LEGAL INDETERMINACY AND CONSTITUTIONAL INTERPRETATION

Law and Philosophy Library

VOLUME 37

Managing Editors

FRANCISCO J. LAPORTA, Department of Law, Autonomous University of Madrid, Spain

ALEKSANDER PECZENIK, Department of Law, University of Lund, Sweden

FREDERICK SCHAUER, John F. Kennedy School of Government, Harvard University, Cambridge, Mass., U.S.A.

Former Managing Editors AULIS AARNIO, MICHAEL D. BAYLESt, CONRAD D. JOHNSONt, ALAN MABE

Editorial Advisory Board

AULIS AARNIO, Research Institute for Social Sciences, University ofTampere, Finland

ZENON BANKOWSKY, Centre for Criminology and the Social and Philosophical Study of Law, University of Edinburgh

PAOLO COMANDUCCI, University of Genua, Italy ERNESTO GARZON VALDES, Institutfiir Politikwissenschaft,

Johannes Gutenberg Universitiit Mainz JOHN KLEINIG, Department of Law, Police Science and Criminal

Justice Administration, John Jay College of Criminal Justice, City University of New York

NEIL MacCORMICK, Centre for Criminology and the Social and Philosophical Study of Law, Faculty of Law, University of Edinburgh

WOJCIECH SADURSKI, Faculty of Law, University of Sydney ROBERT S. SUMMERS, School of Law, Cornell University

CARL WELLMAN, Department of Philosophy, Washington University

The titles published in this series are listed at the end of this volume.

JOSE JUAN MORESO University o/Girona,

Girona, Spain

LEGAL INDETERMINACY

AND CONSTITUTIONAL INTERPRETATION

SPRINGER-SCIENCE+BUSINESS MEDIA, B.V.

A C.l.P. Catalogue record for this book is available from the Library of Congress.

ISBN 978-90-481-5061-8 ISBN 978-94-015-9123-2 (eBook) DOI 10.1007/978-94-015-9123-2

Printed on acid-free paper

English translation of Jose Juan Moreso, La indetenninaci6n deL derecho y La

interpretaciOn de La Constituci6n, Centro de Estudios Politicos y Constitucionales, Madrid, Spain, 1997.

Translated by Ruth Zimmerling

All Rights Reserved © 1998 Springer Science+Business Media Dordrecht

Originally published by Kluwer Academic Publishers in 1998 Softcover reprint of the hardcover 1st edition 1998

No part of the material protected by this copyright notice may be reproduced or utilized in any form or by any means, electronic or mechanical,

including photocopying, recording or by any information storage and retrieval system, without written permission from the copyright owner

To Lidia, sine qua non.

Truth was their model as they strove to build a world of lasting objects to believe in.

w. H. Auden

PREFACE

In this book, I present the results of an investigation which began with an extended stay at Oxford's Balliol College during the first half of 1995. My visit to Oxford was made possible by a grant from the Spanish Ministerio de Educaci6n y Ciencia.

My sincere thanks go to Joseph Raz who served as my supervisor in Oxford. For several points of the present study, conversations with Timothy Endicott in Oxford were also of great help.

The book is part of a larger project of investigation, directed by Albert Calsami­glia, which is a joint effort of a group of legal philosophers from the Universitat Pom­peu Fabra (Barcelona) and the Universitat de Girona, and which also receives financial support from the Spanish Ministerio de Educaci6n y Ciencia.

An earlier version of the manuscript was presented in June 1996 to the selection committee for a tenured professorship in Legal Philosophy at the Universitat de Girona. The members of the committee were Francisco Laporta, Albert Calsamiglia, Gregorio Peces-Barba, Camilo J. Cela Conde, and Francesca Puigpelat. I am grateful to all of them for their comments, which have been extremely useful in preparing the final ver­sion of this book.

I also had the opportunity to discuss parts of my ideas with different audiences: in August 1996, at the Universidad de Buenos Aires and the Universidad de Palermo (Buenos Aires) as well as at the Encuentro Anual de Filosoffa del Derecho at the Va­querfas complex of the Universidad Nacional de C6rdoba (Argentina), on invitations by Eugenio Bulygin, Ricardo Guibourg and Ricardo Carraciolo; in September 1997, at the VII Seminario Eduardo Garcia Maynez organized by the Instituto Tecnol6gico Aut6-nomo de Mexico (Mexico City), on an invitation by Rodolfo Vazquez; and in Decem­ber 1997, at the Instituto de Ciencias Polfticas y Sociales of the Universidad Aut6noma de Barcelona, on an invitation by Isidre Molas.

In addition, comments by Carlos Alchourron - who, unfortunately, is no longer with us -, Ernesto Garz6n Valdes, Daniel Mendonca and Stanley Paulson helped me clarify my own views and thus, I hope, improve the final version. Eugenio Bulygin, Ricardo Caracciolo and Riccardo Guastini read the entire manuscript and formulated sharp objections which helped me correct substantial deficits. With Victor Ferreres, Pablo Navarro and Cristina Redondo, who also read the entire manuscript, I have dis­cussed almost all the questions treated in the book. Working with them has been, and continues to be, one of my greatest intellectuals stimulants.

I also wish to thank all those who made the publication of this English version of the book possible: Ernesto Garz6n VaIdes, who encouraged me to present the manu­script to Kluwer Academic Publishers; Ruth Zimmerling, who translated the book and with keen observations and numerous questions forced me to formulate a number of points more clearly; and, last but not least, several institutions which generously bore the costs of the translation: the Fundaci6n Cultural Enrique Lufio Pefia (Sevilla) and its director, Antonio-Enrique Perez Lufio; the Instituto de Ciencias Polfticas y Sociales (Barcelona), directed by Isidre Molas; and the Departamento de Filosoffa del Derecho at the Universitat Pompeu Fabra (Barcelona) with its director, Albert Calsamiglia. I also

x PREFACE

owe thanks to the publishers of the Spanish version of the book, the Centro de Estudios Politicos y Constitucionales (Madrid), for ceding the rights of the English publication.

Finally, I wish to point out that in Appendices A and B to Chapter I, I use some ideas already presented in MoresolNavarro (1996c) and Moreso (1996), respectively, as well as, in Chapter III, some notions presented earlier in Moreso (1994b), MoresolNa­varro (1996a) and MoresolNavarro (1996b).

Girona, March 1998 Jose Juan Moreso

TABLE OF CONTENTS

PREFACE

INTRODUCTION

CHAPTER I: LEGAL NORMS AND LEGAL SYSTEM

1.

2.

3.

4.

Introduction

The Theory of Sense for Prescriptions

a) Convention-T 81 b) A semantics for deontic logic 9 1 c) Does convention-T provide a complete theory of meaning? 10

The Theory of Force, and Prescriptions

a) The illocutionary force of prescriptions 12 / b) Intentions and meanings 14/ c) Re­levance and implicit meaning 16

Conclusions

ApPENDIX A: A FORMAL THEORY OF SENSE FOR PRESCRIPTIONS

a) Language LN

b) Open Sentences

c) Satisfaction and Compliance

i) A subset of LN 25/ ii) Corresponding d-formulae 25 / iii) Satisfaction and the truth of d­formulae 27 / iv) Compliance and the efficacy of d-formulae and i-formulae 28

ApPENDIX B: RELEVANCE, ARGUMENT A TION AND NORMATIVE SYSTEMS

a) Relevance Beyond Logic

b) The Logic of Relevance

c) Logic Plus Relevance

d) Relevance and Normative Systems

CHAPTER II: A LOGICAL ANALYSIS OF LEGAL PROPOSITIONS

l.

2.

3.

4.

5.

6.

Introduction

Deontic Legal Statements and Conceptual Legal Statements

Pure Legal Statements and Applicative Legal Statements

Truth-Conditions of Legal Statements

Legal Statements and Contradictions: The Law Speaks With Many Voices

Legal Statements and Normative Gaps: The Law Is Silent

ix

3

3

7

12

19

21

21

23

25

32

32

36

40

44

47

47

47

48

49

56

57

xii TABLE OF CONTENTS

a) Kelsen: gaps as fiction 591 b) Dworkin and the completeness of the law 631 c) Legal reasons, sources and gaps 66

7. Legal Statements and Vagueness: The Law Speaks With an Uncertain Voice 72

8.

9.

a) Accepting the conclusion of sorites 751 b) Rejecting the argument: Degrees of truth 77 1 c) Rejecting the premises 79

Excursus on Gaps of Cognition

Conclusions

ApPENDIX:

TRUTH-CONDITIONS AND THE LOGICAL ANALYSIS OF LEGAL STATEMENTS

CHAPTER III: LEGAL DYNAMICS

1.

2.

3.

4.

5.

6.

Introduction: Legal Order and Legal System

Identifying Legal Orders

The Applicability of Legal Norms

Applicable But Invalid Norms

a) Customary norms 1091 b) Standards of morality 1091 c) Irregular norms 110 1 d) Re­ceived Norms 111

Valid But Inapplicable Norms

a) Cancelled norms 1121 b) Irrelevant derived norms 113

Conclusions

CHAPTER IV: THE PRIMACY OF THE CONSTITUTION

1.

2.

3.

4.

The Constitution as Precommitment

Possible Constitutional Worlds

The Thesis of Constitutional Accessibility

Conclusions

CHAPTER V: CONSTITUTIONAL INTERPRET A TION

1.

2.

3.

Introduction

The Noble Dream

a) The Noble Dream of metaphysical realism 1351 b) Dworkin: 'The Noblest Dreamer' 1391 c) Critique 143

The Nightmare

a) Skepticism as therapy: Jerome Frank 1481 b) Skepticism in Critical Legal Studies: .. Law is Politics" 1491 c) Interpretation as creation 1511 d) Critique lS2

86

88

91

101

101

103

105

108

112

114

117

117

119

121

128

131

131

134

147

4.

5.

6.

7.

TABLE OF CONTENTS

The Vigil

Intention in Constitutional Interpretation

Judicial Review of the Constitutionality of the Laws

Conclusions: Modest Objectivity and Limited Indeterminacy

BIBLIOGRAPHY

INDEX OF NAMES

Xlll

156

160

167

169

173

197

INTRODUCTION

In recent years, we have witnessed the resurgence of a strong interest in questions of legal interpretation. I One reason for this may be that interpretation is the key notion when it comes to determining the truth-value of statements like 'Legally, all F have the obligation to pay tax A' or 'Legally, x has the right to obtain damages B', etc. Such statements are commonly used by legal officials when they describe, or inform about, the rights of some social group, and they are what we expect to hear from a lawyer we consult about a legal problem or what we expect to be taught in Law School.

Despite their apparent simplicity, such statements have a complex logical struc­ture, and it is not always easy to determine their truth-conditions. Therefore, the main purpose of this book is to offer a logical analysis of that class of legal statements and to determine their truth-conditions.

Obviously, an answer to the question about the truth-conditions of such state­ments cannot be had without considering the question of the existence of legal norms. Generally, the foundation for the truth of a normative statement is the existence of cer­tain norms. In the words of G. H. von Wright (1983b, 68):

"One important type of answer to the question 'Why ought (may, must not) this or that be done?' is the fol­lowing: There is a norm to the effect that this thing ought to (may, must not) be done. The existence of the norm is here the foundation or truth-ground of the normative statement."

Now, the notion of existence of a norm requires detailed analysis. On the one hand, we must answer the question of what kind of entities norms are; on the other, we need a theory that can account for what it is that makes a norm a legal norm.

In the first chapter of this book, I will try to lay the foundations for an answer to both questions. For this purpose, we must adopt a theory of meaning for certain senten­ces and statements since, as we will see, the question about what kind of entities norms are will prove to be more a semantic than an ontological matter. The question about the legal nature of norms, in turn, will be answered, as it is common practice in legal theory at least since Kelsen, with the help of the notion of a 'legal system'.

The objective of the second chapter is a logical analysis of legal propositions, i. e., propositions expressed in statements like 'Legally, x ought to (may, must not) do 4>'. The analysis will be supplemented by an investigation of the truth-conditions of such statements. This will enable us to take a stand on one of the most important issues in re­cent scholarship on legal interpretation, concerning the notion of legal indeterminacy and its relation to gaps, contradictions and the vagueness of legal concepts.

The third chapter will show that the simple model of a legal system is not suffi­cient to account for the complexity of legal propositions referring to legal systems of some degree of maturity. Several notions from legal dynamics will be presented in

I In a collection of papers published very recently, the editor, A. Marmor (1995b, v) begins his preface say­ing: "Interpretation has become one of the main intellectual paradigms of legal scholarship in the last fifteen years. Like the interest in rules during the 1960s and in legal principles during the 1970s, much of the legal theorizing in the last decade has been built around the concept of interpretation."

2 INTRODUCTION

order to bring to light the importance of concepts like applicability or hierarchy for the determination of the truth-value of a legal proposition.

Once these ideas have been developed, we can, in the fourth chapter, introduce an approach to a central idea in the theoretical reconstruction of most contemporary le­gal systems, namely, that of the primacy of the constitution. A conceptual explication of this idea will be presented, and some conclusions from that explication will be drawn.

Finally, in the fifth and last chapter, a particular conception of constitutional in­terpretation will be proposed, which will then be compared with other, alternative con­ceptions. Special attention will be paid to the relationship between the interpretation and the indeterminacy of the law and, more specifically, to the problem of the discretion en­joyed by the organs entrusted with applying the constitution. We can then also analzye several theses that have been discussed controversially in the context of constitutional interpretation, as, for example, about the relevance of intentions for the interpretation of the constitution and for the justification of judicial review.

I. LEGAL NORMS AND LEGAL SYSTEM

1. Introduction

Norm-authorities perform certain acts the result of which is that certain norm-formula­tions are issued. By attributing meaning to such formulations, we can assess the deontic status of certain actions performed by their addressees or norm-subjects. I

Norm-formulations can consist in at least two basic kinds of statements: (i) pre­scriptions, i. e., statements commanding, prohibiting, or permitting some conduct, and (ii) definitions, i. e., statements explicating the meaning of certain terms used in other norm-formulations. The meaning of a prescription I will call a prescriptive rule or simply a norm; the meaning of a definition I will call a conceptual rule.2 Just as a pro­position is the meaning of a declarative statement (or sentence), a norm is the meaning of a prescription.

One of the questions that immediately arise at this point is that about the rela­tionship between norm-formulations and norms: What is the criterion for specifying norms? I will follow the well-known idea of A1chourr6n and Bulygin (1971, 42) ac­cording to which norms are meanings of sentences correlating generic cases with nor­mative solutions. This conception presupposes a distinction between generic and indi­vidual cases. An individual case is an instance, an occurrence at a certain time and place, of a generic case. Generic cases are classes of events or states of affairs defined by some property. For instance, there is a clear difference between the case of the politi­cal murder of John F. Kennedy and political murder as an abstract category. A1chourr6n and Bulygin introduce the distinction between generic and individual cases as follows:

,,[T]he term 'case' is ambiguous in legal language as it is in ordinary language in general. Thus, for example, we speak about the case of political murder and the case of the murder of Mahatma Gandhi, of the case of divorce and the case of the divorce of Brigitte Bardot ... It is obvious that the word 'case' has not the same meaning in these phrases. Gandhi's murder is a real event, that happened in a certain place and at a certain moment in time. The expression 'case of political murder' does not refer to any concrete event; it is the mere description of certain properties which certain events may have. The property of being a political murder may be instantiated in an unlimited number of concrete occurrences." (Alchourr6n1Bulygin 1971, 28)

Norms, thus, assign normative consequences to generic cases and in this way enable us to know the deontic status of certain individual cases which are instances of those gene­ric cases.3 The deontic status of a generic case is determined by a universe of solutions. The universe of solutions deontically modalizes a universe of act-categories. The pos-

I For the concepts of norm-authority and norm-subject, cf. von Wright 1963a, 75-79.

2 The term 'definition' must be understood in a broad sense, such that it not only includes semantic rules, but also other kinds of rules that can (partially or completely) explicate the meaning of certain concepts. Cf. on this MendoncalMoreso/Navarro 1995, 226 f.

3 Here, I will not go into the question of how generic cases are constructed on the basis of a selection from a universe of relevant properties. On this, cf. Aichourr6n1Bulygin 1971, 22-27.

4 CHAPTER I

sible deontic characters are: P (permitted), Ph (prohibited), 0 (obligatory), and F (fa­cultative), the latter being defined as 'permitted to do and permitted to omit'. Here, it is important to distinguish between maximal solutions and partial solutions. If we take the operator 'permitted' (P) as the basic one, we can formulate the following equivalences:

Op = Pp /\ ""P""p,

meaning that 'p is obligatory' is equivalent to 'p is permitted and not-p is not per­mitted';

meaning that 'p is prohibited' is equivalent to 'p is not permitted and not-p is per­mitted'; and

Fp = Pp /\ P""p,

meaning that 'p is facultative' is equivalent to 'p is permitted and not-p is permitted'. A maximal solution is a solution that determines all the contents corresponding

to some universe of actions. Given a single action p as the possible normative content, maximal solutions are the following: fOp, O""p, Php, Ph...,p, Fp, F...,p}. They can be re­duced still more, to only three elements, since Op is equivalent to Ph...,p, O""p to Php, and Fp to F...,p. A partial solution is one that does not determine all the contents corres­ponding to a universe of actions. Given a single action as the possible normative con­tent, minimal solutions, as a subset of partial solutions, are the following: fPp, P""p, ...,Fp}. In the first case (Pp), we do not know whether p is obligatory or facultative; in the second (P""p) , we do not know whether p is prohibited or facultative, and in the third case (...,Fp), we do not know whether p is prohibited or obligatory.

On the other hand, conceptual rules correlate generic cases with other generic cases. Thus, if we have a norm-formulation like 'Persons aged 18 years or older are of age', the generic case (Cl ) of being 18 years old or older is correlated with the case (C2)

of being of age, i. e., Cl = C2• The correlation must not necessarily be one of identity. Thus, when it is said that 'Rivers are public property', what happens is that the class of rivers (C) is included in the class of public properties (C4) , i. e. CJ C C4 • Generally, therefore, conceptual rules establish the following relationship between two generic cases Cj and Cj : Cj ~ Cj" Conceptual rules thus enable us to identify the content of cer­tain prescriptive rules, i. e., if there are two norm-formulations such as 'Persons aged 18 years or older are of age' and 'Persons of age must vote', then the case of being 18 years or older is correlated with the normative solution 'It is obligatory to vote'.

We now have a criterion for the identification of norms and conceptual rules which associates them with the existence of certain norm-formulations.4 But what gives

4 I have said that norm-formulations are the result of acts performed by some norm-authority. This presup­poses that there are norms only when there are authorities capable of establishing normative relationships

LEGAL NORMS AND LEGAL SYSTEM 5

those entities the property of being legal entities? My answer to this question is to say that what gives them this property is the fact that they belong to a certain set. Besides, that set can be given a structure, i. e., it is systematically ordered. The relation which defines the structure of that set is the relation of logical consequence. Thus, a set of norm-formulations can be regarded as a normative system - as a set, that is, which contains all its logical consequences, where at least one of them is a norm (Alchourr6n1 Bulygin 1971, 54-59). It is the relation of deductibility that converts a set into a system (Caracciolo 1988, 57; MoresolNavarro 1993a, 36 f.). Though one can also establish other relationships in a set of norms (all of them associated with the dynamic character of legal systems; cf. chap. III), in this chapter I will only look at sets of norms as static systems.

From what has been said so far, it should be clear that any set of norms can be structured as a deductive system. But the notion of a legal system is a very general one. We may be interested in the set of norms actually belonging to Spanish law, or in the set of norms actually applied by the courts in Spain (as we will see, these sets are not necessarily identical), or - as is usually the case - in more limited sets of norms (e. g., the set of norms regulating extra-contractual liability). All these sets can be presen­ted as systems structured by the relation of deductibility. The discussion of the criterion for the identification of a legal system must wait until later.

Some authors (e. g., Caracciolo 1996) have asked how there can be an associa­tion of norms (which, being propositions, are abstract entities) and systems (which, be­ing sets, also are abstract entities) with the social facts of norm-creation and norm-eli­mination that - in contrast to abstract entities - exist in a real position in space and time. But note that our linguistic usage may be confusing us here. We usually speak of the creation and elimination of norms, of systems that emerge and disappear, etc. Such statements must be understood in the following sense: We associate certain abstract entities (existing outside of time and space) with certain social facts. Thus, to say that a norm N has been created only means that we associate some act of some authority with a certain abstract entity. Thus, qua abstract entities, norms too cannot be cancelled. But we speak of the cancellation of a norm in the sense that after an act of derogation of norm N, we associate a new system of norms (that does not contain N as an element) with that moment in time. We can draw an analogy that can help us understand this. The set of beliefs of a person x can be seen as a deductive system (Hintikka 1962) con­sisting of the propositional contents of x' s beliefs and closed under the notion of logical consequence. When we say that at some time t x replaces belief p with belief "p, we do not mean to say that x cancels the existence of proposition p and creates the existence of proposition "p. In fact, since propositions are abstract entities, their existence cannot be affected by mental acts of x. What we mean to say is that we associate another set of

with norm-subjects. Cf. von Wright 1963a, 117 f. We can, however, abstract from this pragmatic require­ment and call a 'norm' any meaning of certain linguistic expressions of a prescriptive nature, just as a propo­sition is the meaning of any declarative sentence, even if no-one has ever used that sentence for stating any­thing. Cf. Alchourr6n1Bulygin 1979,1989.

6 CHAPTER I

propositions with time t than the one we associated with time t-l, namely, a set con­taining ""p instead of p.

The purpose of this chapter, however, is not to investigate such ontological ques­tions, but to present a possible explanation of how we can attribute meaning to prescrip­tions. The notion of meaning plays a crucial role in the philosophy of language. It is strongly linked to the notion of understanding. The meaning of an expression in some language is what a competent speaker understands by that expression (Platts 1979,43).

Since Frege (cf. the insistence on this point in Dummett 1978, 105 f. and 449 f.), it is common to distinguish two important elements in all theories of meaning: sense andforce. In Dummett's words:

"We thus arrive at the distinction, originally drawn by Frege, between the sense (Sinn) of a sentence and the force (Kraft) attached to it. Those constituents of the sentence which detennine its sense associate a certain state of affairs with the sentence; that feature of it which detennines the force with which it is uttered fixes the con­ventional significance of the utterance in relation to that state of affairs (i. e., according as the speaker is asserting that the state of affairs obtains, asking whether it obtains, commanding that it should obtain, expressing a wish that it obtain, etc.). It is difficult to see how a systematic theory of meaning for a language is possible without acknowledging the distinction between sense and force ... " (Dummett 1978, 449 f.)

Therefore, the purpose of this chapter is to offer a conception of the sense and a concep­tion of the force of prescriptive sentences.

Conceptual rules, on the other hand, can safely be said to be less problematic. Their role in a normative system is similar to that of definitions in axiomatic theories (where, in a normative system, the role of the axioms is occupied by the prescriptive rules). Actually, the association of certain legal properties, like being 'of age', with cer­tain natural properties, like being 18 years old, can be seen as a relation of superveni­ence. That relation is usually considered to be an a priori relation. According to Kim, the following three elements are distinctive of supervenience:

"Covariance: Supervenient properties covary with their subvenient, or base, properties. In particular, indis­cernibility in respect of the base properties entails indiscemibility in respect of the supervenient properties. Dependency: Supervenient properties are dependent on, or are determined by, their base properties. Nonreducibility: Supervenience is to be consistent with the irreducibility of supervenient properties to their base properties." (Kim 1993, 140)

The idea that the relations of covariance and dependency can be attributed to the rela­tionship between legal properties and natural properties seems to be widely accepted. More controversial is the notion of nonreducibility - in fact, it is a characteristic of su­pervenience that has been much discussed in the pertinent literature (cf. again Kim 1993, 149). But if we define legal properties as the circumstances under which a set of normative consequences comes about (e. g., being of age as the property with which we associate the right to vote, the capacity to make valid contracts, etc.), we can regard them as irreducible to natural properties, since natural properties must be defined in an­other way.

LEGAL NORMS AND LEGAL SYSTEM 7

2. The Theory of Sense for Prescriptions

Some of the most important achievements in the area of the logic and philosophy of language have been made in the field of semantics. In this context, the reconstructions of the concept of truth developed by Tarski (1944, 1956) and Davidson (1967, 1973) for formalized and for natural languages, respectively, deserve special mention. The ob­ject of their investigations was the sense and referent of declarative statements. Other kinds of statements, e. g., prescriptive or imperative statements, have received less attention. The resulting deficit acquires special relevance when the core of a theory of meaning - i. e., the theory of sense - depends on the notion of truth. In proposition 4.063 of the Tractatus, for instance, Wittgenstein asserts that

"in order to be able to say »'p' is true (or false)«, I must have determined in what circumstances I call 'p' true, and in so doing I determine the sense of the proposition" (Wittgenstein 1921, 24).

But if non-declarative statements have no truth-value, we must say what the link be­tween such statements and the theory of truth and, therefore, of meaning, is. Following Davidson (1979, 109), we can say that the question is "how it might be possible to re­present mood within the confines of a theory of truth".

In this section, I will treat the semantics of imperative or prescriptive statements. Some authors (HofstadterlMcKinsey 1939; Ross 1941; Dummett 1978,9; Platts 1979, 63; Searle 1979, 13 f.; Smart 1984, 14-19; Hernandez Marin 1989, 297 ff., 302 f.; Hierro S. Pescador 1990,59) have suggested that the semantic value of an imperative is (non-)compliance, and that this value is parasitic on truth.5

The basic intuition on this point is that an imperative i is effective if and only if the corresponding declarative statement d is true. As is well-known, Tarski used the re­lational notion of satisfaction in order to define the predicate 'to be true'. I will use the relation of compliance in order to define the notion of efficacy.6

The term 'efficacy' has been used in different ways in different contexts, e. g., in the legal or the moral context. This seems to suggest that the word expresses afamily of concepts. A common characteristic of many members of that family is that they refer to some relation R between the content of a prescription, e. g. the content of an imperative, and some state of affairs. One of my main objectives will be to explicate that relation R in an abstract and formally rigourous way. In order to offer a general approach that can serve as a guiding frame for different conceptual investigations, in this reconstruction of the concept of efficacy I will not give privileged treatment to anyone of the uses normally associated with that word. In this sense, it may be useful to recall the following assertion by Rudolf Carnap:

5 Hierro S. Pescador (1990, 59), e. g., asserts: "We need to elaborate, Ii La Tarski, a semantic theory of com­pliance with imperatives, or of any other semantic value we may introduce in order to account for other modes. As long as we use, as I have just done, the concept of truth as the criterion of comparison, the elabo­ration of a semantic theory of the new values introduced may not be difficult. But it must be done."

6 In Appendix A to this chapter, I will try to show how the notion of efficacy can be constructed with the help of a recursive definition of Tarski' s concept of truth.

8 CHAPTER I

"A philosophical thesis on logic or language, in contrast to a psychological or linguistic thesis, is not intended to assert anything about the speaking or thinking habits of the majority of people, but rather something about pos­sible kinds of meanings and the relations between these meanings. In other words, a philosophical thesis does not talk about the haphazard features of natural languages, but about meaning relations, which can best be re­presented with the help ofa constructed language." (Camap 1963, 1002)

a) Convention-T

The meaning of prescriptive or imperative statements is determined by the truth-condi­tions of certain statements that describe a world in which imperatives are always com­plied with - a deontically perfect world.

Earlier, I said that to determine the meaning of a sentence is to establish the con­ditions under which it is true. My starting point was to assume that convention-T

(T) X is true if and only if p

(where X is the name of p) is valid for the descriptive statements of natural languages. Similarly, we can propose a convention-E in order to give meaning to imperative or pre­scriptive sentences. Suppose X is an imperative sentence and X' is the corresponding declarative statement (the name of p'):

(E) X is effective if and only if X' is true.

Now, it is clear that (E) is parasitic on (T). Therefore,

(E') X is effective if and only if p'.

With the ideas presented before, convention-E can be refined for a language of deontic and propositional logic. For this, the following idea of von Wright is useful:

"We introduce the symbol A for 'always' and V for 'sometimes'. The second may also be regarded as an abbreviation for ...,1\...,; and the first as an abbreviation for..., V...,. The notions of 'sometimes' and 'always' I shall call temporal quantifiers. The scope of these quantifiers can also be relativized to a time-span of limited duration. Then 'I\p' says that the state that p obtains always throughout this span. The span can be, for example, the time during which a certain norm exists ... The variable 'p' in 'Op' or 'Pp' ... is a schematic representation for an open sentence expressing a generic proposition, e. g. 'p' = 'it is raining' ... Generic propositions are not 'by themselves' true or false; but 'Vp' and 'Ap' are closed sentences expressing true or false individual propositions - for example that it some­time(s) is raining or is always raining respectively." (von Wright 1983a, 160)

With the help of temporal quantifiers, we can say that the declarative statement corres­ponding to the prescription 'Op' is 'Ap', since for a norm of obligation to be complied with it must be complied with on every occasion as long as the norm exists; and the de­clarative statement corresponding to 'Pp' is 'Vp', since in the case of a permission it suffices that the permission is used on some occasion during the norm's existence. In this way, convention-E can be specified as follows, distinguishing norms of obligation and norms of prohibition from permissive norms:

LEGAL NORMS AND LEGAL SYSTEM

(E) 'Op' is effective if and only if '/\p' is true.? (E2) 'Pp' is effective if and only if 'Vp' is true.

Now, since because of convention-T

(T) '/\p' is true if and only if /\p and

(T2) 'Vp' is true if and only if Vp,

convention-E can be presented as follows:

(E/) 'Op' is effective if and only if /\p and

(E2') 'Pp' is effective if and only if Vp.

b) A semantics for deontic logic

9

This semantics provides an interpretation of deontic logic or the logic of i-formulae (cf. HofstadterlMcKinsey 1939). To say that a norm N implies a norm N' means that if N is effective, N' necessarily is effective too. For instance, 'Op' implies 'Pp', since' /\p' im­plies'Vp'.

Now, something must be said about mixed inferences. An inference is mixed when its premises contain at least (a) one d-formula and (b) one i-formula or m-formula (i. e., a formula containing d-formulae as well as i-formulae, e. g., 'p ~ Oq'). In that case, we can say that a set of true or effective premises necessarily implies an effective solution.

We must, however, point out the following restrictions intended to avoid that the resulting calculus violate Bume's law (cf. Weinberger 1991, 285):

(a) Only d-formulae can be deduced from sets of d-formulae. (b) If a set of premises contains an i-formula or an m-formula, the conclusions can only be i-formulae or m-formulae.

