speeds, or the restriction of forms of conveyance likely to be dan?
gerous to pedestrians. Yet despite their similarity, there appears to be a
difference between the cases. The statue seems to lie more outside the
purpose behind the rule than the golf cart. If we assume that something
like twenty miles per hour is dangerous, a totally immobile vehicle is fur?
ther away from the danger point than one that can go ten miles per hour.
Similarly, a vehicle with a totally inoperative engine makes less noise and
emits no more noxious fumes than even an electric motor.
If the difference between the cases is a matter of degree, is there some
way of empowering a decisionmaker to draw the distinction without at the
same time discarding all of the formalist-inspired virtues from the deci?
sionmaking process? That is, can we empower the decisionmaker to over?
ride the rule when its application would be totally preposterous, but not
when its application, still outside the purpose of the rule, would fall short
of the preposterous?
The question is the same one that arises in discussions regarding higher
court scrutiny of lower court decisions, and lower court scrutiny of admin?
istrative decisions and state laws. Can a decisionmaker distinguish those
state interests that are "compelling" from those that are "important" from
those that are merely "rational"? Can a decisionmaker distinguish "proof
beyond a reasonable doubt" from "clear and convincing evidence" from "a
preponderance of the evidence"? Can decisionmakers distinguish "de
novo" review from review only for "abuse of discretion"? The question
raised by all of these standards and others is the same: Can we admit the
possibility of overriding some judgment while at the same time not open?
ing the door to unconstrained substitution of judgment?
It is debatable whether some form of deferential but genuine review is
possible. It might be argued that deferential but not toothless review is an
illusion. Once the reviewing decisionmaker has the authority to look at the
decision below with at least the possibility of overturning it, deference
becomes largely illusory. This hypothesis equates the review process with
Pandora's box: Once the record below is opened, the review is in reality
de novo, and the language of abuse of discretion?or compelling interest

or whatever?is used merely as a tag line when the decisionmaker wishes

to reach a conclusion different from that reached below.
An alternativehypothesis posits some ground between no review and
unfettered intrusiveness. There might be cases in which the presumption
in favor of the result below would cause the decision to stand. Under this
hypothesis, we can have rebuttable presumptions?cases in which the pre?
sumption might be overcome in particularly exigent circumstances but
nevertheless controls in many or even most cases.
If this latter hypothesis is correct, it is correct as a contingent empirical
matter and not as a necessary truth. My instincts are that it is sometimes
correct?that at some times in some domains presumptions can matter
without being irrebuttable. This conclusion is based on my also instinctive
view that presumptions create attitudes, and that attitudes can matter. I
believe, for example, that I am more likely to admire an item of clothing if
I discover it myself than if my mother tells me she saw it in a shop and it
would look very nice on me. But I might be wrong. Even if I am right
about clothing and mothers, those attitudes might not carry over to real
decisions by real decisionmakers, and even if it does, it might be empiri-
cally false more often than it is empirically true.
Moreover, even if attitudes can be changed, it may be that linguistic
instructions are not particularly effective in accomplishing those changes.
The observation that linguistic instructions to adopt a certain attitude are
in fact potent is not universally proved by the observation that such in?
structions are sometimes potent, any more than the observation that such
instructions are sometimes impotent proves that they are never potent.
Given all of this, let me satisfy myself here with the unproved empirical
conclusion that linguistic instructions are sometimes potent.
If suchinstructions sometimes create presumptions, and if those pre?
sumptions sometimes work, then what does this say about the possibility
of what we might call a presumptive formalism? In order to construct
such a model, we would want to equate the literal mandate of the most
locally applicable written rule with
the judgment of the court below. The
court below can be taken to have
determined, for example, that in one
case operable and operating automobiles are excluded from the park, in
another case golf carts are excluded from the park, and in a third case
immobile statues of trucks are excluded from the park. We can then
equate the reviewing court with a determination of the correct result from
the perspective of the reasons behind the rule rather than the literal lan?
guage of the rule itself. We might conclude that in the first case even a de
novo application of the reasons would generate the same result as gener?
ated by the formalistic reading, and therefore the formal mandate would
prevail uncontroversially. In the second, a de novo application of reasons
would generate a different result than that generated by the rule, but the
result generated by the rule remains "in the ballpark" and therefore is