With these restrictions, deontic logic becomes an extension of classical logic. Its seman­tic legitimation is given by the notion of truth. This allows us to assume that the notion of logical consequence can be applied to imperatives.

Since the notion of a normative system presupposes that there are logical rela­tionships between norms, we must have some notion of logical consequence between norms that provides an escape from the well-known dilemma formulated by Jorgensen:

"According to a generally accepted definition of logical inference only sentences which are capable of being true or false can function as premises or conclusions in an inference; nevertheless it seems evident that a conclusion

7 Since 'Php' is equivalent to 'O-yJ', the convention for prohibitions is only an application of (E,): 'Php' is effective if and only if '/\--,p' is true.

10 CHAPTER I

in the imperative mood may be drawn from two premises one of which or both of which are in the imperative mood." (Jorgensen 1938, 290)

If one accepts this presentation of the dilemma - i. e., if one accepts that the meanings of imperative sentences have no truth-value -, then either a logic of norms is impossi­ble, or there is a logic beyond truth.

There are several ways in which the second horn of the dilemma can be con­ceived. Recently, the interesting version of adopting an abstract general approach and thus abandoning the primacy of semantics for the notion of logical consequence has been proposed (Alchourr6n1Martino 1987/88, Alchourr6n 1993). The version I will suggest here, however, is a semantic version. What must be shown is that in the case of norms efficacy is an appropriate substitute for truth, and that efficacy can be defined with the help of the notion of truth, i. e., that efficacy is parasitic on truth. This also seems to be the version proposed by Jorgensen:

"An imperative sentence has a meaning if and only if the corresponding indicative sentence which may be de­rived from it and which describes its contents is meaningful." (Jorgensen 1938,291)

c) Does convention-T provide a complete theory o/meaning?

Some philosophers, like Quine and Davidson, seem to think that the question of whe­ther convention-T provides a complete theory of meaning should be answered affirma­tively. Thus, Quine writes:

"You have given all the meanings when you have given the truth-conditions of all the sentences. Davidson took the connection to heart and drew this conclusion: the way to develop a systematic account of meanings for a language is to develop Tarski's recursive definition of truth for that language." (Quine 1969,33)

According to Davidson, a theory of meaning proves a sentence of the form's means that p' for every sentence in the object-language. And he adds:

"The theory will have done its work if it provides, for every sentence s in the language under study, a match­ing sentence (to replace 'p') that, in some way yet to be made clear, 'gives the meaning' of s. One obvious candidate for a matching sentence is just s itself, if the object language is contained in the metalanguage; otherwise a translation of s in the metalanguage. As a final bold step, let us try treating the position occupied by 'p' extensionally: to implement this, sweap away the obscure 'means that', provide the sentence that re­places 'p' with a proper semantical connective, and supply the description that replaces's' with its own pre­dicate. The plausible result is

(1) s is T if and only if p. What we require of a theory of meaning for a language L is that without appeal to any (further) semantical notions it places enough restrictions on the predicate 'is T to entail all sentences got from schema T when 's' is replaced by a structural description of a sentence of Land 'p' by that sentence." (Davidson 1967, 23)

In fact, even most critics of this theory of meaning hold it to be adequate, though in­complete (cf. Strawson 1971, 188; Grice 1989, 231-237). The purpose of this book is not to provide a complete theory of meaning; nevertheless, there are certain questions that need to be answered and that seem to be overlooked by theories like that of David­son. More specifically, I will treat the following three questions:

LEGAL NORMS AND LEGAL SYSTEM 11

(a) If we take convention-T seriously, we must conclude that non-declarative sentences have no meaning. We can add convention-E for imperatives, and other con­ventions for interrogative, performative, and other classes of sentences. But then, the following problem arises: How can we distinguish the mood of one sentence from that of another? We know that the syntactic structure of a sentence is not always a sufficient means. It looks as if the theory of sense, without any further complement, cannot an­swer that question.

(b) Some authors (Strawson 1971, 170-189; Peacocke 1976) have expressed the following concern: Suppose we have a theory of the sense of sentences that can show all truth-consequences of those sentences. There may then still be something that needs to be explained, namely, how those sentences acquire that meaning and, therefore, lead to those truth-consequences. We are tempted to say that this is so because people use them in a certain way in verbal communication. That idea can be refined with the notion of intentional activity. Sentences have the truth-consequences they have because people use them with certain communicative intentions (cf. Platts 1979, 86 f.; Marmor 1992, 13-34).

It seems that this aspect too cannot be accounted for with a theory like that of Davidson.

(c) The third question I wish to refer to is that for a theory of truth like David­son's there can be no propositions without a truth-value. Hugly and Sayward recently presented that argument as follows:

"Let ML be a metalanguage that contains L as a fragment. Then consider s is true in L if and only if s

(s is a designator of the sentence s and L designates L), and the following argument: ASSUMPTION 1. If s lacks a truth-value in L (= neither it nor its negation is true in L), then s also lacks truth-value in ML. ASSUMPTION 2. If s lacks a truth-value in L, then

s is true in L is false in ML. ASSUMPTION 3. A biconditional is not true in ML if one side is false in ML and the other side lacks a truth-value in ML. ASSUMPTION 4. A theory (a deductive closed set of sentences) is true only if each member of the set of its nonlogical sentences is true. From these four assumptions plus Davidson's definition it follows that no true theory of truth for L can be couched in ML if L contains sentences without truth-values ... A stronger conclusion is derivable: no true theory of truth of the sort envisaged by Davidson for a language with truth-value gaps is possible." (Hugly and Sayward 1993,551 f.)

This is a particularly serious conclusion since, as will be seen in the next chapter, my analysis of legal propositions presupposes that there are propositions without a truth­value. However, we can adopt a theory of sense that comes close to that of Davidson, and still account for propositions without a truth-value. As Hugly and Sayward (1993, 558) observe: "For gappy languages it is necessary to distinguish between being false and not being true." That is precisely what will be done in section 3 of the next chapter.

12 CHAPTER I

It will require some restrictions in our definiton of satisfaction and truth which I will not elaborate here,8 but which must be accepted in order to develop the philosophi­cal position about legal propositions presented in the next chapter.

For the time being, nothing more will be said about question (c). Concerning questions (a) and ( b), I trust that a theory of the force of linguistic expressions, to which I will now turn, enables us to fill the gaps the theory of sense a la Davidson has left open (cf., however, Davidson's more recent reflexions in Davidson 1990).

3. The Theory of Force, and Prescriptions

The theory of the force of linguistic expressions allows us to complete our theory of meaning. The conception has been presented by Platts (1979, 67) in the following way:

"We therefore have as our final definition of 'Sentence s in language L means that p' the following: There is a truth-theory 8 for L such that: (a) It is a theorem of 8 that s is true if and only if p; 9 and (b) the deliverances of that theory combine with an acceptable theory of force and with observed linguistic and non-linguistic behaviour to license the ascription of plausible propositional attitudes to speakers of L."

One of the objectives of the theory of force is to enable us to indicate the kind of speech-act performed on some occasion of utterance - assertion, question, command, etc. - and to show us how we can obtain a declarative sentence adequately linked to that speech-act (McDowell 1976,42-66; Platts 1979,58-63).

The other objective I will look at is the analysis of intentions in a theory of meaning or, more precisely, in the theory of the force of linguistic expressions. I will analyze mainly the notion of relevance and its relationships with the theory of meaning, as well as the previously sketched notion of a normative system.

a) The illocutionary force of prescriptions

As is commonly known, in contemporary philosophy, we are indebted to Austin (1962) and Searle (1969) for having insisted on the fact that language not only serves for trans­porting information about the world, and for having emphasized the action-component

8 In order to do this, besides the symbols of our languages and the sequences that satisfy them, we need something like Lewis's (1983, 193-197) notion of indices, i. e., factors on which the extensions of our sym­bols can depend. In that way, we can account for propositions without truth-value, e. g., propositions re­ferring to inexistent entities. As Lewis (1983, 196) observes: "A name may not denote anything at a given possible world. 'Pegasus', for instance, denotes nothing at our world, so its intension may be taken as unde­fined at any index having our world as its world coordinate. A sentence that suffers from failure of presup­position is often thought to lack a truth-value. If we adopt this treatment of presupposition, sentences suscep­tible to lack truth-value should have intensions that are undefined at some indices." Cf. an analogous solu­tion in Hugly/Sayward 1993, 558 f.

9 This formulation must be slightly modified if we want to account for propositions without truth-value. Cf. the previous footnote.

LEGAL NORMS AND LEGAL SYSTEM 13

implied in language. IO In their work, they insist on the difference between the proposi­tional content of an utterance and its illocutionary force. Only an adequate theory of force seems to enable us to distinguish whether on uttering some linguistic expression we are asserting something, commanding something, asking something, etc.

Searle (1979) even listed twelve dimensions on which we can distinguish dif­ferent kinds of illocutionary acts. Here, I am not interested in a complete taxonomy of such acts. Therefore, I will refer only to one of the differences pointed out by Searle, which I think is especially important for distinguishing declarative utterances from pre­scriptive or imperative utterances. I am referring to what for Searle - based on Ans­combe (1957) - are differences in the direction of the match between words and the world:

"Some illocutions have as part of their illocutionary point to get the words (more strictly, their propositional content) to match the world, others to get the world to match the words. Assertions are in the former catego­ry, promises and requests are in the latter ... Suppose a man goes to the supermarket with a shopping list given him by his wife on which are written the words 'beans, butter, bacon, and bread'. Suppose as he goes around with his shopping cart selecting these items, he is followed by a detective who writes down every­thing he takes. As they emerge from the store both shopper and detective will have identical lists. But the function of the two lists will be quite different. In the case of the shopper's list, the purpose of the list is, so to speak, to get the world to match the words; the man is supposed to make his actions fit the list. In the case of the detective, the purpose of the list is to make the words match the world; the man is supposed to make the list fit the actions of the shopper." (Searle 1979,3 f.)

Searle adds that in case of a mistake the detective can correct it by changing his list, whereas a mistake by the shopper cannot be corrected by changing the list: He would have to go back to the supermarket and change the products. Thus, it is the illocutionary force which determines how the propositional content relates to the world.

Declarative utterances have a word-to-world direction of fit, prescriptive or im­perative utterances have a world-to-word direction of fit. Thus, we can say that a pro­position - the meaning of a declarative sentence - is the set of possible worlds which make it true, and truth is a word-to-world relation. In contrast, a norm - the meaning of a prescriptive sentence - is the set of possible worlds that make if effective, and effi­cacy is a world-to-word relation.

Suppose now that, in a way similar to that of Searle, through the illocutionary force of utterances we can distinguish whether we have a declarative or a prescriptive utterance. The theory of force must then still provide us with an element that enables us to relate that sentence p with another sentence p' of the object-language. That is, we need a function f which for every sentence in our language enables us to say thatf(p) = p'. In the case of indicative sentences, the function is simple: it is the identity function f(p) = p. But when p is not a declarative sentence, the function will be more complex. Now, for the case of prescriptions we have the notion of a corresponding d-formula for i-formulae as well as for m-formulae. Using the ideas presented before - and now only a language of propositional logic - we can say that

10 Obviously, this topic also plays a central role in the work of the second Wittgenstein (1953).

14 CHAPTER I

(J)f(OP) = Ap

and

(2)f(Pp) = Vp.

Once we have this function, we can introduce the resulting declarative sentence, p', in the theory of sense and analyze it with the help of convention-To

This approach also makes it possible to ascribe propositional attitudes to the per­son uttering the sentences. The standard ascription is that if someone says that p, then he believes that p; and if someone prescribes that p, then the standard ascription is that he wishes that p (Platts 1979,63). In any case, propositional attitudes concern the place of intention in a theory of meaning. That is what I will turn to now.

b) Intentions and meanings

Some 25 years ago, Strawson (1971, 170-189) already examined two contending ap­proaches in philosophical theories of meaning: theories of communicative intention, on the one hand, and theories of formal semantics, on the other. He presented that conflict in the following terms:

"A struggle on what seems to be such a central issue in philosophy should have something of a Homeric quality; and a Homeric struggle calls for gods and heroes. I can at least, though tentatively, name some living captains and benevolent shades: on the one side, say, Grice, Austin, and the later Wittgenstein; on the other, Chomsky, Frege, and the earlier Wittgenstein." (Strawson 1971,172)

Strawson himself advocates an approach in the line of the theorists of communicative action; the main target of his criticism are theorists of formal semantics a la Davidson.

Theorists of communicative action usually present their approach in the follow­ing way: They form a primitive concept of communication (or communicative inten­tion) such that it does not presuppose the concept of linguistic meaning, and they then show that the latter can and must be explicated in terms of the former.

The most elaborate conception of this kind is that of Grice (1989). Grice's cru­cial distinction is the one between the notions of sentence-meaning and utterer's mean­ing, i. e., between 'sentence s means that p' and 'by uttering sentence s, x means that p'. The theory then defines utterer's meaning in terms of the notion of the utterer's inten­tions, and finally attempts to define sentence-meaning in terms of the previously de­fined utterer's meaning.

Here, I do not wish to analyze Grice's several attempts to refine his definition of utterer's meaning,ll as a reaction to the criticisms it provoked (cf. Grice 1989,93-104).

11 Grice's standard definition of 'utterer's meaning' is as follows: " 'U meant something by uttering x' is true iff, for some audience A, U uttered x intending: (1) A to produce a particular response r (2) A to think (recognize) that U intends (1) (3) A to fulfill (1) on the basis of his fulfilment of (2)." (Grice 1989, 92)

LEGAL NORMS AND LEGAL SYSTEM 15

I will only try to show that none of the approaches to meaning by theorists of communi­cative action can substitute the theory of sense in terms of truth-conditions.

According to Grice, sentence-meaning is defined in terms of the intentions with which a sentence is uttered, perhaps together with the receivers' response habitually guaranteed by that utterance. Now, Platts (1979, 89 f.) argues, that explication fails as an explication of the meaning of sentences in natural languages. Most of those senten­ces (of which there is an infinite number in natural languages) will never be uttered. Therefore, neither will they be uttered with any intention, nor will their being uttered in­duce any response in some audience. Thus, Platts continues,

"The obvious move, perhaps the only possible one, is to hold the meanings of such sentences to be definable in terms of the intentions with which they would be uttered were they to be uttered and the responses they would then induce".

Now, Grice confronts a dilemma: Either those intentions and responses are somehow restricted, or they aren't, In the latter case, the meanings of those unuttered sentences will be completely undetermined; in fact, they will not mean anything at all. And in the former case, that restriction can only come from the meaning of the sentence itself. Therefore, the attempt to define the meanings of unuttered sentences through hypothe­tical intentions and responses fails; a notion of sentence-meaning must be presupposed. And such a notion we obtain from the theory of sense in terms of truth-conditions.

But then, what is it that makes the approach of the theorists of communicative action so attractive? Perhaps, Platts (1979, 92 f.) suggests, that it does not distinguish between the following theses:

,,(T.1) A system of utterances would not be a language unless intentions were ascribed to the utterers; more specifically, an utterance is a piece of linguistic behaviour only if it is intentional. (T.2) The notion of sentence-meaning can be defined in terms of the notion of utterers' intentions. (T.3) The meaning of any particular sentence in a language can be determined by reference to the intentions with which it is uttered."

While we can assume the truth of T.l, it in no way implies T.2 or T.3. But the truth of T.2 and T.3 is precisely what an approach to meaning like that of communicative action requires.

Acceptance of T.l does not refute the theory of sense I have presented. Inten­tions must even be given a central place in our theory of linguistic behaviour. That place, however, is not in the theory of sense, but in the theory of force; more precisely, in the component of the theory of force by which we identify the mood of the utterance of a sentence. As we have seen, it is the theory of force that shows us the propositional attitude (the direction of fit) with which a sentence is uttered.

That means that the 'Homeric struggle' between theorists of communicative ac­tion and theorists of formal semantics must be concluded by a hybrid theory of linguis­tic behaviour which gives room to intentions (in the theory of force), but which also preserves an approach in terms of truth-conditions (for the theory of sense) (cf. Rumfitt 1995 on the need for such a hybrid theory).

16 CHAPTER I

There is, however, another idea of Grice's (1989, 22-40) which, I think, deserves closer attention. It is the idea that in any declarative utterance we can distinguish that which is asserted from that which is implied by what is asserted. Something similar can be said of imperative utterances, where we can distinguish that which is explicitly pre­scribed from that which is implied by what is prescribed (it is precisely this distinction which enables us to speak of a normative system as a deductive system). Now, one could think that the classical notion of logical consequence is all that is needed here. If someone says 'There was a table and a chair in the room', he has also (implicitly, as a consequence of his explicit assertion) said 'There was a chair in the room'. Now, since from that last expression we logically derive 'There was a chair or a dog in the room', he would also implicitly have asserted the proposition expressed by that last sentence. Part of Grice's work on so-called "conversational implicatures" attempts to restrict the possibility of obtaining such counterintuitive conclusions.

In the next section, I will propose a notion of relevant conclusion that can miti­gate those counterintuitive conclusions. In order to do that, I will take as a paradigm a certain conception of the justificaton of judicial decisions.

c) Relevance and implicit meaning

In the recent literature on the justification of judicial decisions, we usually find a dis­tinction being made between external and internal justifications (Wroblewski 1971, 411 f.; Alexy 1978, 119 f.; Aarnio 1987, 119 f.). Internal justification concerns the validity of the inference from some premises to a judicial decision, taken as their conclusion. The function of external justification, in contrast, is centered on the control of the pre­mises.

A judicial decision is internally justified if and only if it follows from the norms applicable to the case, and from the sentence(s) describing the facts of the case. On that conception, ajudicial argument (JA) can take the following form:

( 1) If x does A, x must be punished with sanction S. (2) x does A.

Therefore, (3) x must be punished with sanction S.

(3) is justified since it is a deductive consequence of (1) and (2). Despite the numerous criticisms it has evoked, it seems to be widely agreed that deductive arguments play a central role in internal justification (e. g., MacCormick 1994, ix). Thus, internal justifi­cation can be reconstructed as a logical inference.

Most criticisms of deductive logic, however, relate to the insufficiency of the in­struments of formal logic for analyzing the question of the adequacy of the premises (i. e., external justification). And the main problems of justification in the law are often said to concern external justification (Aarnio 1987, 119 f.)

But I will not go into these problems here. Therefore, in what follows, 'justified legal decision' must simply be taken as a synonym for 'internally justified legal deci-

LEGAL NORMS AND LEGAL SYSTEM 17

sion' .12 I will also not treat the problem of whether the insufficiency of deductive logic may be due to the complexity of some particular external justification.13 Instead, I will be concerned with the question of whether deductive logic (and internal justification) does not actually prove too much. It should not be forgotten that any set of premises logically entails an infinite number of consequences and, therefore, an infinite number of sentences would be justified with any judicial argument. In lA, from (1) and (2) we cannot only deduce (3), but also

(3') x must be punished with sanction S or given award A. (3") If it rains tomorrow, x must be punished with sanction S.

and (infinitely) many more. 14

(3') and (3") are logical consequences of (1) and (2), but intuitively they seem to be paradoxical consequences. No jurist would regard (3') or (3' ') as adequate ('justi­fied') sentences grounded in (1) and (2). But if the only criterion of justification for ju­dicial decisions is that of logical deduction, then we must try to find out why these con­clusions seem paradoxical, and to find possible solutions. The paradox perhaps lies in the lack of a connection of (3') and (3") with the premises of the argument, in - as we may say - the irrelevance of those conclusions.

The notion of relevance, however, eludes precise conceptual reconstruction and, what's even more important, an adequate logical treatment. 15

12 Lyons (1993,119-140) prefers to reserve the expression 'justified judicial decision' for morally adequate judicial decisions. Such decisions are not necessarily the same as the decisions that derive from legal norms. But here I will not go into this topic either.

13 In MoresolNavarro/Redondo 1992 we have treated that question under the name of the 'thesis of insuffi­ciency' .

14 fA could be represented as follows:

l)p~Oq

2)p

~3) Oq

I will assume that expressions like 'p ~ Oq' are adequate representations of conditional norms. As is well­known, there is no agreement on this in the literature on deontic logic. Von Wright, to mention an undispu­ted authority, has taken a vacillating stand on this central point of the logical representation of norms. Thus, in 1980 he wrote: "It should be noted that this standard system differs in two important respects from the system proposed in my (1951). First, it admits 'mixed' formulae formed of propositional variables, sentential connectives, and deontic expressions. For example: p ~ Oq, is a well-formed formula of the standard sys­tem ... " (von Wright 1980, 403). But only a few years later, he observed: "On the interpretation of Oq as a prescription (norm), the expression p ~ Oq would consist of a descriptive and a prescriptive ingredient joined by a sentential connective. Does such an expression 'make sense'? The question is not really easy to answer. It seems to me clear, however, that if the standard connective in question is a truth-connective, then this 'linguistic hybrid' is a monster with no place in meaningful discourse." (von Wright 1983a, 151)

15 In Appendix B, I will discuss some of the notions of relevance to be found in the philosophical literature, one of which (the one that seems to me the most adequate) will be applied to the question of the justification

18 CHAPTER I

To reason is to show that from a given set of premises (P), some conclusion C can be deduced. In classical logic, that C can be deduced from (P) means (in terms of classical semantics) that if the premises (P) are true, then C is necessarily true, i. e., that (P) entails C. The notion of entailment is regarded to be the converse of the notion of 'following from' or 'being deduced from'. Apparently, G. E. Moore (1922, 291; cf. also von Wright 1957, 166) was the first one to introduce this notion:

"We require, first of all, some term to express the converse of the relation which we assert to hold between a particular proposition p and a particular proposition q, when we assert that q follows from or is deducible from p. Let us use the term 'entails' to express the converse of this relation." (G. E. Moore 1922, 291)

Thus, if we have two premises

we can deduce

~q,

but we can also deduce

~q v r, ~r~q, ~-,q ~ r,

(2) P

and many other conclusions. For instance, from the premises 'If you come with me, I go to the movie' and 'You come with me', we can deduce 'I go to the movie', but also 'I go to the movie or I stay at home', 'If you insult me, I go to the movie' or 'If I don't go to the movie, you will insult me' .16 The paradoxical note of these conclusions (which are necessarily true if the premises are true) is due to the fact that there is no connection between them and the starting point. Apparently, an argument is irrelevant if there is no connection between the premises and the conclusion; and the connection of logical entailment is not enough to produce relevant arguments.

If to this we add the fact that many logicians mistrust a formal treatment of the notion of relevanceP it seems that a theory of reasoning based on the formal-logical notion of deduction is highly deficient.

of judicial decisions, followed by some reflections, related to that notion of relevance, on the concept of a legal system as a normative system.

16 The paradoxical overtone may become manifest in the following way: I say to a friend 'If you come with me, I'll go to the movie'; he answers 'I'll go with you'; and I say 'If you insult me, I'll go to the movie'. The contexts of human communication in which the use of practical reasoning takes place show that this kind of reasoning is inadequate (for common purposes).

17 Thus, Suppes (1957, 8) writes that ,,00' the notion of connection or dependence being appealed to here is too vague to be a formal concept of logic"; and Susan Haack (1978, 16 f.) holds that "Considerations of rele-

LEGAL NORMS AND LEGAL SYSTEM 19

Among the philosophers interested in reasoning, this leads to three possible atti-tudes:

I) A (possibly partial) rejection of formal logic as a criterion for controlling our arguments. This thesis has two corollaries:

l.a) There are valid arguments that do not follow the patterns of formal logic. l. b) There are invalid arguments that do follow the patterns of formal logic. II) A construction of a new, deviant, non-classical logic that does not produce

the paradoxes of irrelevance. In other words, the elaboration of a (deviant) notion of logical consequence which incorporates the notion of relevance.

III) A restriction of classical logic with additional criteria of relevance. For this conception, there can be no valid arguments that do not follow the patterns of classical logic; but there are formally valid arguments that are irrelevant. For this conception, reasoning consists in logic plus relevance.

In Appendix B, the reader will find a detailed analysis of these three approaches. Suffice it here to point out that, in my view, in order to show that a set of normative and factual premises justifies a conclusion, the conclusion must not only derive logically from the premises; it must also be shown that it is a relevant consequence of that set of premises. For this reason (among others), in a normative system it is important to dis­tinguish the derived norms which are relevant consequences of that system from those which are irrelevant consequences.

4. Conclusions

( 1) The law can be seen as a set of prescriptive and conceptual rules containing all their logical consequences, i. e., as a normative system.

(2) In such a set, it is important to distinguish the consequences that are relevant from those that are irrelevant.

(3) In order to adopt this approach, we need to attribute meaning to norm-formu­lations, i. e., to interpret those formulations, which are the result of certain speech-acts by certain norm-authorities.

(4) Such an attribution of meaning presupposes the attribution of propositional attitudes to the authorities. For this, we need a theory of illocutionary force that can ac­count for the intentions of such norm-authorities.

(5) In order to identify the normative content prescribed by authorities, we need a theory of sense. Such a theory is provided by convention-E, which is parasitic on con­vention-T and, therefore, uses an approach in terms of truth-conditions.

vance are apt to be relegated to the rhetorical rather than the logical dimension of assessment of arguments". This attitude is neatly summarized - if only to submit it to the most important critical revision in logico­philosophical literature so far - by Anderson and Belnap (1975. XXI): .. The difficulty of treating relevance with the same degree of mathematical sophistication and exactness characteristic of treatments of extension­al logic led many influential philosopher-logicians to believe that it was impossible to find a satisfactory treatment of the topic."

LEGAL NORMS AND LEGAL SYSTEM 21

ApPENDIX A: A FORMAL THEORY OF SENSE FOR PRESCRIPTIONS

a) Language LN

In philosophy of language, the question what entities are to be considered to be true or false (propositions, type-sentences, inscription-sentences, etc.) has been much discus­sed. In the same way, one can discuss what entities can be said to have the property of being 'effective'. In this book, I will assume that norms are the vehicles of efficacy. I will also assume that norms can be represented in an artificial language. Therefore, I will hold that the vehicles of efficacy are a subset of the well-formed expressions of an artificial language LN. LN is a language of first-order predicate deontic logic.

Primitive symbols:

a) Constants: a, b, c, a', b', c', a", ... b) Predicates: F, G, H, F', G', H', F" ... c) Connectives: -', /\, V, ~

d) Parenthesis: (, ) e) Variables: x, y, z, x', y', z', x" ... f) Quantifiers: 'if, ::3 g) Deontic operators: 0, Ph, pI8

The well-formed expressions of LN can be characterized recursively as well-formed ex­pressions of a declarative (d-formulae), imperative (i-formulae), or mixed (m-formulae) nature (where the latter are actually of an imperative nature too).

Definition of ad-formula:

1) An n-adic predicate letter, followed by n constants is an atomic d-formula. 2) If A is a d-formula, then -,(A) is ad-formula. 3) If A and B are d-formulae, (A /\ B) is ad-formula. 4) If A and B are d-formulae, (A v B) is ad-formula. 5) If A and B are d-formulae, (A ~ B) is ad-formula. 6) If A is a d-formula, and a is the result of the substitution in A of a constant by a variable, then the expression formed by the universal quantifier (''if') followed by the variable, followed in turn by a also is ad-formula.

18 As we know, the operators (to be read as 'obligatory', 'prohibited', and 'permitted'), are interdefinable: P = -.0-.. P = -.Ph.

There is also the operator 'facultative' (F) which can be defined with the help of 'permitted' and therefore is not needed here:

F = P 1\ P-..

22 CHAPTER I

7) If A is a d-formula, and a. is the result of the substitution in A of a constant by a variable, then the expression formed by the existential quantifier ('3') followed by the variable, followed in tum by a. also is ad-formula.

Definition of an i-formula:

1') If A is an atomic d-formula, then a deontic operator followed by (A) is an atomic i-formula. 2 ') If A is a i-formula, then -,(A) is an i-formula. 3') If A and B are i-formulae, (A /\ B) is an i-formula. 4') If A and B are i-formulae, (A v B) is an i-formula. 5') If A is an i-formula, and a. is the result of the substitution in A of a constant by a variable, then the expression formed by the universal quantifier (''<t') fol­lowed by the variable, followed in tum by a. also is an i-formula. 6') If A is an i-formula, and a. is the result of the substitution in A of a constant by a variable, then the expression formed by the existential quantifier ('3') fol­lowed by the variable, followed in tum by a. also is an i-formula.

Definition of an m-formula:

1 ") If A is a d-formula and B is an i-formula, then (A ~ B) is an m-formula. 2") If A is a d-formula and B is an m-formula, then (A ~ B) is an m-formula. 3") If A is an m-formula, and ex is the result of the substitution in A of a constant by a variable, then the expression formed by the universal quantifier (''<t') fol­lowed by the variable, followed in tum by a. also is an m-formula. 4") If A is an m-formula, and ex is the result of the substitution in A of a constant by a variable, then the expression formed by the existential quantifier ('3') fol­lowed by the variable, followed in tum by ex also is an m-formula.

Comments:

The main objective of language LN is to provide an adequate base for the analysis of the sense of imperatives. In contrast to other artificial languages, the construction of LN is not aimed at the solution of problems of deontic logic; instead, it is intended to facilitate the investigation of the semantics of directive discourse.

Therefore, it may be helpful to clarify some of the restrictions implied by lan­guageLN:

(a) Language LN allows the use of all the logical rules of a natural deductive cal­culus for first-order predicate logic, provided the expression obtained is a well-formed expression (a formula) of LN.

(b) A conditional with a deontic operator in the antecedent, as in 'O(Fa) ~ O(Ga)" is not a well-formed expression of LN. The only conditionals admitted are mixed expressions the antecedent of which is a d-formula and the consequent of which

LEGAL NORMS AND LEGAL SYSTEM 23

is an i-formula or an m-formula. Although syntactically such expressions are mixed, semantically they are of an imperative nature. The restriction arises from pragmatic considerations. In imperative language, it does not seem possible to use conditionals with an imperative antecedent. For instance, the imperative sentence 'If you go to the market, buy oranges!' is a well-formed expression, whereas the sentence 'If, go to the market!, buy oranges!' makes no sense.19 A normative proposition, e. g., a sentence de­scribing the existence of a norm, however, can be the antecedent of a conditional. For instance, 'If you must go to the market, then buy oranges!' has meaning; but the antece­dent must be represented as a d-formula instead of an i-formula.