1988] Formalism 547
upheld despite its divergence from the result that would be reached by
direct application of the reasons. In the third, however, a de novo applica?
tion of the reasons indicates that the result generated by the rule is so far
out of bounds, so absurd, so preposterous that it is analogous to an abuse
of discretion and would therefore be reversed?the rule would not be ap?
plied in this case.
Under a theory of presumptive
such formalism there would be a pre?
sumption in favor
of the result generated by the literal and largely acon-
textual interpretation of the most locally applicable rule. Yet that result
would be presumptive only, subject to defeasibility when less locally appli?
cable norms, including the purpose behind the particular norm, and in?
cluding norms both within and without the decisional domain at issue,
offered especially exigent reasons for avoiding the result generated by the
presumptively applicable norm.
Such a system would bring the advantages of predictability, stability,
and constraint
of decisionmakers commonly associated with decision ac?
cording to rule, but would temper the occasional unpleasant consequences
of such a system with an escape route that allowed some results to be
avoided when their consequences would be especially outrageous. Such a
system would not be without cost. First of all, the escape route would
necessarily decrease the amount of predictability, stability, and deci?
sionmaker restraint. In short, it would diminish the amount of ruleness by
placing more final authority in the decisionmaker than in the rule. Sec?
ond, the presumptive force attached to the formalist reading of the appli?
cable norms would still result in some odd or suboptimal results. In this
sense, such a system would fail to honor all of the goals either of unre?
strained particularism or unrestrained formalism. Finally, such a system
would risk collapse into one in which the presumptions were for all prac?
tical purposes either absolute or nonexistent.
Even on the assumption that such a system might be desirable in some
decisional domains, this does not mean that all or part of what we com?
monly call the legal system might be one of those domains. It might be
that formalism, even only presumptively, is a good idea, but that the goals
of the legal system, in light of the decisions we ask it to make, are such
that it ought not to be designed along such a model. More likely, formal?
ism ought to be seen as a tool to be used in some parts of the legal system
and not in others. Determining which parts, if any, would be susceptible
to such treatment is not my agenda here, for what I have attempted to
offer is only an argument that formal systems are not necessarily to be
condemned. That is not to say they are universally or even largely to be
applauded, nor that they are to be pervasive or even frequent within that
segment of society we call the legal system. To answer this last question
we must ask what the legal system, in whole or in part, is supposed to do,

for only when we answer that question can we determine what kinds of
tools it needs to accomplish that task.
V. Conclusion
I have concluded this analysis by venturing no more than a prolegome-

non to a theory of presumptive formalism, which, to avoid the pejorative
(or at least to select a slightly less pejorative pejorative) might be called a
theory of presumptive positivism. As I have said, to urge the potential
advantages of such a view in some domains is to say little if anything
about whether the domain of decisions of judges or the domain of deci?
sions of the political state backed
by force is amenable to presumptively
positivistic decisionmaking. But even if we put aside the question of con?
crete applications, the presumptiveness that is central to this model may
illuminate one final usage of the word "formalistic" in its pejorative guise.
It may be that, in practice, to condemn an outlook as formalistic is to
condemn neither the rule-based orientation of a decisional structure nor
even the inevitable over- and under-inclusiveness sys? of any rule-based
tem. It may be to condemn such a system only to be when it is taken
absolute rather than presumptive, when it contains no escape routes no
matter how extreme the circumstances. Such a usage of "formalism" is of
course much narrower than is commonly seen these days. But with that
narrower usage we see that formalism is no longer something to be
roundly condemned, but rather, like the relation of fanaticism to enthusi-
asm, or bullheadedness to integrity, merely the extreme and therefore un-
fortunate manifestation of a fundamentally desirable characteristic. If we
recognize that, we may ultimately cease to use the epithetical deployment
of "formalistic" as a substitute for argument and turn instead to the cen?
tral questions involved in determining what, if anything, lies at the heart
of the idea of law.


Frederick Schauer - Formalism

Uploaded by

Copyright:

Available Formats

Frederick Schauer - Formalism

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Frederick Schauer - Formalism

Uploaded by

Copyright:

Available Formats

The Yale Law Journal Company, Inc.