For the same reason, some transformations authorized by the rules of logic are invalid in LN. Thus, in LN it is impossible to go from '(Fa) ~ O(Ga)' to '-.o(Ga) ~ -(Fa)', since the second expression is not a formula of LN.

A certain inconvenience of expressions like 'O«Fa) ~ (Ga))', which is either equivalent to '(Fa) ~ O(Ga)' or incomprehensible, is the reason why LN authorizes deontic operators only in front of atomic d-formulae.

(c) In LN, mixed expressions of a conjunctive or disjunctive nature are not per­mitted, for similar reasons as those pointed out by Smart (where 'Ap' is an assertive or declarative sentence, and 'lp' an imperative sentence):

"I am inclined to think that mixtures of imperatives occur in colloquial language only in the form Ap -7 1q (and perhaps in a rather strained way in the form Apvq which comes to the same as A(op) -7 Iq. We do not get ApAq though we do get Ap and 1q asserted as separate premisses. I do not think we ever get Ip -7 Aq, even though this would be a way of saying A(-,q) -7 I(op)." (Smart 1984, 16)

(d) General norms, e. g., 'Persons of age must vote' are expressed in LN through m-formulae and have the following canonical formulation:

''ilx (Fx ~ O(Gx))'.

(e) Note that the predicates of LN, when modalized by a deontic operator, must be interpreted as properties attributed to human beings, concerning the performance of certain actions. LN is a language that permits quantification only over actors, not over actions (Ziemba 1976, 383 ff.; Makinson 1981, 87-91; Hernandez Marin 1984, 113 ff.). For instance, 'The chairs must recite a Shakespeare sonett' is an expression that makes no sense. This is analogous to the situation generated by expressions such as 'Prime numbers are blue' (cf. Grant 1968, 189 f.).

b) Open Sentences

The expressions which in the definitions above were designed by a., i. e., expressions with free variables, we will call open d-sentences, i-sentences, or m-sentences, respec-

19 Not all scholars of deontic logic share this view. Weinberger (1984,1991), e. g., has insisted that expres­sions of the form 'p -7 Oq' are unsatisfactory reconstructions of normative conditionals.

24 CHAPTER I

tively. Examples of such open sentences are: '(Fx ~ Gx)', '(O(Gx) 1\ P(Hx)) , , or '(Fy

~ Ph(Hy))" etc. Open d-sentences are neither true nor false; rather, they are satisfied or not satis­

fied by certain objects (or pairs of objects, or triplets of objects, etc.). Thus, if 'F' is the predicate 'being a philosopher', the open d-sentence 'Fx' is satisfied by Aristotle, but (perhaps) not by Alexander the Great. If 'G' is 'being the lover of', the open d-sentence 'Gxy' is satisfied by the pair <Marcus Antonius, Cleopatra>, but (perhaps) not by the pair <Caesar, Brutus>.

Open sentences are sentential functions (it is easily seen that the so-called open d-sentences are not d-formulae). The notion of satisfaction can be explicated informally in the following way. An open d-sentence is satisfied by certain objects if the expres­sion becomes a true d-formula on substituting the free variables by the names of those objects. Thus, we can say that the pair <Marcus Antonius, Cleopatra> satisfies the open sentence 'x is the lover of y', since 'Marcus Antonius is the lover of Cleopatra' is a true sentence. But, since - according to Tarski - the definition of truth requires the notion of satisfaction, we cannot (without circularity) use the predicate 'true' for the definition of 'satisfaction' .20

In any case, this seems an adequate way of informally introducing the notion of compliance for open i-sentences and open m-sentences. An expression like 'x should be a movie actress' is complied with by Julia Roberts, but not by Margaret Thatcher, and 'x should kill y' is complied with by the pair <Brutus, Caesar>, but not by many other pairs. Therefore, we can sustain that an open i-sentence (or m-sentence) is complied with by some individuals if on substituting the names of those individuals for the free variables that expression becomes an effective i-formula (or m-formula). Thus, the pair <Kennedy, Johnson> complies with the expression 'If x dies, y should succeed him', since 'If Kennedy dies, Johnson should succeed him' is an effective sentence.

In view of this presentation, it should be clear that efficacy is parasitic on truth: An i-formula or m-formula is effective if and only if its corresponding d-formula is true.21 Let's take the simplest case: that of atomic i-formulae, with expressions such as 'O(Ga)'. The corresponding d-formula is 'Ga'.

In order not to complicate things unnecessarily, I will choose a subset of the i­formulae and m-formulae of LN. Based on them, I will show what their corresponding

20 In the words of Grayling (1982, 160): .. Thus, for example, snow (not the name 'snow' but the actual stuff) satisfies 'x is white' because the sentence 'snow is white' is true. However this is a merely heuristic way of explaining satisfaction, for 'true' is being used here; and because we wish to define 'true', we must seek for an account not involving 'true'."

21 Thus, Ross (1941, 60) writes: ..... an imperative I is said to be satisfied, when the corresponding indica­tive sentence S describing the theme of demand, is true, and non-satisfied, when that sentence is false". Wil­liams (1973, 187 f.; cf. also Hart 1983, 325) formulated a similar suggestion: ..... that corresponding to any imperative 'do x', there is an indicative statement, 'x is done', which might be called its 'obedience state­ment'; and we may say that two imperatives are inconsistent if their obedience statements are inconsistent." A critique of this way of analysis can be found in Kelsen (1979, 163 f.; 173-176).

LEGAL NORMS AND LEGAL SYSTEM 25

d-formulae are. For that subset of formulae, the notion of compliance will be defined recursively, and we will thus arrive at a definition of efficacy.

c) Satisfaction and Compliance

i) A subset ofLN

Let us take a subset of LN in which all predicates of the language are monadic, and let us restrict the connectives to negation and conditional. The negation only precedes atomic formulae, and the conditional links only atomic formulae, or formulae preceded by the negation. Finally, in i-formulae and m-formulae, we will use only the universal quantifier. In this subset, all expressions are of the following form:

d-formulae:

Fa -(Fa) (Fa ~ -Ga) \/x(Fx~-Gx)

3x (Fx /\ Gx)

(Fa ~ O(Ga)) (Fa ~P(Ga))

m-formulae:

ii) Corresponding d-formulae

i-formulae:

O(Fa), Ph (Fa), P(Fa) -(O(Fa)), ... \/x (O(Fx)) \/x(P(Fx))

\/x (Fx ~ O(Gx)) \/x (Fx ~ P(Gx))

I will now show what the d-formulae corresponding to that subset of i-formulae and m­formulae are. The basic intuition is the following: The d-formulae corresponding to i­formulae and m-formulae are true descriptions with respect to some possible worlds. These are deontically perfect worlds with respect to those i-formulae and m-formulae. In those worlds, norms are always effective.

The d-formula corresponding to an atomic i-formula of obligation or permission Xi is obtained by eliminating the deontic operator from Xi. The d-formula corresponding to an atomic i-formula of prohibition Xi is obtained by eliminating the deontic operator and putting the sign of negation in front of Xi.

Let us now look at i-formulae with quantifiers. An i-formula like '\/x (O(Fx))' is effective if it is complied with by all its addressees. Therefore, we can say that \/x (Fx) is its corresponding d-formula. One can argue, however, that this is asking too much, and that the compliance of a norm N by some majoritarian percentage of its addressees is sufficient for the efficacy of N. I will comment on that question later.

Now, when can we say that an i-formula like '\/x (P(Fx))' is effective? Follow­ing von Wright (1983a, 139), I will use the following convention: A permissive norm is complied with if and only if it is used by its addressees on at least one occasion; in other

26 CHAPTER I

words, if the state of affairs permitted by the norm is produced at least once. Von Wright asserts that it is rather reasonable to expect that the states of affairs permitted by an authority will obtain on some occasion, adding:

"I shall say that a permissive norm is satisfiable if, and only if, it is possible that the permitted state of affairs obtains at some time in the history of the norm. And it is satisfied if, and only if, at some time in its history that which it permits actually is also the case." (von Wright 1983a, 139)

A permission given with a certain condition of application, e. g., 'Smoking is permitted during dinner', is complied with (and therefore effective) if and only if some addressee of that permission smokes during dinner.22 If one accepts that convention, then the i­formula ''r/x (P(Fx)), is effective if and only if there is some x that is F. Therefore, we will say that the d-formula corresponding to that i-formula is '3x (Fx)' .23

Let us now analyze the negations of atomic i-formulae. According to an old, but controversial intuition of von Wright (1963a, 138), the negation of a norm is a norm just as the negation of a proposition is a proposition. In view of the interdefinability of the operators, the negation of an i-formula 0- or Ph- is an i-formula P-, and the nega­tion of an i-formula P- is an i-formula 0- or Ph- Thus, the i-formula '-{O(Fa))' is ef­fective if and only if '-{Fa)' is true, and this then is the d-formula corresponding to the above i-formula.

An m-formula like '(Fa ~ O(Ga))' has as its corresponding d-formula '(Fa ~ (Ga)', and the d-formula corresponding to ''r/x (Fx ~ O(Gx))' is ''r/x (Fx ~ (Gx),.24

22 This reconstruction has a counterintuitive consequence. If norms making an action optional (facultative) are defined as the conjunction of the permissions to do and to omit that action, then in this language making an act-individual facultative at a given time would always be ineffective (no-one can do and omit an action at the same time). Of course, this was not the only way of approaching the question of permissions. An alter­native way was to assume that, since permissions in principle cannot be complied with, they do not really express prescriptions in the same way as obligations and prohibitions and that they are perhaps expressions of acts of rejecting (previous) prohibitions or acts of rejecting them in advance. This is a way explored by Alchourr6n and Bulygin (1981). It would lead to an elimination of our P-expressions from language LN.

23 Incidentally, this intuition corresponds to the analogy some deontic logicians (cf. Kalinowski 1975,40 f.) establish between the square of opposition representing the relations of Aristotelian predicate logic (where universal statements imply their corresponding individual statements through a relation of subaltemation) and the square representing the relations between deontic expressions (where expressions of obligation entail permissive expressions). On the origin of that analogy in the work of Bentham, cf. Moreso 1992, 148 f.

24 This implies, of course, that a norm like 'All those with an income of more than 10 million dollars a year will pay an income tax of 50%' is effective if no-one has an income of more than 10 million dollars a year. This may seem paradoxical if efficacy is linked to the motivation of behaviour: in that case, such a norm ap­parently does not motivate any behaviour. But that paradoxical note is similar to that produced by condi­tional statements with a false antecedent, e. g., 'All centaurs are metaphysicists'. Such statements are a con­venient way of increasing our set of truths without increasing our knowledge about the world. However, as has been noted (Dummett 1978, 9 f.), when the truth of the antecedent depends on an action that is in the power of the addressee of the imperative, then we cannot say that this kind of conditional imperatives does not motivate behaviour. Suppose a mother orders her son 'If you go out, wear your coat'. We cannot say that if the son does not go out it is as if the imperative had not been issued, since the reason why he does not go out could be, e. g., that he cannot find his coat.

LEGAL NORMS AND LEGAL SYSTEM 27

The i-formula '(Fa ~ P(Ga))' has as its corresponding d-formula '(Fa A Ga)', while 'Vx (Fx ~ P(Gx)), has the corresponding d-formula '3x (Fx A Gx)'. Note that there is a certain asymmetry between expressions of obligation (or prohibition) and permissive expressions. Conditional formulae of obligation are effective even when the condition of application is absent, whereas conditional formulae of permission are not effective in that case. The analogy between universal sentences vs. singular sentences and expres­sions of obligation vs. expressions of permission is thus maintained.

In summary, we can now produce a list of some correspondences:

i-formulae

O(Fa) Ph (Fa) P(Fa) --{O(Fa))

--{Ph(Fa))

--{P(Fa))

Vx (O(Fx))

Vx (Ph(Fx))

Vx (P(Fx))

Vx --{O(Fx))

Vx --{Ph(Fx))

Vx --{P(Fx))

m-formulae

(Fa ~ O(Ga))

(Fa ~ P(Ga))

Vx (Fx ~ O(Gx))

Vx (Fx ~ P(Gx))

iii) Satisfaction and the truth of d-formulae

corresponding d-formulae

Fa -(Fa) Fa -(Fa)

Fa

-(Fa)

Vx(Fx)

Vx --{Fx)

3x (Fx)

3x --{Fx)

3x (Fx)

Vx --{Fx)

corresponding d-formulae

(Fa ~ (Ga))

(Fa A Ga)

Vx (Fx ~ (Gx)) 3x(Fx A Gxj25

Now, let us recall that satisfaction is a relation between open sentences and ordered n­tuples of objects. According to Tarski, open sentences, like

F(x, x', x' " ... , xn) are satisfied by infinite sequences of objects like

<0, 0', 0", ... , On+l, ... >.

25 I omit the combination with negations since, because of the equivalences between deontic operators, it follows from the application of the mechanism given for i-formulae.

28 CHAPTER I

Since such sentences are satisfied by the first n objects of the sequence, the rest of it can be ignored. Thus, we can say that the negation of an open d-sentence A is satisfied by all sequences that do not satisfy A. Open d-sentences of the form A ~ B are satisfied ei­ther if no sequence satisfies A, or by those sequences that satisfy B.

For any d-formula Xd, all members of any sequence are irrelevant for knowing whether the sequence satisfies Xd• Thus, a true d-formula in LN is satisfied by all se­quences, and a false d-formula is not satisfied by any sequence. For example, '3x (Fx), (where F is 'being a philosopher') is satisfied by any sequence, e. g. the sequence <Oc­tavio, Augustus, ... >, since the open d-formula resulting from the elimination of the quantifier is 'Fx', and that sentence is satisfied by some sequences, like <Aristotle, ... >.

Now we can offer a recursive or inductive definition of satisfaction for our sub­set of LN.26 I will call 'var (i)' the i-th variable of the vocabulary of LN, and Xi the i-th object of a sequence X. If we assume that 'A' and 'B' are monadic predicates of LN, the definition is as follows:

(1) For any i and for any X: X satisfies 'A' followed by var (i) iff Ax,. (2) For any sequence X and for any d-formula A: X satisfies the negation of A iff X does not satisfy A. (3) For any sequence X, for any d-formula A and for any d-formula B, X satisfies 'A ~ B' iff X does not satisfy A or X satisfies B. (4) For any X, for any A, and for any i: X satisfies the universal quantification of A with respect to var (i) iff A is satisfied by any sequence X' such that Xl = X/, for allj withj;>!o i.

With Quine, we can add:

"Taken altogether, the inductive definition tells us what it is for a sentence to satisfy a sentence of the object language. Incidentally it affords a definition also of truth, since, as lately noted, this just means being satis­fied by all sequences." (Quine 1970, 42)

iv) Compliance and the efficacy of d-formulae and i-formulae

The inductive definition of satisfaction, together with the definition of a d-formula cor­responding to an i-formula and an m-formula, allows us to define the notion of compli­ance as a relation between i-formulae or m-formulae and sequences of objects:

For any i-formula A and for any sequence X, X complies with A iff the d-formula B corresponding to A is satisfied by X. For any m-formula A and for any sequence X, X complies with A iff the d-formula B corresponding to A is satisfied by X.

With this, we can now also define efficacy:

An i-formula is effective iff it is complied with by all sequences or, what amounts to the same, iff its corres­ponding d-formula is true.

26 Since the above presentation follows Quine (1970), it should be added that proper names (the constants of LN) are eliminated in the following way: 'Fa' is replaced by the expression '3x (a = x A Fx)" where 'a =' is a predicate. Cf. Quine (1970, 25 0.

LEGAL NORMS AND LEGAL SYSTEM 29

An m-fonnula is effective iff it is complied with by all sequences or, what amounts to the same, iff its corres­ponding d-formula is true.

Since these definitions of compliance and efficacy differ from the usual ones, the following points should be noted:

(a) Norms are not complied with by their addressees, but by ordered sequences of individuals. The addressees of norms are elements of those sequences.

(b) Compliance with d-formulae or with i-formulae depends on the sequences taken into consideration. Those sequences depend on some ascription of the elements of our language to a certain structure of the world in a given domain or universe (an inter­pretation).27 Some sequences obviously satisfy certain i-formulae or d-formulae under one interpretation, but not under another.

( c) The idea underlying this formal definition can be better understood through the following examples:

The norm expressed by 'All persons of age should vote' can be symbolized in LN by the m-formula

(a) Vx (Fx ~ O(Gx)).

The corresponding d-formula is:

(~) Vx (Fx ~ Gx)

(a) is complied with by all the sequences satisfying (~). A world in which no-one is of age, or a world in which all persons of age do vote is a world in which (~) is true and, therefore, (a) is effective.

Similarly, the norm expressed by 'All persons of age may vote' can be symbol­ized in LN by the m-formula

(a') Vx (Fx ~ P(Gx)).

The corresponding d-formula is:

(W) 3x (Fx 1\ Gx)

(a') is effective if and only if (W) is true. The permissive norm expressed in (a') is ef­fective only in the case that some of its addresses make use of the permission to vote. If

27 For an alternative presentation of the notion of satisfaction using the notions of structure (as a set of ob­jects referred to by the universal quantifier - a universe - and the denotation of the predicates of the lan­guage) and interpretation (a function correlating language with structure in such a way that one can say that if an interpretation I satisfies some set of fonnulae rCl, then I is a model of rCl), cf. Enderton (1972, 79-86) and Garrido (1981, 220-230).

30 CHAPTER I

no-one votes, then the permission is ineffective. That is the intuitive reason why (a)

does not imply (a'), since in standard logic (P) does not imply (W)· (d) I will now analyze the possibility that the formulae corresponding to i-for­

mulae and m-formulae of obligation are not d-formulae preceded by the universal quan­tifier. It will then be possible to account for the intuition that in order to predicate the efficacy of a norm we do not require that all its addressees comply with it all the time. I will present two alternative strategies for this.

On the one hand, one can use a numerically defined quantifier in the correspon­ding d-formulae (Quine 1972, 311-316). The idea is as follows: Suppose a norm N is effective if it is complied with by at least 75% of its addressees, and suppose also that N has ten addressees. The i-formula '(a) Vx (O(Fx))' is effective if and only if '3'8X (Fx),

is true. This last expression would then be the d-formula corresponding to the previous i-formula.

In this way, d-formulae corresponding to universal i-formulae or m-formulae would be existential. And, what's more, all corresponding d-formulae in LN would be existential.

Now, as Quine has pointed out, this strategy fails when the number is indefinite, i. e., when the addressees of a norm are indefinite, and that obviously is what happens with many kinds of norms.

On the other hand, we can try to use so-called non-standard quantifiers (cf. Platts 1979, 100-106). Platts introduces the possibility of working with at least two quantifiers between 'all' and 'some', namely, 'most' and 'many'. This means that besides senten­ces like 'All philosophers are boring' and 'Some philosophers are boring', we could have 'Most philosophers are boring' and 'Many philosophers are boring'. I will present Platts's ideas about the 'most' quantifier - which I will call the 'majoritarian quanti­fier' ('M') - with the underlying intuition that the declarative statement corresponding to 'All citizens should vote' could be 'Most citizens vote'.

This idea will enable us to introduce a new clause in the definition of satisfac-tion:

(6) For any X. for any A and for any i: X satisfies the majoritarian quantification of A with respect to var (i) iff A is satisfied by most sequences X' such that Xj = X/. for all j with j ;c i.

Now, we need a canonical formulation for the kind of statements with the majoritarian quantifier. In principle, we have two options. We can either make it similar to a state­ment with the existential quantifier, or make it similar to a statement with the universal quantifier:

(a) Mx (Fx /\ Gx)

or

(y) Mx (Fx ~ Gx)

LEGAL NORMS AND LEGAL SYSTEM 31

(0) is clearly a bad candidate, since it does not mean that most philosophers are boring, but that most things are at the same time philosophers and boring. But (y) does not seem to work any better, since because of the equivalence of 'Fx ~ Gx' and '-,Fx Y Gx' statement (y) would have to be read as 'most things either are not philosophers or are boring'. Therefore, given the truth of the first disjunct, any statement beginning with 'Most philosophers .. .' would be true. This too is an implausible conclusion, because it generates truth-conditions for the majoritarian quantifier that are very far away from our basic intuitions about its use in language.28

We must therefore give up our attempt to represent statements with the majori­tarian quantifier through the use of connectives. This kind of sentences does not include connectives; they are rather relational sentences (similar to 'The moon is far from the sun'), but establishing a second-order relation. They do not establish a relation between individuals, but between predicates. The canonical formulation of 'Most philosophers are boring' thus could be:

(0) Mx (Fx, Gx),

which could be read as 'Most individuals satisfying predicate F also satisfy predicate G'. Therefore, the clause that should be added to our definition of satisfaction is another one:

(6') For any X, for any A and for any i: X satisfies the majoritarian quantification of A - say, 'Mx (Fx, Gx)' - with respect to var (i) iff most individuals satisfying 'Fx' also satisfy 'Gx'.

This seems to give an adequate account of sentences with the majoritarian quantifier, but the formulation is too far from that of the classical quantifiers, and it makes logical relations between those quantifiers impossible. Therefore, it also makes impossible logi­cal relations between the corresponding d-formulae which reflect the relationships be­tween imperative statements.

28 Perhaps an idea of von Wright (l984b, 46) about the possibility of restricting the scope of quantifiers could help to avoid this difficulty. But I will not analyze that possibility here.

32 CHAPTER I

ApPENDIX B: RELEVANCE, ARGUMENTATION AND NORMATIVE SYSTEMS

a) Relevance Beyond Logic

In several relatively recent studies on pragmatics, the notion of relevance plays a central role. They follow up on a suggestion by Grice (1975, 46) who, when speaking about the general conditions of conversation, mentions as one of them (as the category of rela­tion) the maxim 'Be relevant'. The question is what one must do in order to be relevant. If we want to answer that question we need an adequate conceptual reconstruction of the notion of relevance.

The work of Sperber and Wilson is one attempt in that direction. On the one hand, they assert that they "implicitly assumed that the process of inferential compre­hension is non-demonstrative" (SperberlWilson 1986a, 65). On the other, they hold that "utterance comprehension involves a substantial inferential element" (SperberlWilson 1986b, 244). In their view, that inferential element has to do with logic, but

"there is good reason to think that the logic used in utterance comprehension is not a standard one. On the one hand, it must be much more extensive, providing rules for every concept that can playa role in the in­ferential processing of propositions, including many that are of no particular interest to logicians. On the other hand, it must be more restrictive in certain ways" (SperberlWilson 1986b, 247).

Here, I am interested in how these authors think logic should be restricted in order to account for relevance. To this end, they propose that one should accept only those rules of logic (of propositional logic, we must assume) that never lead to 'irrelevant' - in the sense of 'trivial' - conclusions.

They assume that the restriction should apply to systems of natural deduction within propositional logic. Such systems contain rules of introduction and of elimina­tion for each connective.29 The proposed restriction is that all rules are elimination rules (1986b, 248). They thus construct the notion of non-trivial logical implication:

29 For a standard system of natural deduction like that of Gentzen (1934,176-210), these rules are:

Rule for conjunction introduction (AI):

A, B ~ A A B and also B A A Rule for conjunction elimination (AE):

A A B ~ A and also B Rule for disjunction introduction (vI):

A ~AvBandaisoBvA Rule for disjunction elimination (vE):

A v B, A ... C, B ... C ~ C Rule for conditional introduction (or deduction theorem) (~l):

A ... B~A~B Rule for conditional elimination (or modus ponens) (~E):

A~B,A ~B Rule for negation introduction (or reductio ad absurdum) (-J):

A ... BA-JJ ~-.A Rule for negation elimination (-,E):

LEGAL NORMS AND LEGAL SYSTEM 33

"A set of assumptions {P} logically and non-trivially implies an assumption Q if and only if, when {P} is the set of initial theses in a derivation involving only elimination rules, Q belongs to the set of final theses." (SperberIWilson 1986a, 97)

They then proceed to analyze under what conditions adding a new piece of information {P} to a given context (C) - a set of statements representing a store of information­allows one to gain contextual implications. Thus, the notion of contextual implication is defined in terms of the contextualization of {P} in (C) - i. e., of the union-set of {P} and (C):

Contextual implication (SperberlWilson 1986a, 107)

"A set of assumptions {P} contextually implies an assumption Q in context {C} if and only if (i) the union of {P} and {C} non-trivially implies Q, (ii) {P} does not non-trivially imply Q, and (iii) {C} does not non-trivially imply Q."

Their approach to relevance, then, is bound up with contextual implication: The more contextual implications a proposition has in a given context, the more relevant it is (SperberlWilson 1986b, 249):

Relevance (SperberlWilson 1986a, 122)

"An assumption is relevant in a context if and only if it has some contextual effect in that context."

Thus, we can say that the relevant consequences of an argument A are those that are contextual implications of some set of premises {P}. The advantage of this notion is that some of the consequences we had regarded as paradoxical no longer apply. From

(1) p -+ q. (2) P

we can now deduce neither 'q v r' (since we have no rule like that of disjunction intro­duction) nor 'r -+ q' (since we have no rule for conditional introduction), nor many other conclusions which we intuitively regard as irrelevant.

The problems with this conception of relevance are, on the one hand, that the lo­gical system these ideas give rise to is not rigourously spelled out (Sperber and Wilson could counter this30 by saying that they want to offer a pragmatic characterization of

-,-,A I-A Expressions of the kind '@ ... #' should be read as '# is obtained from @ (by rules of logic)' .

30 Instead, they argue that their system is somewhat similar to Parry's system of analytic implication, as pre­sented by Anderson/Belnap 1975,430-432; but although PAl (Parry's analytic implication) does not accept in the system a rule that would enable one to go from 'p' to 'p v q' (i. e., there is no rule for disjunction in­troduction, which is the cause of so many of the irrelevant conclusions we draw), it does have an explicit

34 CHAPTER I

the notion of relevance, rather than try to construct an alternative to the standard notion of logical consequence), and, on the other, that a system without introduction rules loses its inferential capacity.

Sperber and Wilson (1986b, 258 and n. 7; 1986a, 99-101) mention two aspects of their notion that may be problematic:

a) Since there are no introduction rules, the following inference is impossible (cf. n. 29):

(1) P (2) q

This is a case of the rule of conjunction introduction. Sperber and Wilson argue that the deductive capacity of human beings is imperfect, and that there are valid arguments they cannot directly grasp. Allegedly, only by showing that we cannot accept the pre­mises and at the same time deny the conclusion can the validity of the above argument be shown. This assertion looks very much like an ad hoc hypothesis: It is implausible that the only way for us to accept the argument is by showing that it is inconsistent to assert at the same time the truth of the negation of the conclusion and the truth of the premises (this is a kind of a rule of reductio ad absurdum which, besides, in the calculi of natural deduction is also an introduction rule).

b) They also wonder how the following argument, which they regard as valid, can be represented in their terms:

(1) If the trains are on strike and the car has broken down, there is no way of getting to work.

(2) The trains are on strike. (3) The car has broken down.

Therefore, there is no way of getting to work.

A standard system of logic would formalize and solve the case as follows:

(1)(p /\ q) -H

(2) p (3) q

2,3 (4) p /\ q 1,2,3 (5) r

Al2,3 -+E(MP) 1,4

rule for conjunction introduction and thus violates SperberlWilson's criterion for non-trivial implication. For them, in contrast to PAl, one cannot obtain 'p A q' from 'p, q'.

LEGAL NORMS AND LEGAL SYSTEM 35

But step (4) involves the use of an introduction rule. Therefore, they say that we should postulate another rule: the rule of conjunctive modus ponens (CMP)

(A" B) ~ C, A ~ B ~ C,

with which we can now show the validity of the argument as follows:

(1) (p " q) ~ r (2) p (3) q

1,2 (4) q ~ r 1,2,3 (5) r

CPM 1, 2

~E(MP) 3, 4

However, the rule of conjunctive modus ponens is nothing but a rule derived from any standard system of natural deduction presupposing the use of conjunction introduction. Actually, it is an ad hoc elimination rule which only apparently succeeds in renouncing the use of an introduction rule.31

Nor do their reflections on the psychological plausibility of those rules seem very convincing, since we do not seem to have much experimental evidence of the psy­chological accessibility of conjunctive modus ponens, which is accepted by the authors, whereas we do have evidence of the accessibility of the rule for conjunction introduc­tion, which the authors reject (cf. SperberlWilson 1968a, 100 f.).

Besides these problems, the inferential capacity of a system of natural deduction that renounces the use of introduction rules is very limited. Without the rule for condi­tional introduction - the deduction theorem - the validity of the following argument (from deontic logic) cannot be shown:

( 1) If a person is more than eighteen years old, she is of age. (2) If a person is of age, then she has the obligation to vote.

Therefore, if a person is over eighteen years old, she has the obligation to vote.

31 Most calculi of natural deduction would show that conjunctive modus ponens is a rule derived as follows:

3 2,3 1,2,3 1,2

(l)(A /\ B) ~ C (2) A

(3) B (4)A/\B

(5) C

(6) B ~ C

Assumption Al2,3 ~E(MP) 1,4 ~I3,6

This proof uses two introduction rules: the rule for conjunction introduction (step 2), and the rule for condi­tional introduction (steps 3 and 6). Something similar applies to what they call disjunctive modus ponens: (A v B) ~ C, A ~ C. This is a derived rule that presupposes the use of the rule for disjunction introduction.

36 CHAPTER I

In standard (deontic) logic, the validity of that argument can be shown as follows:

(1) p ~ q (2)q~Or

3 (3)p 1,3 (4) q 1,2,3 (5) Or 1,2 (6)p ~ Or

Assumption ~E(MP) 1,3 ~E(MP)2,4

~I3,6

One could say that this deduction can also be done using another rule (the transitivity of the conditional), but again this would only apparently solve our problem, since that is a derived rule and presupposes the more basic rule for conditional introduction.

In summary, Sperber and Wilson's notion of relevance succeeds in avoiding irre­levant consequences only at the forbiddingly high price of a loss of inferential capacity, to the point of obstructing the way to consequences that are clearly relevant. In the pre­vious example, a judge may need to show that the norm 'p ~ Or' is a norm derived from the system, and if we have no introduction rules, we cannot logically justify his decision.

b) The Logic o/Relevance

That an argument is logically valid means that if the premises are true then the conclu­sion will also be true. This requires that the set of premises {P} materially implies con­clusion C. Now, the notion of material implication has occupied a great number of logi­cians, because it gives rise to certain consequences regarded as paradoxical. Thus, Hun­ter (1993, 281) recently remarked:

"What distinguishes a decent logic of conditionals from the logic of material implication is the requirement that the antecedent of a conditional should be relevant to its consequent. Standard truth-functional logic does not meet that requirement ... A logic of conditionals must be a relevance logic of some kind."

In fact, the development of modern modal logic received its impulse from C. I. Lewis's dissatisfaction (1912, 1914, 1918) with the notion of material implication as used by Frege in his Begriffsschrift (1879) and by Russell and Whitehead in Principia Mathe­matica (1910). Lewis observes that this notion enables one to show that the following 'paradoxical' formulae taken from sentential or propositional logic are valid theorems:

p~(q~p)

-,p ~ (p ~q) (p ~ q) v (q ~ p)32

32 See the following proof (without premises) of the ftrst one of these theorems. The reader can easily prove the other two (although in the third the use of derived rules is suggested: De Morgan's laws - which inter-

LEGAL NORMS AND LEGAL SYSTEM 37

In the first case, the paradoxical nature lies in that if a statement is true then it is implied by any other statement; in the second, in that if a statement is false then it implies any other statement; and in the third, in that given any two statements, either the first im­plies the second, or the second implies the first.

Lewis proposes to avoid these paradoxical results with his notion of strict impli­cation, where for 'p' strictly to imply 'q' it is not required that either 'p' is false or 'q' is true (as with material implication), but instead that 'p' cannot be true and 'q' false. Thus, "a strict implication is a material implication which is necessary" (von Wright 1957, 170).

To this end, Lewis introduces a new symbol for strict implication (which I will represent by '-<') and defines it with the help of the modal notion of necessity (where 'it is necessary that' is represented by 'L'):

Lewis's concept, however, has been unable to avoid paradoxes altogether, because they could be reintroduced as paradoxes of strict implication. Thus, the following formulae are theorems of the various modal systems proposed by Lewis:

Lp ~ (q -<p)

L-,p ~ (p -< q).

That means that a necessary statement is strictly implied by any statement, and an im­possible statement strictly implies any statement. Lewis upheld these consequences and thought that those paradoxes are truths about the relation of deduction (cf. Haack 1978, 197 f.). For the second theorem, he proposed the following proof (a proof that consists in showing how an arbitrary conclusion can be deduced from an impossible premise: ex falso quodlibet):

l)PA-,p 2)p

(impossible premise) AE 1

define conjunction and disjunction - and the rules interdefining the conditional with conjunction and with disjunction):

I (1 ) p Assumption 2 (2) q Assumption 1,2 (3)pAq All, 2 1,2 (4)p AE3

(5)q~p ~I2,4

(6)p~(q~p) ~II,5

33 If we add the modal operator of possibility (represented by M) and define it as Mp =df -.L-.p, that defi­nition is equivalent to:

P -< q =df -M(p A -.q).

38

1 1 1

3)pvq

4)-.p 5) q

CHAPTER I

vl2 AEI MTP3,434

If one accepts this form of argument as valid, then that means that totally arbitrary (and, in that sense, irrelevant) conclusions can be deduced from a set of contradictory (im­possible) premises.

Therefore, for some logicians the relation of entailment expressed by the notion of strict implication is insufficient. For them, there must be some connection of mean­ing between the premises and the conclusion in order to say that the former imply the latter. The analysis of this idea has been the task of so-called relevance logic, in which the work by Anderson and Belnap (1975) stands out.35

My intention here is not to present Anderson and Belnap's conception or their ambitious logico-mathematical exposition in great detail. For my purpose, it is enough to sketch their main ideas, partly in an informal manner.

The logic of relevance is designed to guard against the paradoxes of both mate­rial and strict implication. According to AndersonlBelnap, for 'A ~ B' to be true (in the sense that A entails B), A must be relevant for obtaining B. That means, on the one hand, that it must be possible to use A in the deduction of B from A (AndersonlBelnap 1975, 18 and 30 f., construct a formal technique to this effect), and, on the other, that A and B must share some meaning content:

34 This is, in fact, an example of the so-called disjunctive syllogism or modus tollendo ponens [A v B, -.A I­B] which, in Gentzen's system of natural deduction presented in n. 29, is a rule derived from the rule for dis­junction elimination and some other rules:

I)A vB 2)-.4

3 3)A Assumption 2,3 4)AII--.4 Al2,3 5 5)-,8 Assumption 2,3,5 6)(A II -.4) 1\ -,8 Al4,5 2,3,5 7) A 1\ --.4 AE6 2,3 8) -,-,JJ -J5,7 2,3 9)B -,E8 10 1O)B Assumption 11 11) -,8 Assumption 10,11 12) B 11-,8 AlI0, 11 10 13) --,--,]J -J 11,12 10 14)B -,EI3 1,2 15)B vE 1, 3-9,10-14

35 This work, which was co-ordinated by Anderson and Belnap but to which other prominent logicians also made contributions, was to appear in two volumes. Following Anderson's death in late 1973, the publication of the second volume was delayed until 1992.

LEGAL NORMS AND LEGAL SYSTEM 39

"The second formal condition above is suggested by the consideration that informal discussions of implica­tion or entailment have frequently demanded 'relevance' of A to B as a necessary condition for the truth of A ~ B, where relevance is now construed as involving some 'meaning content' common to both A and B ... A formal condition for 'common meaning content' becomes almost obvious once we note that commonality of meaning in propositional logic is carried by commonality of propositional variables. So we propose as a necessary, but by no means sufficient condition for the relevance of A to B in the pure calculus of entailment that A and B must share a variable." (Anderson/Belnap 1975, 32 f.)

Based on these ideas, they construct an axiomatic systems for relevant implication (R-7) (AndersonlBelnap 1975,20):

Ri. A-7A

R2. (A -7 B) -7 ((C -7 A) -7 (C -7 B)) R3. (A -7 (B -7 C)) -7 (B -7 (A -7 C)) R4. (A -7 (A -7 B) -7 (A -7 B)

To this, they add other axioms concerning the other connectives, which allow them to obtain the complete system (E-7). Note especially that the complete system contains the following two axioms (8 and 9) about disjunction (AndersonlBelnap 1975,231 f.):

(E8) A -7 (A vB)

(E9) B -7 (A vB)

I will now try to show the problems this conception raises with respect to the exclusion of the 'paradoxical' logical consequences presented earlier:

i) First, the logic of relevance is unable to eliminate one of the most paradoxical factors about relevance. In that logic, disjunction can be introduced, since 'A -7 (A v B)' is an axiom. Thus, with the addition of the deontic operators to the language of the logic of relevance, the following is a valid inference (in the sense in which Anderson and Belnap redefine the notion of logical validity):

(1) P -7 Oq (2) P

~ O(q v rJ36

That this is a valid argument in the logic of relevance implies that that logic cannot rid us of the problems burdening the logical conception of the justification of judicial deci-

36 Since from (1) and (2), we can deduce 'Oq', and from 'Oq' we can deduce O(q v r). This is one possible formulation of what is called Ross's Paradox (Ross 1941). One reason for Ross's mistrust in deontic logic was precisely the possibility to deduce the norm 'Mail the letter or bum it' from the premise 'Mail the let­ter'. Ross's Paradox, as is known, has given rise to an extensive debate in deontic logic.

40 CHAPTER I

sions.37 'O(q v r)' is a logical consequence (in the logic of relevance as well) of (1) and (2) and, therefore, ajustified decision.

ii) Second, I will now analyze the criticism Anderson and Belnap direct against Lewis's deduction, commented earlier, of 'q' from 'p v -.p'. In the authors' words:

"In rejecting the principle of the disjunctive syllogism. we intend to restrict our rejection to the case in which the 'or' is taken truth functionally. In general and with respect to our ordinary reasoning this would not be the case; perhaps always when the principle is used in reasoning one has in mind an intensional meaning of 'or'. where there is relevance between the disjuncts. But for the intensional meaning of 'or'. it seems clear that the analogues of A ~ (A v B) are invalid. since this would hold only if the simple truth of A were sufficient for the relevance of A to B; hence. there is a sense in which the real flaw in Lewis's argument is not a fallacy of relevance but rather a fallacy of ambiguity. The passage from b to d [i. e .• from 'po to 'p v q'] is valid only if the 'v' is read truth functionally. while the passage from c and d to e [i. e .• from '-,p. and 'p v q' to 'q'] is valid only if the 'v' is taken intensionally." (Anderson/Belnap 1975. 165 f.)

But to reject the disjunctive syllogism also means, as the authors admit (AndersonlBel­nap 1975, 165), to reject modus ponens for material implication (though. of course, not for their relevant implication), since" 'A v B' and '-,A' imply 'B'" is equivalent to" '-,A

~ B' and '-,A' imply 'B'''. And modus ponens is a rule which, in the words of the authors (ibid.), "has perhaps never been seriously questioned before".

With the loss of modus ponens, inferential capacity is greatly reduced. As San­ford (1989, 131) observes, to reject modus ponens (or the disjunctive syllogism) "ap­pears to many a worse cure than the disease it aims to abolish": "Relevance logic rejects too much".38

For these two reasons, the logic of relevance is unable to eliminate those irrele­vancies that are most important for the deductive conception of the justification of judi­cial decisions and, instead, deprives us of important inferential mechanisms without which we cannot justify decisions we all regard as justified.

c) Logic Plus Relevance

In this section, I will try to defend the following thesis: The content of a judicial deci­sion is justified if and only if it is a relevant logical consequence (in the sense of classi­cal logic, extended to account for deontic logic) of the normative premise(s) and the statements describing the facts of the case.

In order to do this, I need a criterion for distinguishing, among the logical conse­quences entailed by a set of premises, those that are relevant and those that are irrele-

37 Hernandez Marin (1989. 306 f.) has used examples like this one to criticize the so-called logical concep­tion of the application of the law.

38 One can agree with Hunter (1993.283) that "one thing a hospital should not do is spread disease; and one thing logic should not do is teach people that invalid arguments are valid"; but it is also true that medicine should not kill the patient. and relevance should not destroy the inferential capacity of logic.

LEGAL NORMS AND LEGAL SYSTEM 41

vant. To this effect, I will use the notion of relevant (logical) conclusion as it appears in a recent paper by Schurz (1991).39 Schurz's idea is simple, yet elegant:

DF 1. Assume {P} ~ C. Then: C is a relevant conclusion of {P} if and only if no propositional variable in C is replaceable on some of its ocurrences by any other propositional variable, salva validitate of {P} ~ C. Otherwise, C is an irrelevant conclusion of {p}.40

The irrelevance of a conclusion C can result from the addition of propositional variables (through disjunction introduction or similar rules). The paradoxical examples of irrele­vant judicial decisions presented above (cf. ch. l.3.c) are of this type. It is true that 'p --+ Oq' and 'p' imply 'O(q v r)' as well as 'r --+ Oq'. But these are cases of irrelevant con­clusions, since in both conclusions 'r' can be replaced by any other propositional varia­ble salva validitate.

Irrelevance, however, can also be due to the presence, as conclusions, of logical truths for which the presence of the premises is superfluous. From 'p --+ Oq' and 'p', 'Oq v -.Oq' can be deduced. But that conclusion is irrelevant because the propositional variable 'q' can be replaced by any other in both its occurences in the formula.

This is an interesting case, because the logic of relevance has attempted to show the irrelevance of those implications (regarded as paradoxes of material or strict impli­cation). For this notion of relevant conclusion, the two inferences

p ~qv-.q P/\-.p~q

(verum ex quodlibet) and (ex falso quodlibet),

which are valid in sentential logic, are cases of irrelevant logical consequences. In both cases, variable q can be replaced (in both its occurrences in the first formula, and in its single occurrence in the second) by any other formula salva validitate of the implica­tion.

This is the kind of irrelevance von Wright intended to avoid with his notion of entailment. As he observes,

,,p entails q, if and only if, by means of logic, it is possible to come to know the truth of p ~ q without coming to know the falsehood of p or the truth of q" (von Wright 1957, 181).

Von Wright adds that the possibility of coming to know, by means of logic, whether or not a proposition is true means that that proposition is demonstrable. Thus, the concept of entailment is intimately linked to the concept of demonstrability. With this idea, the definition can be reformulated in the following way:

39 In that paper, Schurz attempts to show how his notion serves to eliminate a great number of paradoxes re­suiting mainly from the rule for disjunction introduction (also known as addition rule). Ross's Paradox is among the paradoxes analyzed by Schurz.

40 Cf. Schurz 1991,409. His definition is somewhat more complex, in order to be able to apply it to predi­cate logic too. But we can ignore that complication here, since I only use sentential or propositional logic.

42 CHAPTER I

,,[1 entails q, if and only if, by means of logic, p -+ q is demonstrable independently of demonstrating the falsehood of p or the truth of q." (von Wright 1957,181)41

Schurz (1991, 411) points out that von Wright's notion can be explicated with his own notion of irrelevant logical consequence or, more precisely, with his notion of com­pletely irrelevant logical consequence, which is a consequence where all occurrences of propositional variables can be replaced salva validitate.

Now, it may be useful to analyze what is gained and what is lost with that con­cept of relevant logical consequence in comparison with a logic of relevance like that of Anderson and Belnap.

One thing we gain is consistency with our presystematic intuitions about appa­rently paradoxical consequences. Disjunction introduction, which is a valid rule in rele­vance logic, is the paradigm of irrelevance here. And for this gain we do not need to re­duce drastically the inferential capacity of our logic: The disjunctive syllogism (and with it, modus ponens) is not irrelevant in our logic. 'q' is an irrelevant consequence of 'p /\ -p' not because the disjunctive syllogism is irrelevant, but because of the irrele­vance (in Lewis's proof) introduced by the rule for disjunction introduction (Schurz 1991,413).

Furthermore, we gain the possibility of continuing to use the classical concept of logical consequence (extended to account for deontic logic). The point here is not to de­fend the view that the classical notion of logical consequence is the only correct one. It is simply that this notion belongs to the innermost core of our conceptual structure and, following Quine's arguments,42 should be revised only in extremis. As I try to show, the problems we encounter in justifying judicial decisions do not constitute so extreme an emergency as to call for sacrificing the notion of logical consequence.43

However, there is also some loss, as compared to the logic of relevance. Above all, we lose certain elegant properties of the notion of relevant consequence (Schurz 1991,412 f.). Thus, the notion of logical consequence of Anderson and Belnap'S logic of relevance (in what follows: LCAB) - just like the standard notion of logical conse­quence - is closed under substitution, whereas the notion of relevant logical conse­quence (RLC) is not. Hence, while 'p' is an LCAB of 'p /\ -p', because it is an instance

41 Von Wright (ibid.) adds that, using the symbols 'M' for 'possible' and 'D' for 'demonstrated', we can express that definition as follows: 'p entails q' =df 'M (D(p -+ q) A -.D-tJ A -.Dq)'.

42 In Quine's words (1970,100): .. Logic is in principle no less open to revision than quantum mechanics or the theory of relativity. The goal is, in each, a world system - in Newton's phrase - that is as smooth and simple as may be and that nicely accommodates observations around the edges. If revisions are seldom pro­posed that cut so deep as to touch logic, there is a clear enough reason for that: the maxim of minimum muti­lation."

43 Perhaps the philosophical status of my position on relevance is close to that of Orayen (1989, 234-255; this work, besides, contains an excellent exposition of the ideas of Anderson and Belnap, as well as a sharp critique of them): .. I am inclined to think that there are no theoretical reasons that justify abandoning the classical analysis of deducibility for one offered by some logic of the relevant type. This involves rejecting relevant logic as a logic diverging from classical logic: it would not be a good substitute. Still, perhaps rele­vant logic can have some other use, despite of being regarded ill-suited to replace the ojficiallogic."

LEGAL NORMS AND LEGAL SYSTEM 43

of 'A /\ --,A ~ A', it is not a RLC, since 'p' can be replaced by any other propositional variable salva validitate. LCAR satisfies the properties of transitivity and monotonicity, whereas RLC does not.44 Still, this is not so very serious if we recall that the notion of relevance is not intended to replace the notion of logical consequence, but only to con­stitute a criterion for distinguishing relevant from irrelevant logical consequences.

Another problem we must address is particular to deontic logic. Standard deontic logic accepts the inference according to which Op ~ Pp (as a consequence of an axiom or a theorem, depending on how the calculus is presented). And yet we would be sur­prised to see a judge arguing conclusion 'Pq' as a result of the premises {p ~ Oq, p}.45 If we want to say that 'Pq' is an irrelevant logical conclusion of the premises, we must have a notion of relevance that not only affects propositional variables, but deontic ope­rators as well.

One way of responding can be found in the distinction presented at the begin­ning of this chapter, between maximal and partial solutions. It should be recalled that only maximal solutions completely determine actions from the normative point of view.

Thus, we can present the following definition of a relevant logical conclusion with respect to deontic operators:

DF 2. Assume {P} ~ C. A logical conclusion C is relevant with respect to deontic operators if and only if those deontic operators of C that constitute partial solutions cannot be deduced from other formulae that are consequences of {P} as well and that contain deontic operators that constitute maximal solutions.

Thus, a partial solution is a relevant logical conclusion of a set of premises A only when A does not provide sufficient information for deriving a maximal solution. Otherwise, the partial solution is irrelevant. That is what happens with the derivation of 'Pq' from {p ~ Oq, p}, since from these premises we can obtain the maximal solution 'Oq', from which, in turn, 'Pq' can be derived. On the other hand, DF 2 implies that if all deontic operators in C are maximal solutions, then C is relevant with respect to its deontic ope­rators.

44 Schurz (1991, 414) gives the following examples. The first is to show the non-transitivity ofRLC: (p v q)

A r ~ p v (q A r) and p v (q A r) ~ (p v q) A (p v r); in contrast, (p v q) 1\ r Hp v q) 1\ (p v r) is not a case of relevant deduction, because the second occurrence of pin (p v q) A (p v r) can be replaced by any other for­mula salva validitate; that is, it is possible for B to be a relevant consequence of A, and for C to be a relevant consequence of B, while C is an irrelevant consequence of A. Regarding monotonicity (the rule according to which if A ~ B and A ~ C, then C ~ B), the following case shows that the notion of RLC is not monotonous: p v q ~ p v q, but {p, p v q} ~ p v q is not a case of relevant deduction, since q can be replaced by any other formula salva validitate.

45 It would be surprising if a judge argued as follows:

(1) If x does A, then x must be punished with sanction S. (2) x does A.

Therefore, (3) x may be punished with sanction S.

The content of (3) is deduced, in deontic logic, from the premises; but (as in the cases in which the rule for disjunction introduction is used) it is weaker than it needs to be.

44 CHAPTER I

If we call Schurz's definition of a relevant conclusion (DF 1) the definition of a relevant logical conclusion with respect to propositional variables, we can say:

DF 3. A logical conclusion is relevant in deontic logic if and only if it is relevant with respect to proposi­tional variables and with respect to deontic operators (or, what is the same, if and only if it is DF l-relevant and DF 2-relevant).

d) Relevance and Normative Systems

This notion of relevant conclusion in deontic logic can by useful for a revision of our notion of a normative system. As I will explain in the next chapter, legal propositions contained in statements like 'Legally, all F ought to do iP' presuppose that certain norms belong to the legal system - in this case, the norm contained in the norm-formulation 'All F ought to do tp'. Now, since 'All F ought to do iP' implies 'All F ought to do iP or ought to do (1', the proposition contained in 'Legally, all F ought to do iP or ought to do C1' is true as well. But this conclusion is counterintuitive; and one can hardly attribute to a norm-authority the intention (if only an implicit one) of enacting the irrelevant logical consequences of its explicitly enacted norms.

In this context, Raz (1994a, 211 f.) has distinguished between the source thesis according to which all law is based on certain social acts of norm-creation, and the in­corporation thesis according to which all law is either based on sources or implied by law based on sources. In order to explain that distinction, Raz compares sets of norms with sets of beliefs and suggests that one usually does not attribute to a person all the logical consequences of what she explicitly believes. Similarly, Raz thinks that authori­ties do not prescribe all the things implied by the norms they explicitly enact and that, therefore, the incorporation thesis is incompatible with an approach to law as invested with authority.

There are other reasons, however, for which it is important to preserve the notion of a normative system as a deductive system containing all its logical consequences. The most important of those reasons has to do with the questions of legal dynamics which will be treated in Chapter III. But it will be important later to keep in mind the distinction between relevant and irrelevant logical consequences in the set of all logical consequences of a normative system.

In what follows, I will show this importance with respect to the topic of norma­tive contradictions.

One of the aspects that have sometimes been criticized about the conception of legal systems as normative systems is that in the case where a set of formulated norms contains a contradiction, because of the ex falso quodlibet rule, any norm will belong to that normative system. Thus, one can say that a normative system S is inconsistent if and only if any norm belongs to S.

FI/.Illesdal and Hilpinen (1971, 16) define what they call the principle of the con­sistency of a normative system as follows:

LEGAL NORMS AND LEGAL SYSTEM 45

"If a set of sentences A is consistent and (Of,. Of, •...• Of •• Pg) !;;; A. then (f,. f, •...• f •• g) is consistent.,,46

That means that if a set of norms contains a norm ordering not-p as well as another norm permitting p, then it is inconsistent since {p, -,p} is inconsistent. The correspon­ding normative system would thus contain any norm, since a contradictory set implies any statement. As I have shown, however, none of these consequences is a relevant consequence, since in each of the formulae each of the propositional variables can be replaced by any other salva validitate (i. e., they are completely irrelevant consequen­ces). Thus, from the normative system {O-,p, Pp} we can deduce Oq, Phq, Fq, etc., but none of these formulae is a relevant consequence, since variable q can be replaced by any other salva validitate (including, of course, by variable p). We can thus define the notion of an inconsistent normative system (containing a normative contradiction) with the help of the notion of relevant logical consequence:

A nonnative system S is inconsistent if and only if it lacks relevant consequences. or all its consequences are completely irrelevant.

This definition may help us understand in what sense an inconsistent normative system is defective: It contains any norm, but none of these norms is relevant.47

46 The authors add: "It should be observed that [this principle] does not require that all pennitted states of affairs can be realized simultaneously. but only that each pennission is compatible with all obligatory states of affairs" (emphasis added).

47 Recently. Atienza (1992. 1017 f.) has criticized Alchourr6n and Bulygin's notion of a legal system as a nonnative system precisely because. in that case. if a system contains a nonnative contradiction then that would imply that any nonn belongs to that system: .. According to this notion. the jurist who wants to recon­struct some part of the legal order would have to take into consideration that the existence of a contradiction in that part - however small it may be - leads to all kinds of consequences (since from two contradictory statements any other statement follows). For example. if the question is to reconstruct the constitutional sub­system. and he detects a contradiction in it. in the sense that. say. for the approval of a certain type of law one constitutional nonn requires a majority of two thirds. and another only a simple majority. the jurist who strictly wants to apply Alchourr6n and Bulygin's notion of a legal system would have to conclude that in that legal order the constitution stipulates that 'anything goes': laws must be approved by parliament. or not; citizens have the right to habeas corpus. or not; etc. Obviously. no jurist in his right mind - and not even a somewhat deranged jurist - would accept this. So how is it possible that Alchourr6n and Bulygin have -at least implicitly - sustained this. and - for more than twenty years! - have not done anything to correct that notion of a legal system? Can one even correct it without abandoning the classical notion of logical con­sequence?" In my opinion. the introduction of the notion of relevant logical consequence shows that we need not abandon the notion of logical consequence. We must only remember that jurists are interested in the relevant normative consequences of a legal system. and that an inconsistent system lacks such conse­quences. Therefore. what must be done is to 'dissolve' the contradiction. i. e. to refonnulate the inconsistent system - or subsystem - to make it consistent (which is. besides. what jurists usually do).

II. A LOGICAL ANALYSIS OF LEGAL PROPOSITIONS

1. Introduction

The purpose of the present chapter is to present a logical analysis of statements like the following:

(1) Legally, all F ought to do 4>. (2) Legally, x ought to do 4>.

(3) Legally, all F are 4>. (4) Legally, x is a 4>.

Such statements are canonical formulations of expressions that do not always come in the same grammatical form. For instance, the word 'legally' is often omitted. Thus, one says 'x is real estate', where the term 'legally' is implied by the context, or 'x has the obligation to pay that tax', where the same is implied. Obviously, 'legally' in those cases means 'according to some specific legal system'. (1) and (2) are regarded as the canonical formulations of all deontic expressions, such as

and

(la) Legally, all F ought not to do 4> (i. e., all F are prohibited to do 4»; (lb) Legally, all F may do 4> (i. e., all F are permitted to do 4> or, what amounts to the same, it is not obligatory for an F to omit 4»;

(2a) Legally, x is prohibited to do 4>.

(2b) Legally, x is permitted (in some contexts: x has the right) to do 4>.

I will call such statements legal statements. And I will say that legal statements express legal propositions.

2. Deontic Legal Statements and Conceptual Legal Statements

I have assumed that norms belonging to a legal system LS are of two kinds: a) norms in the strict sense or prescriptive rules, and b) conceptual rules.

A legal statement like formulation (1) implies that a certain prescriptive rule is a normative consequence of LS. Now, we can simplify legal statements by realizing that to say that 4> is legally obligatory ('04>') is equivalent to saying that '04>' belongs to the normative consequences of LS: '04>' E NC (LS); and to assert that 4> is not legally ob­ligatory is to assert that '-,04>' belongs to the normative consequences of LS: '-,04>' E

NC(LS).

48 CHAPTER II

Since '-,OI/J' is equivalent to 'p-,I/J', to say that legally I/J is not obligatory is the same as saying that legally it is permitted that not-I/J.l I will call legal statements that re­fer to normative consequences deontic legal statements.

A legal statement like (3) presupposes that a certain conceptual rule is a norma­tive consequence of LS. For reasons of simplicity, I will, for the time being, represent conceptual rules through propositional variables. To say that legally p (where p can be replaced by statements like 'Persons 18 years or older are of age', or 'Killing another person is manslaughter', etc.) is the same as saying that p is a consequence of the re­spective legal system: 'p' E NC (LS). To say that legally not-p is to say that -,p is a consequence of LS: '-,p' E NC (LS).

Legal statements like (3) and (4) will be called conceptual legal statements.

3. Pure Legal Statements and Applicative Legal Statements

The truth of propositions expressed by statements like (1) and (3) depends exclusively on the existence of certain legal norms in a given system; if a norm like 'All F ought to do I/J' is among the consequences of legal system LS, then (1) is true, and if a conceptual rule like 'All Fare I/J' is among the consequences of the legal system, then (3) is true. Such statements will be called pure legal statements.

In contrast, the truth of propositions expressed by statements like (2) and (4) can depend on the existence of certain legal norms and on the truth of certain statements of fact. 2 Thus, (2) can be analyzed in the following terms:

(2c) The norm 'All F ought to do I/J' belongs to NC (LS), and x is an F.

An analysis of (4) would be:

(4a) The conceptual rule 'All Fare I/J' belongs to NC (LS), and x is an F.

Often, whether x has a particular property (whether it is real estate, an owner, an accom­plice, of age, etc.) depends on certain legal qualifications. Thus, if a norm-authority is-

1 If we take into consideration the distinction between strong and weak permission (von Wright 1963a, Al­chourronlBulygin 1984), 'It is legally permitted that not 1/1' must be understood as 'It is permitted in the strong sense'. A conduct is permitted in the strong sense in a system S if, and only if, a normative conse­quence of S explicitly pennits it. A conduct is permitted in the weak sense in S if, and only if, no normative consequence of S prohibits it. Weak permission of a conduct does not guarantee the truth of the correspond­ing legal statement. I will come to this later.

2 If, in the system in question, an individual norm like 'x ought to do 1/1' or a conceptual rule like 'x is a 1/1'

can be deduced, then statements (2) and (4) would also be pure statements (the truth of the propositions they express would depend exclusively on whether or not those individual norms belong to a certain system). For reasons of simplicity, I will assume that legal systems contain only general norms. Thus, the truth of the pro­positions expressed in (2) and (4) will always depend on whether or not certain general norms can be shown to belong to the system and whether or not the propositions expressed by certain statements of fact are true.

A LOGICAL ANALYSIS OF LEGAL PROPOSITIONS 49

sues a norm according to which persons of age have the obligation to vote and a con­ceptual rule like 'Persons aged 18 years or older are of age', then whether x is of age depends on that conceptual rule. In those cases, we must assume that (2c) refers to the normative concequence of both the prescriptive and the conceptual rule, i. e., that 'All F ought to do t/>' stands for 'Persons aged 18 years or older must vote'.

Statements of types (2) and (4) will be called applicative legal statements.3

4. Truth-Conditions of Legal Statements

The proposition expressed by legal statement (l) is true if, and only if, there is a norma­tive consequence of LS that makes it obligatory for all F to do t/>. But when is that pro­position false? I will say that it is false when there is a normative consequence of LS that does not make it obligatory for all F to do t/>, i. e., that permits an F to omit t/>. What happens if in LS there is neither a normative consequence making it obligatory to do t/> nor one permitting to omit t/>? In that case, I will say that the proposition in question has no truth-value: it is then neither true nor false that legally all F ought to do t/>.

But are there propositions without a truth-value? If the answer to this question is affirmative, as I wish to suggest, we must revise our conception of logic: the law of bi­valence, which says that all propositions are either true or false, cannot then be retained unconditionally. In contemporary philosophy, this idea is often linked to Dummett's (1978, 1991) view that there is a strong connection between realism and bivalence. Ac­cording to Dummett, there are three interrelated kinds of realism, such that if you reject one of them, you must also reject the others (cf. Engel 1991, 129):

(I) Metaphysical realism: Independently of our knowledge, there is something in the world which can make our propositions true. (II) Semantic realism: The meaning of a statement is determined by its truth con­ditions, independently of how we may be able to verify them. (III) Realism in logic: The principle of bivalence, i. e., the principle that all pro­positions are either true or false, is accepted as a fundamental principle of logic.

Dummett's position, in short, is that all these forms of realism must be rejected and re­placed by:

(I') Metaphysical antirealism: Our propositions are not true because of a reality that is independent of our ability to verify it; reality is relative to the knowledge we have of it. (II') Semantic antirealism: The meaning of our statements is not determined by their truth-conditions, but by their conditions of assertability and use. (III') Antirealism in logic, or revisionism: The principle of bivalence, as well as the law of excluded middle, is rejected and instead, the revision of the laws of

3 The distinction between pure and applicative statements can be found in Raz (1970, 45-50; 1979, 62; 1980,218; 1994b, 181 f.) and, expressed in different terms, in Hernandez Marin (1989, 270-284).

50 CHAPTER II

classical logic implied by this rejection in favour of an intuitionist conception of logic is defended.