This content downloaded from 147.9.241.57 on Tue, 9 Sep 2014 02:21:01 AM

With accelerating frequency, legal decisions and theories are con?

This content downloaded from 147.9.241.57 on Tue, 9 Sep 2014 02:21:01 AM

This content downloaded from 147.9.241.57 on Tue, 9 Sep 2014 02:21:01 AM

even ever be formalistic. Nevertheless, I do want to urge a rethinking of

I. Formalism as the Denial of Choice

A. Choice Within Norms

Few decisions are charged with formalism as often as Lochner v. New

This content downloaded from 147.9.241.57 on Tue, 9 Sep 2014 02:21:01 AM

it attempts to describe this choice as compulsion.6 What strikes us clearly

6. This was noted by Holmes in his now-famousobservation,"General propositionsdo not decide

This content downloaded from 147.9.241.57 on Tue, 9 Sep 2014 02:21:01 AM

nomic and political class) or intentionally chose to hide it.

This content downloaded from 147.9.241.57 on Tue, 9 Sep 2014 02:21:01 AM

guage of definitional inexorability obscures that choice and thus obstructs

vasively indeterminate. It is not that such terms have no content whatso-

More commonly, however, the indeterminacy to be filled by a deci-

Hart's conception of formalism15 is closely aligned with that under-

11. See infra note 85.

This content downloaded from 147.9.241.57 on Tue, 9 Sep 2014 02:21:01 AM

B. Choice Among Norms

Implicit in Hart's conception of formalism is the view that in the core,

This content downloaded from 147.9.241.57 on Tue, 9 Sep 2014 02:21:01 AM

the necessity of signing forms consenting to his nomination. In fact, these

Ultimately, Hunter prevailed, and it appears that Ryshpan v. Cashman

22. 132 Vt. 628, 326 A.2d 169 (1974).

This content downloaded from 147.9.241.57 on Tue, 9 Sep 2014 02:21:01 AM

C. Is There Always a Choice?

Ryshpan v. Cashman is a trifle obscure, but it is hardly unique. Con?

This content downloaded from 147.9.241.57 on Tue, 9 Sep 2014 02:21:01 AM

unclean hands or laches,80 the legal principle of in pari delicto,*1 or the

This content downloaded from 147.9.241.57 on Tue, 9 Sep 2014 02:21:01 AM

misstated to the disadvantage of another; a decision against the clerk

Alternatively, we could develop a model of a system in which there is

This content downloaded from 147.9.241.57 on Tue, 9 Sep 2014 02:21:01 AM

These two models?one allowing the creation of rule-avoiding norms,

II. Formalism as the Limitation of Choice

A. Can Language Constrain?

Each of my variations on Hunter v. Norman presupposed that the

This content downloaded from 147.9.241.57 on Tue, 9 Sep 2014 02:21:01 AM

rule.39 This rule-independent conclusion presents the possibility that the

This content downloaded from 147.9.241.57 on Tue, 9 Sep 2014 02:21:01 AM

Formalism as the linguistic limitation of choice can be illustrated in a

pretation, 1 Poetics Today 23 (1979).

This content downloaded from 147.9.241.57 on Tue, 9 Sep 2014 02:21:01 AM

conjoins two different elements: mechanical deducibility and the existence

Having posited a model in which the decisionmaker's choice is limited

by rules, we now must determine whether this model is descriptively accu?

up first, is whether such limitation of choice by the words on a printed

This content downloaded from 147.9.241.57 on Tue, 9 Sep 2014 02:21:01 AM

52. Hart, supra note 10, at 607.

This content downloaded from 147.9.241.57 on Tue, 9 Sep 2014 02:21:01 AM

tations of his challenge to the theory of linguistic constraint.54 One inter?

This content downloaded from 147.9.241.57 on Tue, 9 Sep 2014 02:21:01 AM

This content downloaded from 147.9.241.57 on Tue, 9 Sep 2014 02:21:01 AM

This content downloaded from 147.9.241.57 on Tue, 9 Sep 2014 02:21:01 AM

not of frogs or Oldsmobiles precisely because those marks, themselves,

community of speakers of the English language is itself a context. Yet

enough of these understandings exist to create the possibility of literal lan?