Dummett's project is ambitious, and it has been widely reviewed and discussed (see, e. g., Wright 1993, 1-43). My purpose here is more modest. I do not wish to defend a glo­bal antirealist position. Realism may be an adequate conception for some sectors of human knowledge. But since the law is a human construction, it seems plausible to hold an antirealist or constructivist conception with respect to the class of legal propositions.

A parable often used in the literature (e. g., Blackburn 1984, 203-210; Dworkin 1985, 146-166) can throw some light on what I mean. In that literature, statements of literary criticism, referring to objects of fiction,4 are compared to legal statements. What are the truth-conditions of statements referring, e. g., to aspects of the fictional character of Madame Bovary? Obviously, the truth of such statements depends on the story con­structed by Flaubert. Here, it does not matter what theory of literature we subscribe; what matters is only that it seems plausible to assume that statements like 'In Flaubert's novel, Madame Bovary's blood group was A' have no truth-value. What makes this as­sumption plausible is precisely metaphysical antirealism with regard to what we can call the Flaubert-world. The Flaubert-world does not exist independently of our possible knowledge of Flaubert's novel. There is no Flaubert-world independently of Flaubert's novel and, therefore, of our knowledge of that novel, that could make statements about Madame Bovary true or false. This metaphysical antirealism in literature is accompani­ed by semantic antirealism: Only when we are able to assert some property of Madame Bovary we can attribute truth or falsity to the statement that, according to Flaubert's novel, attributes that property to Madame Bovary. But then, there are reasons for re­jecting bivalence, since if it cannot be shown that Madame Bovary, according to Flau­bert's novel, is a iP, nor that she is not a iP, then the statement 'In Flaubert's novel, Ma­dame Bovary is a iP' has no truth-value.5

Something similar happens in law (cf. Patterson 1996, 3-21). The truth of legal statements too depends on whether certain consequences can be obtained in some LS. The truth of the statement 'Legally Gaius is of age' may depend on whether one can ob­tain - whether one can prove - a consequence like 'Persons aged 18 or older are of age' in a particular LS (as in current Spanish law, in virtue of art. 12 of the Constitu­tion), and on whether Gaius is 18 years or older. But if we ask about the truth-value of the statement 'Legally, Gaius is a fan of FC Barcelona', it seems plausible to say (as in

4 For a distinction between literature and fiction, cf. Searle 1979, 58-60.

5 Dummett himself (1978, 230) has held this view for fictional characters: "Thus, to say that the fictional characters are the creations of imagination is to say that a statement about a fictional character can be true only if it is imagined as being true, that a fictional character can have only those properties which it is part of the story that he has; to say that something is an object of sense - that for it esse est percipi - is to say that it has only those properties it is perceived as having: in both cases, the ontological thesis is a ground for re­jecting the law of excluded middle as applied to statements about objects. Thus we cannot separate the ques­tion of the ontological status of a class of objects from the question of the correct notion of truth for state­ments about thse objects; i. e. of the kind of thing in virtue of which such statements are true, when they are true."

A LOGICAL ANALYSIS OF LEGAL PROPOSITIONS 51

the case of Madame Bovary's blood group) that, since there is no corresponding norm in LS, that statement is neither true nor false.

We thus get the metaphysical thesis of legal antirealism or constructivism: There is no legal world, beyond our capacity of knowing the law as constructed by human beings, that can make legal propositions true or false.6

And we also have the semantic thesis of legal constructivism: The meaning of legal statements is determined by their condition of assertability, that is, by the ability to prove that certain consequences obtain in some legal system LS.1

These theses, in turn, imply the logical thesis of legal constructivism: Not all legal propositions are true or false.

But then, what is the logic that applies to legal propositions? Revisionism in logic which proposes to accept an intuitionist logic raises problems I cannot treat here (cf. Haack 1978,216-220). For this and other reasons, I choose what can perhaps be called an intermediate way: the truth-logic (or logics) constructed by G. H. von Wright (1984c, 1988, 1989).8 In truth-logic, a minimal deviation from classical logic brings with it the ability to account for the problems encountered in treating propositions with­out a truth-value.

The only heterodox characteristic of truth-logic (TL) is that the notion of truth is introduced in the object-language with the symbol T. T functions as a modal operator; prefixing it to a well-formed expression produces a new well-formed expression. It is read as 'it is true that' , and forms what I will call T-expressions.

With the help of T, we can distinguish two ways of negating a statement: extern­al negation ...,T, to be read as 'it is not true that', and internal negation T..." meaning 'it is true that not'. An internal negation asserts falsity and can therefore also be read as 'it is false that'. In TL, falsity and not-truth are not the same: not-truth is weaker than falsi­ty. A false proposition is always not-true, but not all not-true propositions are false (as, for example, propositions that are neither true nor false). This is the novelty in TL: it can account for propositions without a truth-value.

But, can there really be propositions without a truth-value? Are not all proposi­tions necessarily true or false? In another paper, von Wright (1984a) demystifies the no­tion of a proposition as follows: The basic notion is that of a sentence (a grammatically well-formed expression). A proposition is then defined as 'a move of language': a grammatically well-formed sentence expresses a proposition if, and only if, the sentence obtained by prefixing it with the expression 'it is true that' is well-formed too. If one

6 This thesis is intimately linked to legal positivism, with its thesis that the existence of law in a society de­pends only on certain social facts, i. e., on human acts. Cf. Moreso 1994a, 353 f.

7 Dworkin (1977a, 8) posits a strong relationship between positivism and semantic antirealism, with special reference to Dummett's characterization: "I think that many positivists rely, more or less consciously, on an antirealist theory of meaning. They think that no sense can be assigned to a proposition unless those who use that proposition are all agreed about how the proposition could, at least in theory, be proved conclusively."

8 Although the most important ideas can already be found in von Wright 1984c, the system I will present is one of those constructed in von Wright 1989, which was already anticipated in von Wright 1988 and which has some important particularities, among them especially the acceptance of mixed formulas.

52 CHAPTER II

accepts that notion of a proposition, perhaps it will be easier to accept that some propo­sitions have no truth-value. Thus, 'Prime numbers are blue' seems to express a proposi­tion without a truth-value. That it expresses a proposition can be verified by noting that 'It is true that prime numbers are blue' is a well-formed sentence.9

Although von Wright himself has developed some ideas for constructing a quan­tificational truth-logic (von Wright 1984b), TL is an extended propositional logic.

The basic symbols of TL are:

1) Propositional variables: p, q, r. 2) Sentential connectives: " 1\, V, ~, H.

3) Brackets: (, ). 4) Operator: T.

The notion of aformula of TL is defined recursively:

I') All propositional variables are formulae of TL. 2') If a is a formula of TL, then so is .a. 3') If a and ~ are formulae of TL, then so are a 1\ ~, a v ~, a ~ ~ yaH b. 4') If a is a formula of TL, then so is Ta. lO

The axioms of TL are the following:

AO. All formulae obtained from tautologies of classical, two-valued, proposition­allogic by putting the letter T immediately in front of every variable occurring in the tautologous formula. AI. Tp H T.op . A proposition is true if, and only if, its negation is false. A2. T(P 1\ q) H Tp 1\ Tq. A conjunction is true if, and only if, all its conjuncts are true. A3. T.(p 1\ q) H Top v T-.q . A conjunction is false if, and only if, at least one of its conjuncts are false. A4. Tp ~ p. If it is true that p, then p.

The rules of inference or transformation are as follows:

RI. Substitution of formulae for variables (a variable is also a formula). R2. Detachment (Modus ponens).

9 Other examples more widely discussed in philosophical scholarship are statements about contingent fu­tures, like 'Tomorrow there will be a naval battle'; statements containing descriptions referring to inexisting entities, like 'The King of France is bald' (cf. Appendix); some counterfactual conditionals, like 'If Kelsen had been born in Barcelona, he would have spoken Catalan'; or statements containing vague terms, when they refer to cases that fall within the penumbra of the referent of those expressions, like 'Maria (referring to a person aged 65) is old'. I will say more about this last type later.

lO The conventions for brackets are the usual ones.

A LOGICAL ANALYSIS OF LEGAL PROPOSITIONS 53

R3. The Rule of Truth: if a is an axiom or theorem of TL, then so is Ta. ll

TL can be interpreted as a three-valued logic: true, false, and neither-true-nor­false- indetermined - (represented in what follows by '+', '-', and'?'). A truth-table may help us understand some of the special characteristics of TL:

p

+

? + ?

Tp

+

Top

+

...,Tp

+ +

...,Top

+

+

In the same way, we can construct the truth-tables for all the other connectives of TL.12 Now, some comments about TL are warranted.: (i) Although propositional variables can express propositions without a truth­

value, T-expressions in TL always express propositions that are true or false. Therefore, the heterodoxy of TL is limited: T -expressions follow the rules of classical logic.

(ii) In classical logic, the law of excluded middle is usually represented by p v op: all propositions are either true or not true. But T(p v op) is not a theorem of TL. The law of excluded middle is retained in TL in a more restricted form: Tp v...,Tp.

(iii) The law of bivalence, according to which all propositions are either true or false - Tp v Top -, is not retained in TL either, since it is possible that ...,Tp /\ ...,Top. That is precisely what happens when p has no truth-value. But we can give a restricted version of the law of bivalence, and that version is retained in TL: T(p v op) H Tp v Top (it is true that p-or-not-p if and only if it is true that p or it is true that not-p). And it

11 This is one of the systems of von Wright 1989 (called TLM). In von Wright 1989, 25 f., it is shown that the systems of 1984c and of 1988 which in addition to AD - A3 contained the axiom

A5. Tp -+ ....,T--.p and (only the system of von Wright 1984c) the axiom

A6. T....,Tp H ....,Tp are contained in the system of 1989. This means that A5 and A6 are demonstrable in that system (as von Wright explicitly acknowledges, the proof is due to Carlos Alchourr6n).

12 Here are the truth-tables for conjunction, disjunction, conditional and biconditional:

p

+ + +

? ? ?

q

+

?

+

?

+

?

+

?

?

?

pvq p-+q pHq

+ + + + + ? ?

+ + + +

? + ?

+ + ? ? ? ? ? ?

54 CHAPTER II

is also a logical law of TL that it is either true that p is true or it is false that p is true: TTp v T -,Tp. That law is perhaps the best-suited candidate for being called the law of bivalence in that logic.

(iv) Since -,T-,p is equivalent to T-,T-,p, from the theorem Tp ~ -,T-,p (cf. n. 11) it follows that Tp ~ T -,T -,po That means that if a proposition is true, then it is false that that proposition is false. But the converse does not hold. As von Wright remarks (1988, 13), this characteristic, together with the rejection of the classical law of bival­ence, is what makes TL similar, though not equivalent, to an intuitionist logic.

(v) It is precisely this similarity to intuitionist logics which makes TL an ade­quate system for the logical treatment of legal propositions in a constructivist approach. Intuitionists like L. E. 1. Brouwer and A. Heyting (cf. Haack 1978, 216-220) claim that classical logic cannot adequately account for mathematical reasoning. Their construc­tivist approach to mathematics sees it basically as a mental activity, and numbers as mental entities. Thus, to say that there is a certain number with certain properties is to say that such a number can be constructed; if we cannot prove that a certain number has property P, then that there is such a number that is P has no truth-value. In the opinion of intuitionists, there is no mathematical reality independently of our capacity to con­struct mathematical systems. And that metaphysical thesis (rejecting Platonism in mathematics) is accompanied by the semantic thesis that truth in mathematics is prov­ability, and by the logical thesis that rejects bivalence in the classical sense (that every proposition is either true or false). Thus, Dummett's interest (1978,215-247) in the phi­losophical consequences of intuitionism in logic is not surprising.

In my view, in that sense, legal propositions are similar to mathematical proposi­tions. Thus, it is legally true that all F ought to do t{J if, and only if, there is (can be proved to be) a normative consequence of LS making it obligatory for all F to do t{J, and it is legally false that all F ought to do t{J if, and only if, there is a normative conse­quence of LS permitting an F not to do t{J. If in LS there is no norm making it obligatory to do t{J or permitting not to do t{J, then I will say that legally it is not true that all Fought to do t{J, and that legally it is not false that all F ought to do t{J.

The distinction introduced in Chapter I between relevant and irrelevant logical consequences of a set of statements can help sharpen this approach to legal proposi­tions. It would seem absurd to say that since the norm 'All F must vote' implies 'All F must vote or sleep fifteen hours a day' the legal proposition expressed by the statement 'Legally, x (who is an F) must vote or sleep fifteen hours a day' is true. In order to avoid such absurdities, I will say that the truth of legal propositions depends on whether some relevant consequence can be proved to exist in a system. Therefore:

A legal proposition is true with respect to LS if and only if the nonn it refers to is a provable relevant con­sequence in LS, and it is false if the negation of the nonn it refers to is a provable relevant consequence in LS. If neither the nonn nor its negation are provable relevant consequences in LS, then that legal proposition has no truth-value. I3

13 For reasons of simplicity, unless otherwise indicated, in what follows the expression 'provable in the sys­tem' should be understood to mean 'provable as a relevant consequence in the system'.

A LOGICAL ANALYSIS OF LEGAL PROPOSITIONS 55

We can now use TL for a logical analysis of legal statements like (1) 'Legally, all F ought to do cp'. If we insert the expression 'it is (not) true that' or 'it is (not) false that' between 'Legally' and the rest of (1), we get what I will call legal T-statements. Legal T-statements always have a truth-value. The legal T-statement corresponding to (1) has the following possible forms:

(LTSla) Legally, it is true that all F ought to do cpo

(LTSI b) Legally, it is false that all F ought to do cpo (LTSlc) Legally, it is not true that all F ought to do cpo

(LTSld) Legally, it is not false that all F ought to do cpo

(LTSla) implies (LTSld), and (LTSlb) implies (LTSlc). But neither does (LTSld) imply (LTSla), nor does (LTSlc) imply (LTSlb). In the case where (1) has no truth-value, although legally it is not true that all F ought to do cp, neither is it false that all Fought

to do cp, i. e., (LTSlc) and (LTSJd) can both be true. The same can be said for statements like (2). The legal T-statements correspon­

ding to (2) are:

(LTS2a) Legally, it is true that all F are cpo (LTS2b) Legally, it is false that all Fare cpo (LTS2c) Legally, it is not true that all Fare cpo (LTS2d) Legally, it is not false that all Fare cpo

Note that 'Legally, it is true that' and 'Legally, it is false that' do not exhaust the logical space, whereas 'Legally, it is true that' and 'Legally, it is not true that' do.

The analysis of legal T-statements corresponding to (3) and (4) must be more elaborate. As I have shown, (3) and (4) are conjunctive statements.

Let us begin with (3). The analysis of (3) must be in terms of

(LTS3a) Legally, it is true that x ought to do cp,

because (LTS3) must show the molecular structure of (3) in a form like

(LTS3a') Legally, it is true that all F ought to do cp, and it is true that x is an F.

Thus, for

(LTS3b) Legally, it is false that x ought to do cp

we have correspondingly:

(LTS3b') Legally, it is false that all F ought to do cp, or it is false that x is an F.

56 CHAPTER II

This is merely an application of axioms A2 and A3 of truth-logic, according to which T(P 1\ q) f-7 Tp 1\ Tq and T-.(p 1\ q) f-7 T-,p v T-.q, and the assumption that the expres­sion 'legally' only is prefixed to 'it is (not) true' or 'it is (not) false' in the case of state­ments referring to normative consequences of LS.

(LTS3c) Legally, it is not true that x ought to do ¢

will be analyzed with the help of

(LTS3c') Legally, it is not true that all F ought to do ¢ or it is not true that x is F.

And finally,

(LTS3d) Legally, it is not false that x ought to do ¢

is to be analyzed with

(LTS3d') Legally, it is not false that all F ought to do ¢ or it is not false that x is anF.

With the necessary changes, the same applies to the legal T-statements corresponding to (4).

Now, is it possible that a statement like 'x is an F' has no truth-value? In other words: Is the expression 'It is true that x is an F' equivalent to 'It is not false that x is an F'? Factual propositions attributing properties to individuals (e. g., the property of be­ing 18 years or older to x) or events (e. g., the property of being a murder to an action of x) may behave according to the laws of classical logic, and their truth-value may de­pend on their correspondence with empirical reality, since truth in that case is not equi­valent to demonstrability in a legal system. But as we will see later, it is important to preserve the possibility that such propositions have no truth-value, because it allows us to approach the treatment of factual propositions containing vague concepts.

5. Legal Statements and Contradictions: The Law Speaks With Many Voices

If it could be shown in LS - if it were a consequence of LS - that all F are ¢ and that all F are not ¢, or that all F ought to do not-¢ and that all F may do ¢, then the proposi­tions expressed by the legal T-statements 'Legally, it is true that all Fare ¢' and 'Legal­ly, it is true that all F ought to do not-¢' would be at the same time true and false. This idea can be expressed by saying that the legal world corresponding to that LS is an im­possible legal world. There is no logical reason that could prevent an authority to issue commands or qualifications (through conceptual rules) that are contradictory (in the sense, already explained, that there is no possible world that could make such a norma­tive system effective). Although this leads to unfortunate consequences for the addres­sees of such norms, it is a rather common situation in complex legal systems.

A LOGICAL ANALYSIS OF LEGAL PROPOSITIONS 57

But, as I have shown in Chapter I, those normative consequences are irrelevant, since an LS that contains a contradiction has no relevant consequences. If the truth­value of a legal proposition depends on the demonstrability in LS of a relevant conse­quence, then the legal statements 'Legally, all F ought to do not-</>' and 'Legally, all F may do </>' express propositions without a truth-value, since all the logical consequences of LS are irrelevant.

The conclusion of all this is that all legal propositions, expressed by legal state­ments, that refer to a legal system LS containing contradictions, i. e., an LS with anti­nomies, have no truth-value; and all propositions expressed by legal T-statements like 'Legally, it is not true that .. .' and 'Legally, it is not false that .. .' referring to that LS are true.

To say that a normative system that contains contradictions has no relevant con­sequences can perhaps explain why such a system is not an adequate system for the re­gulation of human behaviour: it cannot serve as a guideline for action.

6. Legal Statements and Normative Gaps: The Law Is Silent

The question of whether legal systems are, by definition, complete or whether they can have gaps has been - and still is - widely discussed in legal philosophy.14

Since the word 'gap' has a wide variety of uses in legal theory, an attempt to re­construct and elucidate those different uses seems to be needed. In my view, the best in­vestigation of gaps in the law is that of Alchourr6n and Bulygin (1971, 17-21, 31-34, 94-115,116-143).

What the different uses of the word 'gap' in legal contexts seem to have in com­mon is that whenever gaps are mentioned, there is some degree of indeterminacy in the application of the law. Alchourr6n and Bulygin (1971, 31) observe that this indetermi­nacy can derive from "problems of a conceptual type arising at the level of generic cases and general norms" or from "empirical and semantic problems arising from the application of general norms to individual cases". These two cases must be carefully distinguished - which has not always been done in legal theory - in order to clarify the different questions treated under the label of 'gap'.

One of the most important problems arising in the application of general norms to individual cases is the classification of the individual case, i. e., what jurists call 'sub­sumption' or 'legal qualification'.

The problems arising in the context of subsumption can have one of two possible sources. The first is lack of information about the facts of the case. Alchourr6n and Bu­lygin call such cases gaps of knowledge. 15 If a general norm prescribes that all persons who in a given year earned more than $ 10.000 have the obligation of presenting a de­claration of income, our ignorance of whether or not Ticius had a higher income may

14 For the thesis that legal systems can have no gaps, cf. Kelsen 1960,251-255; Dworkin 1989,128-134.

IS AlchourronlBulygin 1971,33: "Where, through ignorance of some of the properties of the fact, we do not know whether a certain individual case belongs to a certain generic case, we shall speak of a gap of knowl­edge."

58 CHAPTER II

prevent us from knowing whether his individual case can be subsumed under the gene­ric case regulated by that general norm. We then do not know the truth-value of the le­gal statement 'Legally, Ticius has the obligation of presenting a declaration of income' because we do not know whether the statement 'Ticius earned more than $ 10.000' is true. That statement, however, is true or false with respect to the real world; therefore (assuming a realist conception of the real world) it does have a truth-value, although we may not know which.

But the difficulty may also arise, not out of our ignorance of the facts of the case, which perhaps we have perfect knowledge of, but because we cannot determine, e. g., whether certain gifts Ticius received from some of his clients should count as income. The problem then is one of semantic indeterminacy or vagueness, not of ignorance, be­cause the problem subsists even if we know perfectly well how much the gifts received by Ticius are worth. Alchourr6n and Bulygin call such cases gaps of recognition. 16

In the next section, I will argue that in that case the statement 'Ticius had an in­come of more than $ 10.000' has no truth-value and, more generally, that problems of vagueness stemming from the use of certain general expressions referring to cases of penumbra give rise to statements which express propositions that have no truth-value. Here, I only wish to stress that problems from a lack of knowledge, or ignorance, are different from problems from vagueness, or semantic indeterminacy.17 We will very likely never find out the truth-value of the statement 'On September 25, 1995, there was an uneven number of blades of grass on the Bellaterra Campus of Barcelona's Auto­nomous University'; 18 yet, that statement is either true or false (to that class of state­ments, bivalence applies). This question is different from that of whether or not a parti­cular blade of grass on the Bellaterra Campus is green. If it is greenish, with a yellowish overtone, it may be impossible to decide whether or not it is green, even though we possess all the information one can have. Our doubts in that case do not arise from a lack of information.

The problem I wish to treat in this section, however, concerns another use of the expression 'gap', namely, what Alchourr6n and Bulygin call normative gaps, i. e. situa­tions where a generic case of a universe of cases is not correlated with any maximal normative solution.19 This, again, is not a problem of the application of general norms to individual cases, but a conceptual question: the absence of a normative solution for a generic case.

16 Alchourr6n1Bulygin 1971, 33: .. Where, through semantic indetenninacy of the concepts which charac­terize a generic case, we do not know whether a certain individual case belongs to it, we shall speak of a gap of recognition."

17 Recently the thesis has been defended, however, that problems of vagueness are merely a special kind of problems of ignorance. Cf. Williamson 1994. I will come back to that question in the next section.

18 Cf. Quine 1981,31-37.

19 Strictly speaking, I should add: 'or where a generic case is correlated with a disjunction of two or more maximal solutions' (this would be what Alchourr6n1Bulygin [1971, 20 f.] call a panial gap). For reasons of simplicity, in this chapter I will omit this complication. - Alchourr6n and Bulygin analyze still another use of the word 'gap', i. e., what they call 'axiological gaps'. I will come to them in Chapter IV.

A LOGICAL ANALYSIS OF LEGAL PROPOSITIONS 59

The Spanish Civil Code, for instance, regulates the reimbursement of expenses and improvements in case of a transference of possession. Now, does a former posses­sor have the right to be reimbursed for expenses and improvements (or to withdraw im­provements consisting in separable objects)? The Spanish Civil Code regulates such cases (arts. 453-455), taking into account the following criteria: whether the expenses were necessary, useful or superfluous, and whether the claimant possessed the object in good or in bad faith. I am not interested here in the complex solution the Code gives to those cases; I only wish to point out that the case of useful expenses made by a posses­sor in bad faith has no normative solution (Alonso 1995).20 Thus, while the legal state­ment 'Legally, Gaius (a former possessor in good faith who carried out necessary im­provements of the object in his possession) has the right to be reimbursed for what he spent on the improvement of the object' (which means that the new possessor has the obligation of paying Gaius back what he spent) is a statement expressing a true proposi­tion, the legal statement 'Legally, Ticius (a former possessor in bad faith who carried out useful improvements of the object) has the right to be reimbursed for what he spent on the improvement of the object' expresses a proposition without a truth-value, since there is no normative consequence in the normative system in question that correlates the fact of useful expenses made by a possessor in bad faith with a normative solution.

Therefore, when there are legal gaps, legal statements referring to cases falling within those gaps express legal propositions without a truth-value.

But this is a very controversial conclusion. In what follows, I will briefly con­sider the positions of Kelsen, Dworkin and Raz who, for reasons that do not fully coin­cide, reject that conclusion and assert the contrary: that the law is always complete, and that in cases falling within an area of normative gaps, legal statements always express propositions that are either true or false.

a) Kelsen: gaps as fiction

Kelsen's thesis that legal systems never have (i. e., cannot have) gaps is related to the question of a judge's obligations in the application of the law. However, in my presen­tation I will try to keep the two questions apart, because in my view the answers are not necessarily connected.

Kelsen's conception can be summarized in the following two theses (cf., e. g., Kelsen 1945, 146-148; 1960,251-255):

20 Simplifying somewhat, one can say that the Spanish Civil Code regulates those cases by conceding the more rights the more the claimant possessed the object in good faith and the greater the need of the expenses and improvements. Thus, if the expenses were necessary and the claimant possessed the object in good faith, then he has the right to be reimbursed, and to withhold the property until the new possessor has paid him for the expenses. In the case of a former possessor in bad faith who carried out unnecessary improvements, he only has the right to keep those improvements that consist in separable objects, provided the new possessor prefers not to buy them from him at the price of the value of the objects at the time of transference. The problem is that in arts. 453-455, which regulate that situation, no mention is made of useful expenses or improvements made by a possessor in bad faith.

60 CHAPTER II

(I) Legal system cannot have gaps, because a behaviour that is not prohibited by the law is permitted.21 (II) In cases where there is no norm obligating a particular person to a certain be­haviour, if that person is sued by another, then the competent judge applies the law by dismissing the case.22

The link between (I) and (II) is expressed by Kelsen in two different ways: Until 1960, he maintained that a judge acquits the defendant (or dismisses the case) by applying a negative rule according to which no-one can be obligated to behave in ways that are not prescribed by law (cf. Kelsen 1945, 147); but in 1960, Kelsen asserts that "by dis­missing the case or acquitting the defendant, the court applies the legal order which per­mits the defendant the behaviour against which the action or charge, without foundation in the legal order, was directed".23 That means that in 1960 Kelsen holds that judges ap­ply the legal order as a whole, and no longer uses the argument of the negative rule for acquitting the defendant or dismissing the suit in the case of a legal gap.

Thesis (I) has been convincingly criticized by Alchourr6n and Bulygin (1971, 119-124). The Argentine authors distinguish two versions of thesis (I) which they call the Principle of Prohibition (,Everything which is not prohibited is permitted'). The two versions correspond to two possible meanings of the term 'permitted' in the Prin­ciple of Prohibition: 'permitted' in the strong sense, and 'permitted' in the weak sense (von Wright 1963a, 86-90; Alchourr6n 1969, Alchourr6n1Bulygin 1971, 121 f.; Her­mindez Marin 1989,337 f.; MoresolNavarro 1992):

A behaviour b is strongly permitted in the (generic) case c in a legal sys­tem LS if, and only if, a normative consequence can be inferred from LS which permits b in case c.

A behaviour b is weakly permitted in the (generic) case c in a legal system LS if and only if no normative consequence can be inferred from LS which prohibits b in case c.

21 Cf. Kelsen 1945. 147: "Just because no nonn exists which obligates the defendant to the behavior claimed by the plaintiff, the defendant is free according to positive law, and has not committed any delict by his behavior"; and Kelsen 1960,251, where he asserts that "wenn die Rechtsordnung keine Pflicht eines In­dividuums zu einem bestimmten Verhalten statuiert, sie dieses Verhalten erlaubt".

22 Kelsen 1945, 147: "If the judge dismisses the suit, he applies, as it were, the negative rule that nobody must be forced to observe conduct to which he is not obliged by law"; Kelsen 1960,251: "Die Anwendung der geltenden Rechtsordnung ist in dem Fall, in dem die traditionelle Theorie eine Lucke annimmt, nicht 10-gisch unmoglich. Zwar ist in diesem Faile die Anwendung einer einzelnen Rechtsnonn nicht moglich, aber die Anwendung der Rechtsordnung, und auch das ist Rechtsanwendung, ist moglich. Rechtsanwendung ist nicht logisch ausgeschlossen."

23 Kelsen 1960,248: "Indem das Gericht die Klage abweist oder den Angeklagten freispricht, wendet es die Rechtsordnung an, die dem Beklagten oder Angeklagten das Verhalten erlaubt, gegen das sich die in der Rechtsordnung nicht begriindete Klage oder Anklage gerichtet hat."

A LOGICAL ANALYSIS OF LEGAL PROPOSITIONS 61

The two meanings of 'permitted' give rise to two versions of the Principle of Prohibi­tion: the weak version of that Principle says that 'Everything which is not prohibited is weakly permitted', and the strong version says that 'Everything which is not prohibited is strongly permitted'.

The weak version of the Principle of Prohibition is an analytic statement that is necessarily true. It says that all behaviour that is not prohibited in a legal system is not prohibited in it. But the truth of the weak version of the principle does not guarantee that there are no gaps. Rather, it is compatible with the existence of gaps: Whenever there is a normative gap (a generic case without a normative solution), the correspon­ding behaviour is weakly permitted.

If the strong version of the Principle of Prohibition, according to which for every behaviour that is not prohibited in a legal system LS there is a normative consequence in LS expressly permitting that behaviour, were true, it would guarantee that all norma­tive systems are closed, i. e., have no gaps. But the truth of the strong version of the principle is contingent. In the words of Alchourr6n and Bulygin (1971, 127):

,,For from the mere fact that a certain norm (the norm to the effect that p is prohibited in q) does not belong to a certain system, it does not follow that another, different norm (namely, the norm to the effect that p is permitted in q) belongs to the system. Therefore the Principle of Prohibition, in its strong version, is not necessarily true."

To sum up: the Principle of Prohibition - Kelsen's thesis (I) - does not preclude the presence of gaps in legal systems. In its weak version, though expressing a proposition that is necessarily true, it is compatible with the existence of unregulated cases, of gaps. In its strong version, it expresses a proposition that is only contingently true and there­fore only guarantees the absence of gaps for those systems where it expresses a true proposition.

For the analysis of Kelsen's thesis (II), according to which in cases where there is no legal norm that makes a particular behaviour obligatory the competent judge ap­plies the law in dismissing the case, we will need to introduce the notion of applicabili­ty and the distinction between norms that belong to a legal system and norms that are applicable in a legal system (which will be developed in the next chapter). For now, it is sufficient to recall a few obvious points:

Not all branches of the law behave in the same way (that is, have the same crite­ria of applicability) in unregulated cases: While in criminal law, under the rule of law judges are prohibited (by the so-called principle of legality) to sanction behaviour that is not subject to any norm, in civil law they may (or must) make use of the so-called analogical application of norms, with the consequence that suits based on unregulated facts can (or must) be accepted (as in the civil-law case I referred to earlier). To say that a judge applies the law in those cases by dismissing the suit would be regarded as a highly counterintuitive consequence. Kelsen himself seems to be aware of this:

,,But it is also possible that the legal order empowers the court not to dismiss the case or not to acquit the defendant, in case it cannot find a general legal norm imposing on the defendant the duty violation of which

62 CHAPTER II

is alleged by the private plaintiff or the public prosecutor, and instead to admit the suit or convict the ac­cused, if it regards the lack of such a general legal nonn as unjust, or unfair, i. e. as not satisfactory. ,,24

As I have shown, Kelsen's transition from thesis (I) to thesis (II) is supported either by the doctrine of the application of a negative rule (until 1960) or by the doctrine of the application of the legal order as a whole (after 1960; cf. also 1979, xxxi). Alchourr6n and Bulygin maintain that this shift of argument is due to an evolution in Kelsen's con­ception of the meaning of what I have called thesis (I). According to Alchourr6n and Bulygin (1971, 130-134), until 1960 Kelsen interprets the Principle of Prohibition in its strong version (i. e., 'Everything which is not prohibited by the law is explicitly per­mitted by the law'), whereas beginning in 1960 Kelsen interprets it in its weak version (i. e., 'Everything which is not prohibited by the law is weakly permitted'). The reason they give for this change in Kelsen's thinking is the following: Kelsen (1960, 249) now recognizes that there may be a behaviour that is not prohibited, and in that sense is per­mitted, to some individual and which prevents another individual to perform another behaviour which also is not prohibited and, therefore, is permitted. In that case, Kelsen adds, "there is ... a conflict of interest not prevented by the legal order; and no legal or­der can prevent all possible conflicts of interest".25 That is why Alchourr6n and Buly­gin (1971, 132 f.) hold that Kelsen changed his position and came to adopt the weak version of the Principle of Prohibition, implicitly admitting that there may be gaps in the form of conflicts of interest not prevented by the legal order.

More recently, Ruiz Manero (1990, 41-45) has rejected the interpretation of Al­chourr6n and Bulygin, arguing that the version of the Principle of Prohibition of the Kelsen of 1960 also is the strong version. According to Ruiz Manero, in the cases Kel­sen calls "conflicts of interest not prevented by the legal order", all behaviour in ques­tion is permitted in the strong sense, and therefore judges in such cases have the duty to dismiss the case (if it is a civil suit) or to acquit the defendant (in case of criminal pro­ceedings). In that sense, Kelsen still would not distinguish cases of strong and of weak permission. For Kelsen, the two following situations would be equivalent: A legal sys­tem LSI containing a norm permitting behaviour p by person x, and another norm per­mitting behaviour q by person y which prevents the performance of p; and a legal sys­tem LS2 containing no norm prohibiting behaviour p by person x and no norm prohi­biting behaviour q by person y, where the latter prevents the performance of p. For Kel-

24 Kelsen 1960, 249: ,,Es ist aber auch moglich, dass die Rechtsordnung das Gericht enniichtigt, falls es kei­ne generelle Rechtsnorm feststellen kann, die dem Beklagten oder Angeklagten die PfIicht auferiegt, deren Verietzung der private Kliiger oder offentliche Ankliiger behauptet, die Klage nicht abzuweisen oder den Angeklagten nicht freizusprechen, sondem, wenn es das Fehlen einer solchen generellen Rechtsnonn fUr un­gerecht, unbillig, das heisst fUr nicht befriedigend halt, der Klage stattzugeben oder den Angeklagten zu ver­urteilen." Kelsen's repeated quotation (1945, 147; 1960, 252) in this context of the first paragraph of the Swiss civil code ("A defaut d'une disposition legale applicable, Ie juge prononce selon Ie droit coutumier, et au defaut d'une coutume, selon les regles qu'il etablirait s'il avait a faire acte de legislateur") also seems to point in that direction.

25 " ... liegt ... ein Interessenkonflikt vor, dem die Rechtsordnung nicht vorbeugt; und keine Rechtsordnung kann allen moglichen Interessenkonflikten vorbeugen".

A LOGICAL ANALYSIS OF LEGAL PROPOSITIONS 63

sen, the two systems would be equivalent, and insofar, Ruiz Manero is right in saying that there is an interpretation of the expression 'conflicts of interest not prevented by the legal order' which does not imply the acceptance of gaps. System LSI has no normative gap, though there is a conflict of interest left without a solution, whereas system LSz does have gaps and, therefore, unresolved conflicts of interest. The fact that Kelsen re­gards the two systems as equivalent shows that his interpretation of the Principle of Pro­hibition is, as Ruiz Manero claims, the interpretation of the strong version.

Still, Alchourr6n's and Bulygin's doubts seem justified since, as we have seen, Kelsen himself recognizes that a legal order may contain a norm authorizing judges to accept a suit or convict a defendant in the case of a system like LSz- This assertion by Kelsen suggests that he does see a difference between a system like LSI and a system like LSz- In LSI judges must dismiss a suit of y against x (or of x against y) because of the norm that permits y to do q (or x to do p). In LSz the duties of a judge depend on what other norms possibly say. If some person z is sued for having done r, and r is per­mitted by the legal system in the strong sense, then the judge must dismiss the case (otherwise we would not say that r is permitted); in contrast, if z is sued for having done r, and r is permitted in the weak sense - there is no norm prohibiting r - then the duties of the judge will depend on other norms of the system which - as a contingent matter - not always require him to dismiss the case against z.26

We can thus conclude that Kelsen did grasp, albeit not very clearly, the two sen­ses of the Principle of Prohibition, as Alchourr6n and Bulygin say; but, as Ruiz Manero holds, he still did not change his interpretation of the principle in his work of 1960.

Finally, Kelsen (1945, 147-149; 1960, 253-255) points out - and his commen­tators (Ruiz Manero 1990,45-48) emphasize - that the traditional doctrine of gaps is a fiction designed to authorize judges to issue individual norms whose content is not de­termined by general norms, in order to avoid results that would be unsatisfactory and unjust in the light of their moral and political beliefs. Without questioning the correct­ness of Kelsen's opinion on this matter, I only wish to point out that his idea (e. g., 1960, 251) that unregulated cases and (according to some criterion of justice) unjustly regulated cases are equivalent is totally unfounded. We can, and I think we should, dis­tinguish the two cases: We do have a clear criterion for determining cases of normative gaps (and we do not need any criterion of value for this); and they are different from cases that are regulated unjustly, which Alchourr6n and Bulygin call 'axiological gaps' (here, we do need a criterion of value). I will treat that matter in Chapter IV.

b) Dworkin and the completeness of the law

At least on one occasion, Dworkin (1989, 127 f.) has raised the question of gaps in a way very similar to the one used here:

26 In the case mentioned earlier, concerning the Spanish Civil Code, it would be considered contrary to law if a judge were to dismiss a suit filed by a former possessor in bad faith against the new possessor for the reimbursement of his expenses.

64 CHAPTER II

"I would like to explain logically what is understood by a 'gap' in the law. In order to do this, I think we must start from a specific proposition. Thus, let us take Neil MacCormick's example: 'The law authorizes me to ride my bicycle through the Bois de Boulogne today.' According to the thesis admitting the existence of gaps in the law, there are certain - maybe many - equally specific propositions that are neither true nor false. That is what the theory of gaps in the law says, and it is essential to distinguish this thesis - which I think is wrong - from a radically different one according to which it may tum out to be uncertain whether or not a legal proposition is true or false, there may be room for controversy about that question, and it may even be doubtful whether it can be solved. I wish to point out that this last kind of propositions, which speak of uncertainty and controversy, are compatible with realist theses about the ontology of law; they represent a realist approach, whereas the theory of gaps - a proposition is neither true nor false - represent an anti­realist approach ... 27

As can be seen, Dworkin uses the term 'gap' in a wide sense, including all cases in which legal statements express propositions without a truth-value (and which, therefore, lead to indeterminacy in the application of the law). Dworkin's arguments against an antirealist (or constructivist) conception of legal propositions, like the one advocated here, are of a mixed kind, but they are based on an interpretive conception of the law according to which a legal proposition is true if and only if it forms part, or is a conse­quence, of the best possible interpretation of a community's legal-political history (Dworkin 1989, 129, and esp. 1986, chs. 2 and 3). In any case, the analysis and critique of that conception and of how it guarantees the completeness of the law will be offered in the last chapter.

For the time being, I only wish to analyze one of Dworkin's arguments against the idea that the law is incomplete (and that, therefore, there are legal propositions which are neither true nor false), and that is an arguement which is independent of his interpretive conception.28

I mean the following argument: Dworkin (1985, 129 f.) holds that even if there are legal statements of the form 'x is q>' which are true, others which are false, and still others which are neither true nor false, "indeterminacy will not result if a principle of legislation is adopted which requires that if 'x is q>' is not true, it be treated as false". This is how bivalence is restored; 'Legally, it is not true that x ought to do q>' becomes

27 "Je voudrais expliquer de maniere logique ce qu'on entend par 'Iacune' du droit. Pour cela, je crois qu'il faut partir d'une proposition concrete. Ainsi, prenons I'exemple de Neil MacCormick: 'Le droit m'autorise a circuler en bicyclette dans Ie Bois de Boulougne aujourd'hui.' D'apres la these qui admet des lacunes dans Ie droit, il existe certains propositions, peut-etre nombreuses, d'un type aussi concret qui ne sont ni vraies ni fausses. Voila la theorie des lacunes du droit, et il est essentiel de distinguer cette these - pour moi erronee - d'une these radicalement differente selon laquelle il peut se reveler incertain de dire si une proposition juridique est vraie ou fausse, il peut y avoir place a controverse sur cette question, et meme il peut se reveler douteux de pouvoir Ie trancher. Je tiens a signaler que ces demieres propositions, qui parlent d'incertitude et de controverse, sont compatibles avec les theses realistes en matiere d'ontologie du droit; elles representent une approche realiste, tandis que la theorie des lacunes - une proposition n'est ni vraie ni fausse - repre­sente une approche antirealiste."

28 Dworkin actually presents his argument in the context of a discussion of the vagueness of certain con­cepts used in the law (of gaps of recognition). I will say more about his point of view in the next section. I do think, however, that his argument also applies to so-called normative gaps.

A LOGICAL ANALYSIS OF LEGAL PROPOSITIONS 65

equivalent to 'Legally, it is false that x ought to do ~' (which implies the truth of 'Le­gally, it is permitted for x not to do ~').29

Two brief comments on this argument of Dworkin: (i) Dworkin presents the argument against V, a fictitious author who defends the

trivalence of legal propositions. V could reject Dworkin's argument by grounding his position on the truth-logic TL and holding that the restauration of bivalence is what his constructivism tries to avoid. He could add that if the principle of legislation proposed by Dworkin is understood to say that if a proposition is not true, then it is false, it leads to a contradiction in TL. In the notation of TL, the principle of legislation (PL) can be represented as follows:

(PL) -.Tp ~ T-,p.

Now, V assumes that there are legal propositions without a truth-value, that is,

(1) -.Tp 1\ -.T-,p

(1) and (PL) are contradictory, as can easily be seen:

(2) -.Tp (3) T-,p

(4) -.T-,p

(5) T -,p 1\ -.T-,p

(elimination, (1» (modus ponens, (2) and PL)

(elimination, (1» «3) and (4»

Hence, we must reject either (1) or PL. V would conclude that PL, as proposed by Dworkin, does not restore bivalence in a trivalent system, but rather presupposes it, which means that it is inconsistent with a trivalent system.

(ii) Perhaps PL should not be interpreted in this way. Dworkin could be suggest­ing that PL is a norm for judges which prohibits them to convict people who have per­formed actions that are not explicitly prohibited (cf. Blackburn 1984,206-210).30 Now, Dworkin's argument resembles Kelsen's Principle of Prohibition in its strong version. As I have argued before, whether or not that principle (sometimes called a closure rule) belongs to a legal system is a contingent matter. What's more, in many spheres of the law, there is no such principle; thus, in civil law, that there is no norm explicitly pro­hibiting some conduct does not mean that suits filed against persons for performing that conduct always ought to be dismissed. The application of norms by analogy is groun­ded precisely on the opposite assumption.

29 For a more detailed discussion of Dworkin's argument, cf. Endicott 1997.

30 From this, Blackburn (1984, 209 f.) infers that it is wrong to identify a bivalent logical practice (like that of the judge in Dworkin's case) with the acceptance of realism.

66 CHAPTER II

c) Legal reasons, sources and gaps

Like Dworkin, Raz (1979,53-77) uses a broad notion of legal gaps. Let us begin with his conclusion (1979, 77):31

"The outcome of this discussion is that legal gaps are not only possible but, according to the sources thesis, inescapable. They arise, however, where the law speaks with an uncertain voice (simple indeterminacy) or where it speaks with many voices (unresolved conflicts). Contrary to much popular imagining, there are no gaps when the law is silent. In such cases closure rules, which are analytic truths rather than positive legal rules, come into operation and prevent the occurrence of gaps."

I agree with Raz that there is indeterminacy when we are confronted with problems of vagueness (gaps of recognition, as I will try to show in the next section) and also when there are genuine normative contradictions, which Raz calls "unresolved conflicts" (as I tried to show in the previous section). I also agree that in those cases the corresponding legal propositions have no truth-value (Raz 1979, 70 f.). Contrary to Raz, however, I do think that there can be normative gaps in the law and that, therefore, the legal proposi­tions corresponding to these also have no truth-value. In what follows, I will try to ar­gue that Raz cannot show at the same time that the law is complete (has no normative gaps) and that there are legal propositions without a truth-value, since these two theses are incompatible.

According to Raz (1979, 66), legal statements are statements about reasons for action:

"Legal sources are reasons for action. If s is a variable ranging over statements of social sources, then sLRx,1/I ('S is a legal reason for x to 1/1') is the general form of statements of legal reasons. Such statements are true if and only if s is true and is a statement of the appropriate social condition according to the doctrine of identity. Because the existence of the appropriate source is a reason for x to 1/1, its existence is the ground for the truth of statements of the form 'Legally x ought to 1/1' (LRx, 1/1)."

Thus, what Raz calls the sources thesis implies that

"Statements of the form pLRx, 1/1 are true only if statements of social facts specifiable without recourse to moral arguments are substituted for p".

Raz adds that legal statements are about conclusive reasons for action (a conclusive rea­son for x to do ¢ is "a reason which overrides all conflicting reasons, is not excluded by exclusionary reasons, nor cancelled by any cancelling conditions"; Raz 1979,64).

Raz links the problems of legal gaps to the application of the law. This leads him to distinguish jurisdictional gaps from legal gaps (Raz 1979, 70):

,,Jurisdictional Gaps. A legal system is jurisdictionally complete if its courts have jurisdiction over all legal questions. It has a jurisdictional gap if its courts lack jurisdiction over certain legal questions.

31 This is also the source of the expressions used in the titles of sects. 5-7 of the present chapter.

A LOGICAL ANALYSIS OF LEGAL PROPOSITIONS 67

Legal Gaps. A legal system is legally complete if there is a complete answer to all the legal questions over which the courts have jurisdiction. It contains a legal gap if some legal questions subject to jurisdiction have no complete answer."

Obviously, what we are interested in here are legal gaps (normative gaps) (modern legal system usually are jurisdictionally complete). According to Raz (1979, 71), there are two possible answers to a legal question:

(A) LR;C, ,p, the law conclusively requires the action; and (B) LPer;c, -,,p, i. e. the law conclusively permits the omission of the action.

Raz goes on to say that there is a legal gap if the law does not require any decision. And he adds:

"It follows that there are two possible kinds of legal gaps: (I) 'LR;C, ,p' is neither true nor false and 'LPer;c, ."p' (i. e. 'L-.R;C, ,p') is neither true nor false. (2) '(-.LR;C, ,p) A (--.LPer;c, ."p)' is true."

Now, Raz only accepts the existence of gaps in case (1). These are cases of legal inde­terminacy due to semantic indeterminacy (i. e., because of the vagueness of the con­cepts used in the law) or unresolved conflicts (i. e., because of incompatible reasons). Since in none of these cases there is a genuine normative gap (Raz's notion of 'gap' in­cludes all the cases of legal indeterminacy), we can ignore them for the time being.

In contrast, according to Raz, normative gaps would arise if (2) could be main­tained. Now, (2) is equivalent to

But, according to Raz (1979, 76), it is a logical truth that

From (3) and (4), it follows that

The proof of (5) is easy:

[a] ~,;X, tP [b] L-,R.,;X, tP [c] ~-,R.,;X, tP [d] (L-,R.,;X. tP) A (~-,R.,;X. tP)

Elimination of A in (3). Modus ponens in (4) and [a]. Elimination of A in (3). Introduction of A in [b],[c].

32 Because of the equivalence of '(--.LPer;c, ."p)' and '(-,L-,R.;C, ,p)'.

68 CHAPTER II

(5) is a contradiction. Thus, the logical truth of (4) implies the rejection of (2), i. e. the truth of (4) implies the rejection of the thesis of normative gaps.

If Raz's reasoning were sound, there would be no room for normative gaps in legal systems, since it would be impossible that the law at the same time neither prohi­bits an action nor permits its omission. Now, we must look at how Raz tries to prove the logical truth of (4).

He holds that the proof of (4) is easy. First, he tries to prove the biconditional from right to left, that is:

(6) L-.R,x, I/J ~ -,LR,x, ¢.

The proof starts with the following assertion (1979, 76): "It cannot be true that legally there is a conclusive reason to perform an act and at the same time that legally there is no such conclusive reason." This means that

(7) LR,x, ¢ ~ -,L-.R,x, ¢

and, by counterposition, (6) is proved. Now, let us look at the derivation of the bicondi­tional from left to right:

(8) (-,LR,x, I/J) ~ (L-.R,x, ¢)

Raz argues as follows (1979, 76):

"Assume that the antecedent is true. i. e. that LR"x. tfJ is false. Can it also be that L-.R"x. tfJ is false? If --.LR"x, tfJ. then it is not the case that there is a fact which is a conclusive legal reason to tfJ. But if there is no con­clusive reason to tfJ. can it fail to be the law that there is no conclusive reason to tfJ? This would have been a real possibility had there been a need for a special kind of fact to make true negative statements of reasons of the form --.Rx, tfJ ... But it has already been established that such propositions are verified by the absence of reasons only and it is precisely that absence which is entailed by --.LR,x. tfJ. Hence --.LR"x. tfJ ~ L-.R"x, tfJ. Q. E.D."

Raz then asserts that (8) is equivalent to

(9) (-,LR,x, I/J) ~ (L-,Per,x, --.I/J)

And he adds: "That is rendering of the familiar closure rule that whatever is not legally prohibited is legally permitted."

I think the reason why Raz regards (4) to be an analytic truth is his peculiar con­ception of permissive legal statements. Raz (1979, 64) distinguishes two basic kinds of permissions: explicit permissions and conclusive permissions:

(10) x has an explicit permission to ¢ (Per,x, ¢) =def There is a fact with some force to cancel reasons for no-</>-ing (--.¢-ing).

A LOGICAL ANALYSIS OF LEGAL PROPOSITIONS 69

And

(11) x has a conclusive permission to IP (Per~, IP) =def It is false that there is a conclusive reason for x to -,IP (i. e. -,Rc x, -,IP).

This means that "A conclusive permission to act is the contradictory of a conclusive reason for refraining from the act". The following expression is a logical truth:

(12) (Rc x, IP) ~ -,(Per~, -,IP)

It is the peculiar conception of conclusive permissions which leads Raz to assume the analytic truth of (4). Since an action is permitted conclusively when there is no conclu­sive reason prohibiting the performance of the action, one can show that all actions are conclusively determined by a normative system: They are either prohibited (by a con­clusive reason), or not prohibited (i. e., permitted by a conclusive reason). It is the logic of conclusive reasons which allows Raz to hold that all generic actions are normatively determined.

What happens, however, is that the logic of conclusive reasons can support only the weak version of the Principle of Prohibition, because, according to Raz, that a be­haviour is conclusively permitted means that its omission is not conclusively required. Thus, it is easy to see that his definition (2) of a gap leads to a contradiction. On the one hand, the analytic truth that all behaviour is either conclusively prohibited or not is asserted. On the other, a normative gap is defined as 'it is true that some conduct c is neither conclusively prohibited nor conclusively not prohibited'. Hence, there are no normative gaps. Actually, the definition of a gap is already contradictory (given the meaning of 'conclusively prohibited').

But the thesis of normative gaps presupposes that there is a gap whenever a con­duct is neither prohibited nor permitted (in the strong sense) by a normative system. In view of this, Raz's definition (2) of a gap is inadequate. According to Raz, there are no gaps because if a conduct is not regulated, then there is a conclusive reason permitting its performance, since there is no conclusive reason prohibiting it.

Using the logic of normative propositions, part of which is presented in the ap­pendix to this chapter, and given that for Raz the external and the internal negation of his legal statements of reasons are equivalent, the logical truth of (4) - -,(LR~, IP) ~ (L-,R~, IP) - can be represented as follows (where 'Op' is equivalent to IP being re­quired for x):

(4a) -,'Op' E L ~ 'Op' ~ L.

Obviously, (4a) is an analytic statement. Also, it does not come as a surprise that (3)­-,(LR~, IP) 1\ (-,L-,R~, IP), which is equivalent to the definition (2) of a gap - contains a contradiction. (3) is equivalent to

(3a) -,'Op' E L 1\ -,'Op' ~ L.

70 CHAPTER II

Since ,,-,'Op' e; L" is equivalent to" 'Op' E L", (3a) is plainly a contradiction. But the adequate definition of a gap actually corresponds to

(3b) -,'Op' E LA -,'-,Op' E L.

In order to show that (3b) is contradictory, one must prove the analytic truth of

(4b) -,'Op' E L ~ '-,Op' E L.

Neither the logic of conclusive reasons nor any other logic can guarantee the analytic truth of (4b), because the absence of a norm making p obligatory is perfectly compati­ble with the absence of a norm permitting the omission of p.

Let us now see whether the acceptance of the analytic truth of (4) is compatible with definition (1) of a gap, as Raz holds. Raz accepts the possibility that legal state­ments of reasons, like 'LR~, t/J', have no truth-value. That possibility cannot be expres­sed with his logical symbolism. For Raz, 'It is not true that legally there is a reason .. .' is equivalent to 'It is false that legally there is a reason .. .'. That is what is expressed in his logical truth in (4). His symbolism thus presupposes the bivalence he denies with his words. If, as in von Wright's truth-logic, Raz would distinguish between 'It is not true that .. .' and 'It is false that .. .', his conclusions would have to be different. Raz holds that

(1) 'LR~, t/J' is neither true nor false and 'LPer~, -,t/J' (i. e. 'L-.R~, t/J') is neither true nor false

can be true. Using the truth-logic TL, we can refute that assertion. Let us denote 'It is not true that .. .' by -,T, and 'It is not false that .. .' by -,T-,. We can then represent (1) through expressions (13) and (14):

(13) -,T(LR~, t/J) /\ -,T-, (LR~, t/J) (14) -,T(L-.R~, t/J) /\ -,T-, (L-,R~, t/J)

It can be shown that the acceptance of the logical law (4) - which I will call 'Raz's law' -is incompatible with (13) and (14) in TL.

First, the proof for (13):

1) -,T(LR~, t/J) A -,T-, (LR~, t/J) 2) -,(LR~, t/J) ~ (L-,R~, t/J) 3) -,(LR~, t/J) ~ (L-,R~, t/J) 4) (LR~, t/J) v (L-,R~, t/J)

Definition of gap Raz's law Elimination of ~ in 2) Definition of ~ through v in 3)

A LOGICAL ANALYSIS OF LEGAL PROPOSITIONS 71

5) T(LR~, tfJ) v T(L-,R~, tfJ) Axiom AO of TL33 in 4) 6) T(LR~, tfJ) v T~(LR~, tfJ) Substitution in 5)34

7) ~(T(LR~, tfJ) v T~(LR~, tfJ) Morgan's Law in 1) 8) T(LR~, tfJ) v T~(LR~, tfJ) A ~(T(LR~, tfJ) v T-,(LR~, tfJ)

Since 8) is a contradiction (resulting from the conjunction of 6) and 7», we have shown that (13) is incompatible with Raz's law.

Now the proof for (14):

1) ~T(L~~, tfJ) A ~T~ (L~~, tfJ) Definition of gap 2) ~(L~~, tfJ) ~ (LR~, tfJ) Raz's 1aw35

3) ~(L~~, tfJ) ~ (LR~, tfJ) Elimination of ~ in 2) 4) (L~~, tfJ) v (LR~, tfJ) Definition of ~ through v in 3) 5) T(L~~, tfJ) v T(LR~, tfJ) Axiom AO of TL in 4) 6) T(L~~, tfJ) v T~(L-,R~, tfJ) Substitution in 5)36

7) ~(T(L~~, tfJ) v T~(L~~, tfJ) Morgan's Law in 1) 8) T(L~~, tfJ) v T~L~~, tfJ) A ~(T(L~~, tfJ) v T~(L~~, tfJ)

Since 8) is a contradiction - resulting from the conjunction of 6) and 7) -, we have shown that (14) too is incompatible with Raz's law.

That means that Raz cannot maintain the truth of (1) and the analytic truth of the logical law contained in (4). Raz's law implies that there can be no legal propositions without a truth-value. Since Raz holds that legal propositions referring to borderline cases and to unresolved conflicts have no truth-value, he should abandon the claim that (4) is an analytic truth. And he should distinguish the external negation from the inter­nal negation of legal propositions of reasons, just like truth-logic distinguishes 'It is not true that .. .' from 'It is false that .. .'. 'It is not true that .. .' corresponds to his external negation

(15) ~(LR~, tfJ)

and 'It is false that .. .' to his internal negation

(16) (L~~, tfJ).

33 This is the axiom according to which all tautologies remain tautologies when their variables are prefixed with T.

34 Substitution of the second member of the disjunction, given the equivalence of 2).

35 2) is derived from Raz's law by contraposition.

36 Substitution of the second member of the disjunction, given the equivalence of 2).

72 CHAPTER II

Thus, to say that 'LR~, t/J' has no truth-value would mean that it is neither true nor false, that is:

(17) -,(LR~, t/J) A -,(L-,R~, t/J).

But (17) is equivalent to (3). That means that gaps are possible, and it is the logical truth contained in (4) that should be abandoned, because it is a logical truth only in a logical system that presupposes bivalence. In Raz's case, he thus smuggles back in the bi­valence he had denied before.

Raz's thesis that there are no normative gaps in legal systems (and that (4) is analytically true) relies on two assumptions that should be made explicit:

(I) The logic of conclusive permissions according to which that an action is con­clusively permitted means that there is no conclusive reason prohibiting it. (II) The bivalent foundation of Raz's symbolism which renders any formulation of a proposition without a truth-value inconsistent.

Raz thus confronts a dilemma: Either he accepts the logic of conclusive reasons, and re­jects the possibility of legal propositions without a truth-value; or he rejects the logic of conclusive reasons and, therefore, also the logical truth of (4).

The first horn of the dilemma leads to the rejection of legal constructivism and, therefore, of legal positivism. Only a position that asserts the existence of a normative reality independently of social facts can make such a position plausible. What is needed in order to opt for the first horn of the dilemma is an ontological, not a logical thesis.

Now let's turn to the second horn: Once it is guaranteed that the weak version of the Principle of Prohibition (which is a version of Raz's law) is analytic, but irrelevant and, above all, that it does not help to maintain the existence of gaps, one can say that the thesis best fitting to legal positivism is the one asserting that completeness (and, therefore, also the existence of gaps) is a contingent matter.

In summary, like Kelsen and Dworkin, Raz too has not shown that it is a 'logical truth' that legal systems have no normative gaps; the existence of a closure rule like the one expressed in the strong version of the Principle of Prohibition is a contingent ques­tion. Thus, contrary to Raz's belief, there can be gaps when the law is silent.

7. Legal Statements and Vagueness: The Law Speaks With an Uncertain Voice

Even more than with normative antinomies or gaps, the indeterminacy of the law is connected with the problem of the vagueness of linguistic expressions, and especially the vagueness of the generic terms (or, more precisely, of the concepts expressed by the generic terms) used in the formulation of general rules.37 Thus, we can say that

37 As is commonly known, an important part of the controversy in legal philosophy provoked by the work of Hart is precisely about this question. Cf. for example Fuller 1958, Dworkin 1985, Lyons 1982. The dis­cussion of Hart's thesis of the indeterminacy of the law because of the open texture of language has recently

A LOGICAL ANALYSIS OF LEGAL PROPOSITIONS 73

A linguistic expression E expresses a vague concept if and only if in some cases it poses the problem whe­ther or not a particular object belongs to the referent of E.

Generic terms like 'woods', 'bald', or 'heap'38 pose a problem when it comes to deter­mining their referent in many doubtful cases. In the law, there are terms like 'at night' (what about dusk?) or 'habitually' (how many times must a behaviour have been re­peated in order to be a habit?) which suffer from vagueness in that sense. In fact, the problem is even more severe since, as has repeatedly been shown (Russell 1923, Wil­liams 1946, 181 ff.; Waismann 1951; Wittgenstein 1953, sects. 76 and 80; Carnap 1955; Ross 1958, 114 f.; Hart 1961, 121 ff.; Carri6 1965,31 ff.; Scheffler 1979,50-65), all generic terms are potentially vague. This is what is known as the open texture of lan­guage: one can always imagine an object for which we have no clear criteria for know­ing whether or not it belongs to the referent of the respective expression.39 The same is true for artificial languages. Generally, we have no great difficulties in applying the term 'cat' (when referring to a certain mammal); but would we keep using the word 'cat' for referring to an animal with the shape of a cat, but which walks upright, wears a grey suit, and says 'Nice to meet you'? Similarly, since the law uses rules which contain general terms that may be applied to a wide variety of cases, there are many cases that are doubtful or difficult because of the vagueness of the terms used in the formulation of the rules. Thus, in order to know whether an aggravating circumstance applies in a criminal case, one can ask: Is an individual carrying a hypodermic syringe armed? Or, in order to know whether an unwed mother who kills her baby to hide her disgrace has committed infanticide: How long is a child newborn? Or how valuable must a hidden and unknown object be in order to be considered a 'treasure' to the effects of civil law?

Thus, in the famous words of H. L. A. Hart:

"All rules involve recognizing or classifying particular cases as instances of general terms. and in the case of everything which we are prepared to call a rule it is possible to distinguish clear central cases. where it certainly applies and others where there are reasons for both asserting and denying that it applies. Nothing can eliminate this duality of a core of certainty and a penumbra of doubt when we are engaged in bringing particular situations under general rules." (Hart 1961.119)

What happens, then, with the truth-conditions of legal propositions when norm formula­tions are expressed using vague concepts? If a norm requires that all F do q" and x is a

returned to the center oflegal-philosophical attention: cr. Schauer 1991. 34-37; Marmor 1992. 132-134; Bix 1993.7-35; Waldron 1994.509-540.

38 The latter two examples were already mentioned by the Stoics; thus. Eubulides said: "Would you say that a man was bald if he had only one hair? Yes. Would you say that a man was bald if he had only two hairs? Yes. Would you .... etc. Then where do you draw the line?" (quoted from Kneale/Kneale 1962. 114).

39 Potential vagueness is also known as intensional vagueness (the terminology is due to Camap 1955); we have intensional vagueness when it is possible that there are objects which give rise to doubts about whether or not they belong to the referent of the respective predicate. If such objects do. in fact. exist. the vagueness is also extensional. If a predicate is extensionally vague. it also is intensionally vague. but the converse does not necessarily hold.

74 CHAPTER II

borderline case of F, what are the truth-conditions of the proposition expressed by 'Le­gally, x ought to do iP'?

Actually, the question is even more pressing if we take into account that vague predicates give rise to the kind of paradoxical arguments known since the times of anti­quity by the term of sorites ('soros' in Greek meaning 'heap'). Recent philosophical discussion on vagueness takes the problem posed by the paradoxes arising from argu­ments of this kind very seriously.40

Let us look at some examples: Imagine a heap of sand. Now, if we take one grain from that heap, it will still be

a heap of sand: taking one grain from it does not convert a heap into something else. Thus, if two sets of grains of sand differ only by one grain, either both are heaps, or none is. These apparently obvious assertions lead to the conclusion that all sets of grains of sand, even those consisting only of a single grain, are heaps.

Of someone who has ten million dollars, we would say that he is rich. If some­one is rich, he will still be rich if we take one dollar from him. Therefore, we are all rich (even if we have only one dollar, or no dollar at all).

Suppose there is a red spot. If a coloured spot cannot be distinguished from a red spot, it is red. Now, between two spots of whatever colour that can just barely be dis­tinguished, there is always a third possible spot whose colour is indistinguishable from that of the others. Between a red spot and an orange spot (or a yellow, or green, or blue one, ... ), there is a sequence of barely distinguishable spots. Therefore, all coloured spots are red.

o is a small number. If n is small, then n+ 1 is also small. Therefore, all numbers are small. This version of the paradox is known as 'Wang's paradox' (Dummett 1978, 250).

If someone is put into solitary confinement for a thousand days, we would call that humiliating treatment. If a certain period of solitary confinement is humiliating, then solitary confinement of one hour less is also humiliating. It follows that all solitary confinement (for instance, one of only a few minutes) is humiliating.41

What these arguments have in common is the presence of vague concepts ex­pressed through terms like 'heap', 'rich', 'red', 'small', 'humiliating treatment', etc. So­rites arguments depend on the fact that these expressions are tolerant, i. e., small chan­ges do not affect the applicability of the concept. The paradox arises because big chan­ges, which do affect the applicability of the concept, can be constructed - because of the tolerance of those concepts - as a succession of very small changes.

More formally, the argument can be presented as follows:

40 I here follow Sainsbury's illuminating discussion of the problem (1995, 22-51).

41 Note that what is paradoxical here is the step from the premises to the conclusion (this is valid even if one thinks that some solitary confinement is not humiliating). See the example in Endicott 1997. The Spanish Constitution, in its art. 15, prohibits inhuman or humiliating punishment and treatment; and the Prison Re­gulations (Royal Decree 1201 of May 8, 1981, modified by Royal Decree 787 of March 28, 1984) in their art. 111 permit the sanction of solitary confinement, though only for a maximum of 14 days.

A LOGICAL ANALYSIS OF LEGAL PROPOSITIONS 75

(1) A person who has 10 million dollars is rich. (2) If someone with n dollars is rich, then someone with n-l dollars is rich too.

Therefore, a person who has 1 dollar is rich.

Premise (1) of the argument can be called the categorical premise; premise (2) actually stands for a whole number of conditional premises which permit the application of the corresponding number of applications of modus ponens, until we reach a paradoxical conclusion. To such an argument, one can react in one of three possible ways:

(a) Accept the conclusion. (b) Reject the argument as unsound. ( c) Reject one or more premises of the argument.

I will now consider each one of these possible reactions, and their significance for the truth-conditions of legal propositions containing vague concepts.

a) Accepting the conclusion of sorites

To accept the conclusion of sorites seems contrary to common sense. But it has been used as a challenge to our ordinary conception of the world.

Unger (1979a, 1979b) has presented the following sorites argument:

(1) You cannot make a table out of just one gramme of wood. (2) If you cannot make a table out of n grammes of wood, then you cannot make a table out of n+ 1 grammes.

Hence you cannot make a table (i. e, there are no tables).

This argument poses a challenge to our conception of the world. Our vague concepts are deeply deficient, to the extent that they undermine our firmest intuitions and lead us to assert absurdities (that a grain of sand is a heap of sand, or that someone with only one dollar is rich). In Unger's words,

"we may begin by supposing that there are heaps, and that a million beans typically arranged gives us an instance of that concept. But, then, removing a single peripheral bean gently from such a typical heap, it seems, will not leave us with no heap before us. Hence, we must conclude that even when we have but one bean left, or none at all, we still have a heap of beans. But this is absurd. Hence, we have reduced the ori­ginal supposition of existence to an absurdity, and we may generalize accordingly. This, we may say, is an indirect argument, that there are no heaps" (Unger 1979b, 118).

This nihilist and skeptical conclusion challenges our ordinary conception of the world, suggesting a scientific conception for which there are no heaps, no table, no rocks, and not even human beings (in 1979b, Unger reaches the conclusion that he does not exist).

Dummett (1978, 248-268) and Wright (1975; 1976; 1993, 107-174) too have taken the challenge posed by this paradox seriously, and have come to conclusions lead-

76 CHAPTERll

ing to some form of nihilism. For these authors, the problem does not arise from the conflict between our ordinary conception of the world and the scientific conception, but from the incoherence of our language or, more precisely, the incoherence of the rules governing our use of language. According to Wright, the incoherence arises out of two incompatible theses about the presence of observational predicates in our languages: (i) It seems to be incompatible with their meanings to draw precise limits for their applica­tion; and (ii) if we do not draw precise limits for their application, predicates are tole­rant and can be applied indiscriminately. Wright suggests that the point of view accord­ing to which our language is rule-governed should be abandoned.42

In fact, Unger's conception as well as those of Dummett and Wright restore Fre­ge's skepticism about the possibility of vague languages. As Frege (1970, 159) said: "A concept that is not sharply defined is wrongly termed a concept. "43 He gave two rea­sons for his rejection of quasi-conceptual constructions: (i) they lack meaning (Bedeu­tung), and (ii) the laws of logic fail when applied to them. For both reasons, and be­cause they lead to incoherence, Frege regarded languages that use vague concepts as deficient.

The consequences of accepting that nihilist conception of ordinary language are devastating for legal theory. Since legal norms often use observational predicates, the introduction of sorites makes our legal language incoherent. Take the following exam­ple: According to some Criminal Codes, to commit a crime at night is an aggravating circumstance. With this example, we can produce a sorites paradox:

(1 ') On some territory, say, in Spain, it is night at two in the morning. (2') If at h o'clock it is night, then it is also night at h o'clock minus one second.

Therefore, at two o'clock in the afternoon it is night.

Thus, legal propositions expressed by applicative legal statements (containing observa­tional predicates) seem to be incoherent, and the legal propositions referring to them lead to inconsistent propositions.

42 Burns (1991, 128 f.) proposes to substitute Wright's strict rules of tolerance by other, broader ones, in order to avoid the incoherence those rules lead to when applied to observational predicates. A strict rule of tolerance could have the following fonn: "If one thing is a heap and a second differs from it in containing only one less grain, then the second is a heap also." Now, Bums proposes the following fonnulation for rules of tolerance: "If one thing is a heap and a second differs from it in containing only one less grain and in any other ways dependent on this minor difference, but the two do not differ detectably in any other respects re­levant to the application of the predicate 'heap'. then the second is a heap also." The possibility that this ceteris paribus clause in rules of tolerance helps avoid sorites paradoxes will not be analyzed here. A some­what different critique of Wright's ideas can be found in Platts 1979,217-249.

43 And he added (ibid.): "A definition of a concept (of a possible predicate) must be complete; it must un­ambiguously determine, as regards any object, whether or not it falls under the concept (whether or not the predicate is truly assertible of it) ... We may express this metaphorically as follows: the concept must have a sharp boundary."

A LOGICAL ANALYSIS OF LEGAL PROPOSITIONS 77

With sorites, we seem to reach conclusions like those advocated by the authors of Critical Legal Studies, with their insistence on the radical indeterminacy of the law and the incoherence of legal reasoning. In one of the papers most frequently quoted in critical scholarship, Singer (1984) attempts to give a foundation to nihilism in legal theory, a nihilism with two components: an epistemological and a moral one. Here, I am interested in underscoring the epistemological component which Singer elaborates as follows:

"As a theory of knowledge, nihilism claims that it is impossible to say anything true about the world. No one can properly claim to describe the world accurately ... If one takes nihilism seriously, it is impossible, or in any event fruitless, to describe the world ... " (Singer 1984, 4)

The acceptance of sorites arguments gives plausibility to this radically skeptical thesis. However, that position clashes with two obvious truths about ordinary language:

(i) We do use vague predicates in our ordinary (and in our legal) language. And (ii) our ordinary (and our legal) language permit sufficient understanding in human communi­cation.

Of course, we could be under a total illusion; but before we accept such a disap­pointing conclusion we should analyze other ways of escaping the sorites paradox. Since accepting that conclusion is counterintuitive, we should accept it only if there is no other way of detaining the subversive force of the paradox.44 So let us now look at such alternative solutions.

b) Rejecting the argument: degrees of truth

Although we may hardly be inclined to question the rule of modus ponens, some theo­rists have made it responsible for the problem of sorites and the paradoxes it produces.

It may seem natural to say that the proposition according to which someone with a certain amount of money is rich is true to some extent, that there is a certain amount of truth in it. This has led some authors to construct a logic of vagueness, so-called fuzzy logic.45 The semantic notions of truth and falsity are replaced by the notion of de­grees of truth. If the predicate 'rich' definitely applies to some person, then the proposi­tion attributing richness to that person would be said to have the highest possible truth­value (i. e., 1). If the predicate definitely does not apply to some other person, then the proposition attributing richness to that person would be given the smallest possible truth-value (i. e., 0). Between 0 and 1, there is a continuum on which the different ob-

44 Unger's position could also be challenged as absurd in the following way: From the premise 'With 100 kg of wood one can make a table' we can conclude, by way of a sorites argument, that with I gramme of wood one can make a table; now, from the premise 'With I gramme of wood, one cannot make a table' we conclude, also by way of a sorites argument, that with 100 kg of wood one cannot make a table. Unger could argue in favour of the falsity of the categorical premises of any sorites argument; but that would mean to reject the premises (at least sometimes), rather than accept the conclusion of such arguments; and the re­jection of the truth of the premises would have to be argued for independently of the argument.

45 Cf. Zadeh 1965. For a recent systematic presentation of fuzzy logic(s), see Trillas/Alsinafferricabras 1995.

78 CHAPTER II

jects (in our case, persons) being borderline cases of the predicate are located. This gives us an idea of how degrees of truth are assigned to atomic propositions. But we also need a method for assigning degrees of truth to molecular propositions. Disjunction and conjunction can be composed in a fashion similar to that of classical bivalent logic. With a metaphor owed to Quine (1972, 30), we can say that while truth is dominant (and falsity recessive) in disjunction, truth is recessive (and falsity dominant) in con­junction. Thus, we have:

(a) Val [p v q] = Max {Val [P], Val [q]}

(b) Val [p /\ q] = Min {Val [P], Val [q]}

Negation also behaves as in classical logic, i. e., the value of the negation of a proposi­tion p is the complement of the degree of truth (in the set {O, I}) of p:

(c) Val [-,p] = 1 - Val [Pl.

As in classical logic, conjunction and disjunction are interdefinable with the help of (a), (b) and (c). The conditional, however, behaves differently. The idea is that if the ante­cedent of a conditional is more true than its consequent, then the conditional cannot be completely true, and if the antecedent is only marginally more true than the consequent, then the conditional must be almost totally true. The formula for the degree of truth of a conditional can be the following:

(d) Val [p --+ q] = 1 - (Val [P] - Val [q]), if Val [P] > Val [q];

= 1 in all other cases

This approach enables us to avoid the paradox by postulating that the validity of modus ponens presupposes that the propositions it is applied to have the extreme degrees of truth, 1 or O. That means that one cannot reach conclusions of a degree of less than 1 from premises of degree 1. In the intermediate degrees, however, the application of mo­dus ponens can bring with it the filtration of truth. The filtration can be very small for each application, but it will be big if the number of applications is big, as in the case of sorites; and that is enough to render modus ponens invalid (Sainsbury 1995, 41 f.; En­gel 1991, 208 f.).

Here, I will not analyze in detail the advantages and disadvantages the degrees­of-truth approach may have for the analysis of propositions that contain vague concepts, since that approach would lead to severe problems for my general analysis of legal pro­positions.46 If I were to analyze legal propositions containing vague predicates using the conception of degrees of truth, I would have to say that there are true legal proposi­tions, false legal propositions, legal propositions without a truth-value, but also legal propositions that are absolutely true, true to a high degree, more or less true, almost

46 An analysis of legal propositions and of legal reasoning in general with the help of fuzzy logic can be found in Mazzarese 1996, 1997.

A LOGICAL ANALYSIS OF LEGAL PROPOSITIONS 79

true, almost untrue, absolutely untrue, etc. The truth-logic TL would be an insufficient instrument for analyzing those last propositions. But in the case of normative gaps, for instance, the problem is not that legal propositions are more or less true, but that they have no truth-value at all.

Besides, the logic of degrees of truth is itself insufficient for accounting for pro­positions without a truth-value, since those propositions fall outside the range of truth­values between I and O. Therefore, I will try to show how the truth-logic TL can ac­count for propositions containing vague concepts.47

c) Rejecting the premises

Before analyzing how TL can handle vague predicates, I must present - if only very briefly - the so-called epistemic theory of vagueness. It holds that vagueness consists in our ignorance of the precise boundaries of our concepts and that, therefore, it does not call for a revision of classical logic. Although this theory is accepted only by a small minority of philosophers, it has lately received a great amount of attention.48

The epistemic theory regards the paradoxes created by sorites as arguments that prove its own conception. Since the conclusion of a sorites argument is false, the episte­mic theory argues, at least one of the premises must be false - a kind of reductio ad absurdum. Thus, it is argued that a sharp boundary divides heaps from not-heaps and the rich from the not-rich, and also that there is at least one number n for which it is not true that n grains are a heap, or that a person with n dollars is rich. Hence, for some values of n it is false that

If n grains of sand are a heap, n-l grains also are a heap.

That means that vagueness arises out of our ignorance about borderline cases. But why are we so inevitably ignorant about such cases? The epistemic theory's answer is as fol­lows (Williamson 1994, 185-243): Our cognitive mechanisms, i. e. our senses, neces­sarily come with a margin of error. Suppose we are looking at a spectrum of coloured spots starting with red and gradually turning into orange, yellow, and so on. Suppose also that we believe a certain spot s to be the last red spot of the spectrum. According to the epistemic theory, that belief cannot count as knowledge, because the truth of our belief to a certain degree is a matter of luck, since we cannot really distinguish that spot

47 I do wish to point out at least one problem with the logics of degrees of truth. Take the following legal statement: 'Legally, x ought to convict y of murder at night'. Let us assume that the degree of truth of 'x committed murder' is 0.7 (we are not absolutely sure whether it was an act of intentional killing) and that of 'x acted at night' is 0.2 (though it was beginning to get dark, there was still a lot of light). The degree of truth of the statement 'x committed murder, and he did it at night' is then 0.2, since the conjunction, so to speak, follows the worse of the two parts. Now, the degree of truth of 'x did not commit murder, and he did it at night' also is 0.2 (since 0.2 is lower than the 0.3 degree of truth of 'x dit not commit murder'). But that both statements should have the same degree of truth is highly counterintuitive and would lead to a rather extravagant analysis of the degrees of truth of legal propositions.

48 The most complete defence of that theory formulated in recent years is that of Williamson 1994.

so CHAPTER II

from its neighbours in the spectrum. But knowledge is justified true belief and thus requires that truth is not encountered accidentally. Therefore, the fact that there is a margin of error in our perceptive apparatus means that we will never know which one is the last red spot in the spectrum.

What the epistemic theory does not tell us is what we would need to know in order to have a justified true belief in borderline cases. It presupposes, rather than proves that our concepts have sharp boundaries.

Advocates of the epistemic theory often argue with some kind of petitio princi­pii. Thus, for instance, Horwich's argument (Horwich 1990, SO-S7; cf. also Williamson 1994, IS7-190) for rejecting sorites, based on the epistemic theory, depends on his un­questioned acceptance of bivalence. The argument runs as follows (Horwich 1990, SO):

"How can truth-value gaps be admitted? They can't be. Given any logic that licenses the principle of con­traposition:

(9) (a is F ~ a is G) ~ (a is not G ~ a is not F),

we can go from the minimal theory of falsity (10) <p> is false H <p> is not true to (11) <p> is not false H <p> is not not true.

Therefore (12) <p> is not true and not false ~ <p> is not true 1\ <p> is not not true."

Horwich concludes: "Thus we cannot claim of a proposition that it has no truth-value, for that would imply a contradiction." Now, let us take a closer look at this argument. (10) and (11) are formulations of bivalence, according to which a proposition is either true or false. That means that there are no propositions without a truth-value. However, the antecedent of (12) implies that such propositions do exist, and thus, by way of (10) and (11), a contradiction is produced. But that contradiction does not arise because of an intrinsic inconsistence of the rejection of bivalence, but because the acceptance of bi­valence is inconsistent with the existence of propositions without a truth-value. We could also accept the antecedent of (12) and reject the bivalence presupposed by (10) and (11). And that is precisely what TL does. In the notation of TL, (10) would be equi­valent to

(10') T-,p H ...,Tp

and (11) to

(11 ') ...,T-,p H -,-,Tp.

(10') is not universally true in TL, since it may be the case that it is not true that p and it is not false that p. That is, the biconditional of (to') is valid from left to right, but is not

A LOGICAL ANALYSIS OF LEGAL PROPOSITIONS 81

universally valid from right to left: '-,Tp' is compatible with '-,T -,p'. Inversely, (11 ') is valid from right to left, but not from left to right.49

Hence, Horwich's argument does not prove that the rejection of bivalence is in­adequate because it produces a contradiction; he only proves that the acceptance of bi­valence in (10) and (11) is inconsistent with its rejection, i. e., the antecedent of (12).

Now, for the epistemic theory of vagueness it is not enough to show that if one assumes bivalence, then it is inconsistent to hold that propositions about borderline cases containing vague concepts have no truth-value; it must also - independently -show that the rejection of bivalence is inconsistent, or at least inadequate.

The epistemic theory of vagueness presupposes a realist conception of proposi­tions with vague concepts: realist, that is, from the metaphysical point of view - since it accepts that, independently of our knowledge, there is something in the world that can make our propositions about borderline cases true or false; from the semantic point of view - since it accepts that the meaning of propositions about borderline cases is de­termined by their truth-conditions, independently of whether or not we are able to find out what they are (in fact, it maintains that we are not); and from the logical point of view - propositions containing vague concepts behave according to bivalence: they are all either true or false.

It should be noted that the epistemic theory is compatible with my analysis of the truth-conditions of legal propositions. One can hold that there are pure legal proposi­tions without a truth-value - in the case of an antinomy or a legal gap -, whereas all propositions of fact, which are an element of all applicative legal propositions, are bi­valent, i. e., are always true or false.

However, I will hold that the source of vagueness is not epistemic, but seman­tic.50 In von Wright's words (1984c, 39):

"A great many concepts used in discourse about contingent matters of experience are not sharply bounded but have a 'fringe of vagueness' ... which accounts for the existence of 'borderline cases' of which it is hard to tell whether they fall under the concept or not. 'Hard to tell' does not here point to limitations of our epistemic faculties of ascertaining and observing things. The phrase refers to the absence of criteria for applying the concepts to the case at hand."

In such instances of a proposition containing vague concepts and referring to cases that fall into the zone of penumbra of such a concept, I will say that the proposition has no truth-value.

We may even want to say that in the zone of penumbra of some concept F, in a certain sense x is F, and in some other sense x is not F. Von Wright (1984c, 38) distin-

49 The acceptance of propositions without a truth-value ("Tp " ,Top', in terms of TL) is inconsistent­as I have shown in 6.b - with the law of bivalence ('Tp v T,p')and, therefore, that law is not valid in TL.

50 It has also been suggested (Lewis 1983, 228 f., together with Lewis 1969, ch. 5) that the source of vague­ness is not semantic, but pragmatic. The idea is this: All languages are precise (in Frege's sense), i. e., they contain no vague concepts; but people use several languages, rather than a single one. The linguistic conven­tions of a population do not select one particular point, but a fuzzy region in the space of precise languages. Our language is only a "hybrid resonance" (Lewis 1969,201) of the possible languages that shape it.

82 CHAPTER II

guishes two senses of truth: truth in the broad sense (T), and truth in the strict sense (T, which behaves just as in TL). It is true in the broad sense that x is F if it is not false that x is F. Thus, in the broad sense it may be the case that it is true that x is F and that it is false that x is F, since 'true in the broad sense' is equivalent to 'not false in the strict sense', i. e.,

and 'false in the broad sense' is equivalent to 'not true in the strict sense':

Top = -,Tp.

Thus, to say that it is true in the broad sense that p and that it is false in the broad sense that p is the same as saying that in the strict sense it is not true that p and it is not false that p, i. e., that p has no truth-value.

Some of the intermediate conditional premises of sorites arguments, thus, have no truth-value and cannot filtrate truth to the conclusions, which are clearly paradoxical. But to say, for example, that there are values of n for which

If n grains of sand are a heap, n-l grains also are a heap

has no truth-value seems to presuppose that we can precisely divide groups of grains of sand into heaps, not-heaps, and an intermediate zone of penumbra in which the attribu­tion of the predicate 'heap' to a particular object generates a proposition without a truth­value. In fact, we may then not only have doubts about whether h is a heap, but also about whether h is a borderline case of the borderline cases of application of the predi­cate 'heap'. It may be that not only 'heap', but also 'definitely a heap' is vague. This is known as higher-order vagueness - since just as one can generate second-order vague­ness (,definitely a heap'), one can also generate third-order vagueness ('definitely defi­nitely a heap'), and so on.51

In a critique of Dworkin (l977b),52 Raz (1979, 73) insists that a conception of vagueness that does not take higher-order vagueness into account is inadequate:

"This suggestion rests, however, on a fallacious view of vagueness. It assumes that whereas a term which is not vague divides all cases into those to which it applies and those to which it does not, a vague term divides all cases into three sets: those to which it applies 'by its indisputable core of meaning', those to which it clearly does not apply, and those in between. It is as if a term is vague because it draws two sharp dividing lines instead of one. The truth is that all, and not only some, nouns, verbs, adverbs, and adjectives of a natural language are vague. And though a vague term clearly applies to some cases, clearly fails to apply to some and doubtfully applies to others, yet it is often impossible to draw general boundary lines between the

51 For a discussion of higher-order vagueness, cf. the recent works of Sainsbury 1991, Engel 1992, Wright 1992a, Heck 1993, Edgington 1993, Hyde 1994, and Tye 1994. I will not pursue this discussion here.

52 This is a first version of his 1985, 119-145. See Dworkin's reply in the later version, 1985, 130 f. and 405, n. 3.

A LOGICAL ANALYSIS OF LEGAL PROPOSITIONS 83

three categories. It is a test of adequacy of any account of vagueness that it recognizes as a central type of it the cases where vagueness is ·continuous'."

Although it does not yet solve the problem of higher-order vagueness, one way of draw­ing up such a tripartite division about vague concepts - a positive extension, a nega­tive extension, and a zone of penumbra - can be found in the so-called theory of superevaluations.53

That theory assumes that our perplexity about vague predicates originates not in "the realm of how things are with the object; rather ... in the realm of how we choose to speak of the objects" (Sainsbury 1995, 34). When we are uncertain whether seventy grains of sand are a heap, then we also tend to admit the principle of tolerance (the con­ditional sorites premise); but we do not have to do this, once we realize that in the pen­umbra of 'heap' we can count those seventy grains under the heaps or the not-heaps ad libitum. Any choice is a precision of the concept of a heap and is admissible as long as it draws the line in the penumbra of the concept. Thus, many precisions are admissible. In an application of this general idea, we can say that

A proposition p - containing a vague concept - is true if and only if it is true for all its precisions; it is false if and only if it is false for all its precisions; and it has no truth-value in all other cases.

A precision is a way of converting a vague concept into a precise one. So now we must distinguish two senses of 'true': 'true' according to a particular precision, and 'true' ac­cording to all precisions, or supertrue. If a number x of grains of sand is in the penum­bra of the concept of a heap, then it will be true for some precisions and false for others that x is a heap and, therefore, it will neither be supertrue nor superfalse.

The paradox has thus been solved: the principle of tolerance is false, since it is not true that

For all n, if n grains of sand are a heap, n-l grains are also a heap. 54

Since that proposition is false in all precisions, it is, we can say, superfalse. Thus, the second sorites premise is false, and a sorites argument is unable to filtrate truth to the conclusion.

53 The theory was first formulated by Mehlberg (1958) and van Fraassen (1966) in the context of philoso­phy of science. The application of the theory to the problem of vagueness is due to Fine (1975), Kamp (1975), Dummett (1978,340-342), Lewis (1983). Cf. also Williamson 1994, ch. 5.

54 Putnam (1983, 285 f.) has argued, in a slightly different way, that with an intuitionist logic one can ac­cept the truth of

It is not the case that for all n, if n grains of sand are a heap, then n-l are a heap too

without being committed to accept that

There is an n such that if n grains of sand are a heap, then n-l are a heap too.

Here, the law, valid in classical logic, that enables one to go from ·...,'"i/x (Fx)' to '3x (-,Fx), is rejected.

84 CHAPTER II

Similarly, in a superevaluationist conception the law of excluded middle is re­tained: 'h is a heap or it is not a heap' is supertrue, because it is true for all precisions, although 'h is a heap' may have no truth-value.

The theory of superevaluations retains a great part of classical logic. Thus, for instance, all tautologies of classical logic are supertrue in a superevaluationist theory. However, it does not retain the classical assignation of truth to molecular propositions. A conjunction or disjunction of propositions without a truth-value can be supertrue, or superfalse, or lack truth-value. Thus, if h is a collection of grains in the penumbra of 'heap', then even though 'h is a heap' and 'h is not a heap' have no truth-value, 'h is a heap and h is not a heap' is superfalse (false in all admissible precisions) and 'h is a heap or h is not a heap' is supertrue (true in all admissible precisions); but 'h is a heap or h is a heap' has no truth-value (being equivalent to 'h is a heap', which is neither supertrue nor superfalse).

This characteristic of superevaluations unfortunately makes them incompatible with our truth-logic TL, since T ('it is true that') has characteristics that are incompati­ble with supertrue. Thus, for the theory of superevaluations, the law of excluded middle is supertrue. In TL, that would be equivalent to accepting

(1) T(pv-,p),

which (by axiom A3 of TL) implies

(2) Tp v T-,p,

and (2) is incompatible in TL with accepting propositions without a truth-value:

(3) -,Tp /\ -,T-,p.55

Perhaps one could try to construct a superevaluationist approach for all legal proposi­tions (not only for those containing vague concepts). The idea is the following: One could regard as precisions of legal propositions all those possible worlds that could make effective the norms of the legal system the legal propositions refer to. In this way, a pure legal proposition is true if the norm it refers to is effective in all admissible possible worlds; it is false if the norm is ineffective in all admissible possible worlds, and it has no truth-value in all other cases.56

55 Superevaluationist conceptions, however, do not share this characteristic of TL; for them, the law of ex­cluded middle does not imply bivalence; cf. on this Day 1992.

56 In the case of applicative legal propositions, the definition would become more complicated because one would have to look at the supertruth of propositions of fact with the superevaluationist theory of vague pro­positions. A proposition of fact is supertrue if, and only if, it is true in all admissible precisions of all admis­sible possible worlds, superfalse if, and only if, it is false in all admissible precisions of all admissible possi­ble worlds, and neither supertrue nor superfalse in all other cases.

A LOGICAL ANALYSIS OF LEGAL PROPOSITIONS 85

In the case of normative gaps, there are several ways of completing the norma­tive system. Suppose there is a case C that has no normative solution in a system of norms S. We can then complete S by assigning to case C, e. g., the solution Php or the solution Pp. Now, in the admissible possible worlds in which S is effective with one completion, it is ineffective with the other. Therefore, legal propositions referring to C are neither true nor false.

In the case of antinomies or normative inconsistencies, one could regard as ad­missible possible worlds all those that manage to eliminate at least one of the inconsis­tent norms.57 Since there is a plurality of possible worlds satisfying this requirement, legal propositions referring to inconsistent systems also have no truth-value.

However, to accept a superevaluationist approach to legal propositions would mean to give up TL, which expresses my own approach to the problem. Therefore, the possibility is only mentioned here.

Now, the theory of superevaluations as well as TL have problems with higher­order vagueness. In the case of superevaluations, this is because it is not accepted that the notion of admissible precision can be regarded as vague too (Williamson 1994, 156-161; Sainsbury 1995, 38 f.). In the case of TL, the reason is that, although p (which con­tains vague concepts) may lack a truth-value, Tp always has one. The predicate 'it is true that' of TL is not vague. And, again, sorites can threaten the language in which our semantic is expressed. There seem to be only three options:

(i) to accept sorites and to hold either that our language is inconsistent or that the world of our ordinary life does not exist;

(ii) to accept the epistemic theory, which - in fact - is hard to refute, although it seems to be on the wrong track; or

(iii) to accept the semantic theory, in the hope that some reconstruction of super­evaluationism or TL will be able to meet the challenge of higher-order vagueness.

The semantic theory can confront the challenge either by showing its irrelevance or by constructing a semantic that can be expressed in a vague language which, nevertheless, does not allow the paradox to arise; for some propositions, such a semantic would nei­ther imply that they are true, nor that they are false, nor that they are neither true nor false (Tye 1994; Sainsbury 1995,46).

One way of trying to show the irrelevance of higher-order vagueness could be grounded on the fact that if that vagueness depends on the possibility that one is uncer­tain about the classifications one has performed, then it may not be enough to stop at vagueness of level 1. We may, for instance, have classified the expression 'x is a heap' as definitely true ('Tp', in terms of TL). We then begin doubting whether, definitely, it is definitely true that x is a heap. Thus, perhaps we should revise our assignation and re­gard x as a borderline case of a heap, which would mean that in TL we now have ',Tp'. But the possibility of revising our assignations does not force us to accept higher-order

57 One could consider it a criterion of adequacy to eliminate the least possible number of norms in order to keep the system consistent.

86 CHAPTER II

vagueness. We only need to accept three things: definitely true ('Tp', in TL); definitely false ('T -,p', in TL); and neither definitely true nor definitely false ('...,Tp /\ ...,T -,p', in TL). Our doubts about whether or not certain objects have a certain property thus do not lead us to ever more precise precisions, but only to a revision of our precisions (cf. Burns 1991,79 f.).58

I prefer (iii), but, as I already said, my analysis of legal propositions is compati­ble with (ii) as well. Only (i) is excluded, because it entails the idea that our legal pro­positions are inconsistent, or that they are all false.

8. Excursus on Gaps of Cognition

In section 6, I have distinguished between gaps of cognition (or 'gaps of knowledge', in the terminology of Alchourr6n and Bulygn) - when we lack information about whe­ther or not a particular individual case is an instance of some generic case - and gaps of recognition - when we lack a clear semantic criterion that would enable us to deter­mine whether or not a particular individual case is an instance of some generic case be­cause it is a borderline case of the generic case -, and I suggested that only in the sec­ond case the propositions describing the case may lack a truth-value. The intuition un­derlying this distinction presupposes a realist conception of the world of fact. In legal decisions, propositions referring to facts are a necessary element for the foundation of the conclusion reached by an organ of application. Moreover, they are propositions re­ferring to facts of the past, and therefore presuppose a realist conception of the past.

Dummett (1978, 358-374), however, has challenged that realist conception of the past. He argues that antirealist arguments apply even to assertions referring to the past (Dummett 1978, 363):

"The anti-realist's case consisted of an application to statements about the past of the general form of anti­realist argument. We learn the use of the past tense by learning to recognise certain situations as justifying the assertion of certain statements expressed by means of tense ... The only notion of truth for past-tense statements which we could have acquired from our training in their use is that which coincides with the justifiability of assertions of such statements, i. e., with the existence of situations which we are capable of recognising as obtaining and which justify such assertions."

The proposition expressed in the example of section 6 - 'On September 25, 1995, there was an uneven number of blades of grass on the Bellaterra Campus of Barcelona's Autonomous University' - would, thus, have no truth-value, since we have no way of justifying it, and certainly never will.

If we would adopt a position like that of Dummett, propositions about the past could also have no truth-value and would have to be treated like propositions referring

58 Though controversial, this is probably the approach most closely in accordance with TL. In fact, it is the reply Dworkin (1985, 130 f. and 405, n. 3) puts in the mouth of V - a legal positivist arguing in favour of vagueness and likely to share TL - in order to refute the attacks of R - a discussant pointing out to V that he cannot account for higher-order vagueness. Dworkin himself, however, shares neither V's nor R's argu­ments.

A LOGICAL ANALYSIS OF LEGAL PROPOSITIONS 87

to borderline cases. They would constitute a new ground of indeterminacy for legal pro­positions. But then, if we have a legal statement like 'Legally, x ought to be punished with 10 to 15 years, for committing manslaughter' and there is no way of justifying the assertion that x committed manslaughter, that legal statement would express a proposi­tion without a truth-value. That means that propositions referring to so-called perfect crimes would have no truth-value.

Legal constructivism does not force us to accept such an antirealist position with respect to the past. In fact, I very much doubt that it is necessary to adopt such a broad antirealist conception. But it can perhaps give plausibility to certain conceptions usually found in legal theory. Thus, for example, Kelsen holds the thesis of what he calls the constitutive nature of judicial sentences (Kelsen 1960, 242-246). In his view, judicial sentences do not have a merely declarative nature; rather, they first of all constitute the applicable norms, and - what is more important for the present purpose - the verifi­cation of the facts also has a constitutive function. In Kelsen's words (1960, 245):

"The facts of a case are not understood to hold only from the moment of their determination, but from the time determined by the law-applying organ, that is, from the time at which the natural facts of the case -according to the determination of the law-applying organ - held. Thus, the determination of the condition­ing facts of the case by the court is constitutive in every sense. ,,59

Kelsen's position becomes more plausible if one accepts that propositions about events of the past are true or false according to the capacity to justify the statements expressing them when they are asserted. But even then, the constitutive nature of these propositions is controversial, since it may well be that without having an adequate foundation the organ of application asserts that certain events have - or have not - taken place (cf. Alchourr6n and Bulygin 1991a, 309-313). But even with respect to propositions re­ferring to facts, there may be indeterminacy.

As I already said, I do not wish to defend a global form of antirealism including that antirealist conception of the class of propositions referring to facts of the past (cf. the discussion in Wright 1993, 176-203). The constructivism adopted here for legal pro­positions is compatible with a realist conception of propositions about the past. But should someone wish to incorporate that antirealist conception, TL would offer the ade­quate logical instruments. For example, it would be possible to say that

(1) x did tP in t

has no truth-value in tl' which is posterior to t. And that, therefore,

(2) It is true in t\ that x did tP in t

59 "Der Tatbestand gilt nicht erst a1s mit dem Zeitpunkt der Feststellung gesetzt, sondem a1s mit dem von dem rechtsanwendenden Organ festgestellten Zeitpunkt, das heisst mit dem Zeitpunkt gesetzt, mit dem der natiirliche Tatbestand - dec Feststellung durch das rechtsanwendende Organ zufolge - gesetzt wurde. Die Feststellung des bedingenden Tatbestandes durch das Gericht ist also in jedem Sinne konstitutiv."

88 CHAPTER II

is false. Or that it is true that

(3) It is not true in tl that x did tP in t and it is not true in tl that x did not do tP in t.

Perhaps if, in spite of the truth of (3), an organ of application maintains that (2) one can understand Kelsen's thesis that such an organ constitutes the truth of the proposition ex­pressed in (1). The thesis is incompatible with a realist conception of propositions about the past, but it could be made compatible with an antirealist conception of such proposi­tions - a possibility I will not analyze here.

9. Conclusions

In this chapter, I have defended a constructivist or antirealist conception of legal pro­positions that can show how a legal system can leave the deontic status of certain states of affairs indetermined. We can give the following definition of normative determinacy (Alchourr6n 1969, Bulygin 1993):

"A state of affairs p ist normatively determined in a normative system a if and only if p is either positively permitted or prohibited in a, i. e., when the expression 'Pap v Oa-.p' is true."

When that expression is false, the normative system is indeterminate. The expression is false when the legal propositions referring to state of affairs p have no truth-value, that is, when the propositions expressed in statements like 'Legally, p ought to be done' have no truth-value and, therefore, the propositions expressed in 'Legally, it is not true that p ought to be done' and 'Legally, it is not false that p ought to be done' are true.

This happens mainly when system a has a normative gap with respect to p. Thus, incomplete legal systems give rise to indeterminacies, and this is shown in that some legal propositions referring to incomplete systems have no truth-value.

But because of our requirement that a state of affairs p is correlated with some normative situation in a if, and only if, among the consequences of a there is a relevant norm correlating p with some solution, and because if a is an inconsistent system, it has no relevant consequences, in an inconsistent system any state of affairs is normatively indeterminate in a relevant way, although any state of affairs is normatively determined by an infinite number of irrelevant consequences.

Thus, legal propositions about an inconsistent system also have no truth-value and, therefore, such systems too give rise to indeterminacies. In other words, that a le­gal proposition is true not only presupposes that certain normative consequences belong to a certain normative system, but also that certain relevant consequences belong to that system, i. e., it presupposes, among other things, that the system is consistent.

In conclusion, we can say that

A generic case Ci in a legal system LS is legally indeterminate if and only if either LS is inconsistent or there is a gap in case C,.

A LOGICAL ANALYSIS OF LEGAL PROPOSITIONS 89

Only consistent and complete systems can be described by a class of legal propositions for which bivalence is valid, and since both properties are contingent in legal systems, there is always a possibility of indeterminacy.

We now need a notion of indeterminacy for individual cases. Despite the obser­vations in section 8, I will adopt a realist position with respect to the world of fact. The only basis of indeterminacy regarding individual cases is then semantic indeterminacy, i. e., the vagueness of our concepts. We can, thus, formulate the following definition:

An individual case c,' which is an instance of a generic case C, is legally indeterminate if and only if either C is legally indeterminate or Ci is a borderline case of C.

Thus, the legal proposition expressed in 'Legally, x ought to do t/J' can lack a truth-value if, although there is a relevant consequence obligating all F to do t/J, we are not sure whether predicate F applies to x, i. e., if x is a borderline case of the application of pre­dicate F.

Before concluding the chapter, I wish to add two considerations ex abundante cautela:

(i) Controversies about legal indeterminacy usually concern doubts about whe­ther certain meanings should be given to certain norm-formulations. Although this is an important question, to which I will return later, the purpose of this chapter has been more modest: Even in a set of norm-formulations that have already been interpreted­with unequivocally assigned meanings -, there is room for indeterminacy. This is one of the most important conclusions of this chapter.

This may also explain why the question of ambiguity has not been treated in this chapter. We can say that a linguistic expression is ambiguous if, and only if, it has more than one meaning. Thus, a term is ambiguous if it expresses more than one concept; a statement is ambiguous if and only if it expresses more than one proposition; a norm­formulation is ambiguous if, and only if, it expresses more than one norm. When we are dealing with an already interpreted set of norm-formulations, we presuppose that we have unequivocally assigned one single norm - or, if the system is consistent, more than one, but mutually compatible norms - to each norm-formulation.

We can say that while ambiguity is a property of terms, statements, and norm­formulations, vagueness is a property of concepts, propositions, and norms (cf. von Wright 1963b, 13).

(ii) The logical analysis of legal propositions performed in this chapter is based on legal constructivism. But, as can be seen in the appendix to this chapter, one does not need to adopt that position in order to accept the chapter's conclusions about legal indeterminacy. All that has been said about legal propositions without a truth-value can be upheld, without giving up bivalence, by distinguishing the internal from the external negation of such propositions. And with respect to semantic indeterminacy, one can ac­cept an epistemic conception of vagueness - vagueness as a phenomenon of ignorance - in order to retain bivalence. In that case, it should be clear that vagueness no longer is a foundation of indeterminacy in a strong sense: the truth-value of the propositions at-

90 CHAPTER II

tributing some property to borderline cases is no longer indeterminate, only our knowl­edge is.

Even those who, like Quine (1970, 1-3), doubt that there exist such entities as propositions can accept the conclusions of this chapter about indeterminacy. Nothing of what has been said presupposes that the theses advanced here could not be reformulated in a nominalist language that only accepts particulars in its ontology and speaks of sen­tences or token sentences rather than of propositions.

As will be seen later, only two philosophical positions on the analysis of legal propositions are excluded: (a) some especially robust forms of philosophical realism for which states of affairs are normatively determined by some normative reality which is independent of the norms issued by the authorities, and (b) some radical forms of skep­ticism60 for which all our legal propositions lack a truth-value.

I think that the exclusion of both positions is justified since they cannot account for certain platitudes underlying our use of legal propositions, especially the platitude that norm-authorities can change the normative status of our actions.

All other philosophical positions can be seen as in a kind of overlapping consen­sus61 which, despite their deep philosophical disagreements on questions of ontology, logic and semantics, can provide a reconstruction of our use of legal propositions and of the place occupied by indeterminacy in the law.

60 It should be noted that what is commonly known as legal realism actually is a form of skepticism totally opposed to philosophical realism.

61 As is commonly known, the expression is due to Rawls (1987; 1993, 131-172) who applies it to his con­ception of justice. Here, it is used only in the sense of a consensus between 'comprehensive and reasonable doctrines' .

A LOGICAL ANALYSIS OF LEGAL PROPOSITIONS 91

ApPENDIX: TRUTH-CONDITIONS AND THE LOGICAL ANALYSIS OF LEGAL STATEMENTS

In this appendix, I will present an alternative way of reaching the same conclusions as in section 4, where the truth-conditions of legal statements were examined, on the basis of a paradigmatic example from logical analysis. I mean the controversy between Rus­sell (1919) and Strawson (1950) about the analysis of defined descriptions.

Russell and Strawson want to analyze statements like

(1) The present King of France is wise.

They agree on the following:

(i) The sentence has meaning, and if someone were to pronounce it now, he would be pronouncing a sentence that makes sense. (ii) If someone were to pronounce that sentence now, he would be formulating a true sentence only if there really existed one single King of France now, and that King were wise.

But Strawson criticizes Russell - among other things - for endorsing the following assertion:

(iii) Whoever pronounces statement (1) is formulating an assertion that is either true or false.

As is commonly known, Russell's analysis of (1), which I will call a predicative sen­tence, is as follows:

(1 a) There is one and only one x such that x is King of France and x is wise.62

Thus, for Russell, (1) implies

(2) There is one and only one King of France.63

(2) will be called an existential sentence. Thus, if there is no King of France, then, ac­cording to Russell's analysis, (1) is false.

According to Strawson, the relation between (1) and (2) is not one of implica­tion, but of presupposition. Strawson understands that relation of presupposition in the

62 In predicate logic, (la) could be expressed in the following way:

(lb) 3y (';Ix (x is the present King of France H x = y) 1\ (y is wise»

Cf. Quine (1970,25 f.; 1981,299-304).

63 That is:

(2a) 3y (';Ix (x is the present King of France HX = Y»

92 CHAPTER II

following way: (1) presupposes (2) if, and only if, (2) is true if (1) is true or false; or, what amounts to the same, (1) presupposes (2) if, and only if, (1) is neither true nor false if (2) is not true. This enables him to conclude (Strawson 1950,66):

"So when we utter the sentence without in fact mentioning anybody by the use of the phrase, 'the King of France', the sentence does not cease to be significant: we simply fail to say anything true or false because we simply fail to mention anybody by this particular use of that perfectly significant phrase."

Statements like (1) assume the existence of certain entities, just as legal statements as­sume the existence of certain norms. Take the following legal statement:

(LSt) Legally, all F ought to do 1/>.

As we have seen, (LSt) can be analyzed in the following terms:

(LSt) There is a norm N in LS, and N stipulates: 'All F ought to do 1/>'.

Hence, according to Russell, (LSt) implies

(LSt2) Norm N exists in LS, i. e., N belongs to the normative consequences of LS.

In contrast, according to Strawson, (LSt) does not imply, but rather presupposes (LSt) Which means that if (LSt2) is not true, then (LSt) has no truth-value. A1chourr6n (1991, 535-543) has tried to show, among other things, that Strawson's theory adds nothing to that of Russell. One of the reasons that assertion rests on is the distinction between two senses of falsity in Strawson's thesis (as Strawson himself admits in Strawson 1967, 76 f.). For Strawson, a predicative sentence, i. e., a sentence containing a defined de­scription, can be false in two senses. Hence, (1) can be false in two senses:

And

(3) (1) is false! if and only if the description it contains describes one single ob­ject that does not have the property indicated in the predicate.

(4) (1) is false2 if and only if the condition of application for (1) to be true does not hold.

With these two senses of falsity, there are then two senses of implication underlying Strawson's theses:

(5) (1) implies] (2) if, and only if, (1) isfalse] if (2) is not true. (6) (1) implies2 (2) if, and only if, (1) isfalse2 if (2) is not true.

Hence, Russell is right in saying that (1) not only presupposes, but also implies2 (2). The second sense of implication means that if (1) is true, then (2) is true; but that (1) pre-

A LOGICAL ANALYSIS OF LEGAL PROPOSITIONS 93

supposes the truth of (2) also entails that if (1) is true, then (2) is true too.64 That means that if (1) presupposes (2), then (1) implies2 (2).

Now, the sense Strawson gives to implication is probably the first of the two sen­ses - the one reflected in (5). The notion of falsity in the first sense of implication is that of falsity!. And we can very well hold that (1) can presuppose (2) without implying! (2). For (1) to imply! (2), the not-truth of (2) must lead to the falsity! of (1); but it is possible that (2) is not true - because there is no King of France - and (1) is not false! - because the description it contains does not describe anything. That means that the law of excluded middle does not hold for falsity l' since

(7) (1) is true or is false!

does not hold. There can be predicative sentences without a truth-value, and in that sense, (1) presupposes (2) without implying! it. Strawson's theory of predicative senten­ces can be summarized as follows (cf. Alchourr6n 1991,539):

(a) (1) presupposes (2). (b) If (1) is true or falsel' then (2) is true. (c) If (1) is true, then (2) is true, that is, (1) implies2 (2). (d) If (1) is falsel' then (2) is true. (e) If (2) is not true, then (1) is not true and (1) is not false!. (g) (1) is true or false2•

(h) (1) does not imply! (2).65

I think that truth-logic can account for this twofold behaviour of falsity in Strawson, and therefore also for his notion of presupposition. I will denote (1) with p and (2) with q. Then, to say that p presupposes q means that

(a') Tp v T-,p ~ Tq.

From (a'), we can deduce

(b') Tp ~ Tq and (c') T-,p ~ Tq.

Also,

(d') -.Tq ~ -.Tp A -.T-,p.

And even if it holds that

64 If it is true that (1) presupposes (2) if and only if (2) is true if (1) is true or false, then it is also true that if (1) presupposes (2), then if (1) is true, then (2) is true too.

65 When (2) is not necessarily true.

94 CHAPTER II

(e') Tp v -,Tp,

it does not hold that

(f') Tp v T---,p.

Strawson's falsity! is equivalent to (T ---,p) being false in TL, and falsity 2 is equivalent in TL to (-,Tp) not being true. In this way, one can understand that while (I') is not valid - it is not a theorem of TL -, (e') - the law of excluded middle in TL - is valid. Also, one can understand a formulation like (d') which says that when some sentence is not true, then some other sentence has no truth-value. And one can also understand the notion of presupposition expressed in (a'), and the fact that this notion does not entail the notion of implicationp although it does entail that of implicationr The notion of im­plication! can be expressed in TL as follows:

(g') p implies] q if, and only if: if -,Tq, then T---,p.

(g') is invalid since, according to (d') - which expresses Strawson's idea of presuppo­sition -, if -,Tq, then -,Tp and -,T---,p.

In contrast, the notion of implication2 is valid in TL:

(h') p implies2 q if, and only if: if -,Tq, then -,Tp.

(h') is derived from (d') - according to which if -,Tq, then -,Tp and -,T---,p. This en­ables us to say that in TL Strawson's conclusion can be accepted: (I) presupposes (2), which entails that (1) implies2 (2), but that (1) does not imply! (2).

Alchourr6n (1991, 539-542) has shown that these conclusions do not differ in any significant way from Russell's conclusions, even if Strawson did not see it this way. Strawson does not pay attention to a passage in Russell's work that shows how Straw­son's ideas were already implicit in Russell (1919, 55):

"When descriptions occur in propositions, it is necessary to distinguish what may be called 'primary' and 'secondary' occurrences ... An instance will make this clearer. Consider 'the present King of France is bald'. Here 'the present King of France' has a primary occurrence, and the proposition is false. Every proposition in which a description which describes nothing has a primary occurrence is false. But now consider 'the present King of France is not bald'. This is ambiguous. If we are first to take 'x is bald', then substitute 'the present King of France' for 'x', and then deny the result, the occurrence of 'the present King of France' is secondary and our proposition is true; but if we are to take 'x is not bald' and substitute 'the present King of France' for 'x', then 'the present King of France' has a primary occurrence and the proposition is false. Con­fusion of primary and secondary occurrences is a ready source of fallacies where descriptions are concer­ned."

As Alchourr6n (1991, 540 f.) observed, Russell's notions of primary and secondary oc­currence correspond to Strawson's falsity! and falsitY2' respectively. Russell, of course, symbolizes predicative sentences as:

A LOGICAL ANALYSIS OF LEGAL PROPOSITIONS 95

(8) G(u)Fx

which is defined as

(9) 3y ('ix (Fx ~ x = y) /\ Gy).

Russell then considers what happens when a predicative sentence is negated. According to him, there are two senses in which we can conceive of the negation of (8).66 First, there is the sense of negation corresponding to the so-called secondary occurrence. This can be symbolized by

(10) -,G(u)Fx.

(10) is true if and only if (9) isn't:

(11) -,3y ('ix (Fx ~ x = y) /\ Gy).

That means that (10) is true if, and only if, (8) is false2, in Strawson's terminology. But there is another sense in which (8) can be negated, which corresponds to

Russell's primary occurrence. This sense can be symbolized by

(12) lG(u)Fx.

(12) is true if, and only if, it is also true that:

(13) 3y (\Ix (Fx ~ x = y) /\ -,Gy).

This means that (12) is true, if and only if, (8) is false!, in Strawson' terminology. From this we can conclude that there are no relevant differences between Rus­

sell's and Strawson's analyses of predicative sentences. In Russell, we must also distin­guish two senses in which a predicative sentence can imply an existential sentence. Rus­sell symbolizes existential sentences by

(14) 3!(u)Fx

which is defined by

(15) 3y (\Ix (Fx ~ x = y)).

66 For the same two senses of the negation of predicative sentences, cf. also Searle 1969, 159; Grice 1989, 271 f.

96 CHAPTER II

Strawson's conclusions according to which a predicative sentence not only presupposes but also implies2 an existential sentence, but does not imply! an existential sentence, are also true in Russell's theory.

That (8) implies2 (14) can be shown as follows: (8) implies2 (14) if, and only if, (8) is false2 if (14) is not true. (14) is not true when:

(16) --,3y (\:Ix (Fx H x = y»,

and (8) is false2 if, and only if, the predicative sentence is negated in its secondary oc­currence, that is, if (10) holds, which is equivalent to (11):

(11) -.3y (\:Ix (Fx H x = y) 1\ Gy).

That (8) presupposes (14) means that if (8) is true or false!, then (14) is true, or that if (14) is not true, then (8) is neither true nor false!. (14) is not true when:

(16) --,3y (\:Ix (Fx H x = y»,

and thus, as can easily be seen, when it is not true that

(9) 3y (\:Ix (Fx H x = y) 1\ Gy),

which is equivalent to (8). But (8) is also not false!, because that is the case when (8) is negated in its primary occurrence, i. e., when

(13) 3y (\:Ix (Fx H x = y) 1\ -,Gy).

Finally, it can also be shown in Russell's theory that (8) does not imply! (14). That (8) implies! (14) means that if (14) it not true, then (8) is false!. We have seen that (14) is not true when (16) holds, and that this does not entail the falsity! of (8), as reflected in the negation of its primary occurrence in (13).

Neither

(17) G(tx)Fx v lG(tx)Fx,

the law of excluded middle with falsity!' nor

(18) 3!(tx)Fx -+ lG(tx)Fx,

Straws on 's first sense of implication, are theses of Russell. Since, if presented in this way, Russell's and Strawson's theses are equivalent,

one can assume that it is possible to reconstruct a parallel version of them for legal pro­positions.

Thus, we can distinguish two senses of falsity for

A LOGICAL ANALYSIS OF LEGAL PROPOSITIONS 97

(LSt) Legally, all F ought to do l/J.

In order to do this, we must reformulate the two senses of falsity:

(F) (LSt) is false J if and only if the negation of norm N belongs to LS. (F) (LSt) is false2 if and only if the condition of application for (LSt) to be true does not hold.

The negation of norm N - 'All F ought to do l/J' - is 'An F may do not-l/J'.67 There­

fore, (LSt) is false! if there is a norm in LS which permits an F not to do l/J. Also, we can distinguish two senses of implication between (LSt) and the legal

statement

(LSt2) Norm N exists in LS,

according to the two senses of falsity:

(I) (LSt) impliesJ (LSt) if and only if (LSt) isfalseJ if (LSt) is not true. (I) (LSt) implies2 (LSt) if and only if (LSt) isfalse2 if (LSt2) is not true.

We can now conclude that although (LSt) not only presupposes but also implies2 (LSt2), it does not imply! (LSt2).

On the other hand, Alchourr6n and Bulygin (Alchourr6n 1969; Alchourr6n1Bu­lygin 1989; Bulygin 1993) have constructed a logic of normative propositions that can be regarded as the Russellian version of the analysis of legal propositions. In their view, we must distinguish between normative operators in a prescriptive sense ('0' for 'obli­gatory', and 'P' for 'permitted' will be enough, since 'prohibited' is equivalent to 'not permitted')68 and in a descriptive sense. The latter are contained in expressions that enunciate propositions about a norm or set of norms. An expression like 'Oap' means

that in the system of norms a there is a normative consequence obligating one to do p:

Oap =def. 'Op' E NC(a).

In the case of permission, the situation is more complex since to say that p is permitted in a is ambiguous: It can mean that there is a norm in a permitting p - strong or posi­

tive permission -, or that there is no norm in a prohibiting p, i. e., '0,1' does not be-

67 This conception of the negation of a nonn corresponds to the one presented by von Wright 1963a, 138 f., and satisfies the requirements stipulated there, one of which is that just as the negation of a proposition is a proposition, the negation of a nonn is a nonn. In von Wright's own words: "On our suggestion the negation of a positive command is a negative pennission and conversely, and the negation of a negative command is a positive permission and conversely."

68 Actually, we only need one operator, since they are interdefinable; but for the sake of simplicity of the presentation, I will use the two operators mentioned.

98 CHAPTER II

long to the consequences of a - weak or negative permission. The two kinds of per­mission can be defined formally as follows:

P+ap =def. 'Pp' E NC(a).

P-ap =def. 'O-,p' Ii!: NC(a).

The only part of the logic of normative propositions I am interested in here is the one referring to the logical behaviour of negation. In the logic of normative propositions, there are two kinds of negation: external negation (-.) and internal negation (1). Let us look at how these negations behave in the case of operator O. First, external negation:

-.Oap =def. 'Op' Ii!: NC(a).

Then, internal negation:

lOap =def. '-,Op' E NC(a) =def. 'P-,p' E NC(a).

Note that '-.Oap' is equivalent to the negative permission not to do p, i. e., to 'P-a-,p', and that 'lOap' is equivalent to the positive permission not to do p, i. e., to 'P+a-,p'.

We can now say that (LSt) is synonymous to 'Oap', and that that expression implies2 (LSt2), sind if (LSt2) is not true - if norm N does not exist - then (LSt) is false2, i. e., we have the external negation of (LSt), '-.Oap'.

It is also correct that 'Oap' presupposes the truth of (LSt), i. e., the existence of N among the normative consequences of a. That means that if 'Oap' is either true or false! - i. e., the internal negation of 'Oap': 'lOap' is true, the permission not to do p is a normative consequence of a -, then (LSt) is true.

However, 'Oap' does not imply! that norm N exists. Because in that case the non-existence of norm N - i. e., that (LSt2) is not true - would entail the falsity! of (LSt) and, thus, the truth of the internal negation of 'Oap': 'lOap', which means that the permission not to do p belongs to the set of normative consequences of a. But that (LSt2) is not true is compatible with the fact that this permissive norm does not exist. In fact, in the logic of normative propositions the principle of excluded middle is valid with respect to the external negation of 'Oap' , but not to its internal negation. Thus,

(19) Oap v -.Oap

is a valid formula, whereas

(20) Oap v lOap

is invalid, since it is possible that neither a norm obligating one to do p nor a norm per­mitting one not to do p belongs to a. That is the reason why it is maintained that the corresponding normative proposition has no truth-value, and therefore, just as in Straw-

A LOGICAL ANALYSIS OF LEGAL PROPOSITIONS 99

son's analysis, internal negation, i. e., the first sense of falsity, acquires a privileged position.

This shows that the logic of normative propositions can represent the relations between legal propositions, and that it is equivalent to the presentation in terms of the truth-logic TL in the previous section. The only reason for preferring the presentation I have chosen here is the purpose of the present investigation which is to show the sour­ces of the indeterminacy of the law. This seems to be clearer if one can say that some legal propositions have no truth-value (thus giving privilege to the first notion of falsi­ty), rather than that there are several senses of negation. However, from what has been said it follows that the notions of internal and external negation of normative proposi­tions have their correlates in our vocabulary in the expressions 'Legally, it is false that all F ought to do l/J', and 'Legally, it is not true that all F ought to do l/J'.

In conclusion, then, although Strawson thought that his theory differs from Rus­sell's more than it actually does, in my view, Strawson's reasons for saying that predi­cative sentences have no truth-value when the corresponding existential sentences are not true - reasons which stem from a particular conception of what it means and en­tails to assert a proposition - are equally valid for the relation between legal proposi­tions and the sentences that assert the existence of certain normative consequences in a certain normative system.

Dworkin (1977c, 289 f.) too has noted the similarity between legal propositions and propositions referring to non-existent entities:

"I should mention, however, one possible objection to this portion of my argument not foreseen in that dis­cussion. This appeals, in a general way, to an argument which is familiar among philosophers of language, namely that propositions about non-existent entities are neither true nor false. There is a tradition which ar­gues that the proposition that the present King of France is bald is neither true nor false (though there is also a tradition which argues that this proposition, properly understood, is simply false). The proposition about the King of France ... (according to one view) is neither true nor false. So (I have heard it argued) proposi­tions of law ... may also be neither true nor false. U

Dworkin thinks, however, that the analogy is useless because propositions referring to non-existent entities presuppose, but do not assert, the existence of the proposition's subject (which is non-existent), whereas what legal propositions in hard cases assert or deny is "the existence of a legal right or some other legal relation" (Dworkin 1977c, 290). And he adds:

"The comparable proposition is the proposition that there is now a King of France. No one supposes that that proposition is neither true nor false. It is either (as most of us think) false, or (as some extreme supporters of the Comte of Paris believe) true. U

Dworkin's position on this point presupposes that legal propositions assert the existence of rights, obligations, and other legal relations, i. e., that they assert the existence of some normative reality. The thesis supporting that assertion is a (Platonic) realist thesis about that normative reality and will be discussed in Chapter V. But it can already be announced that it is a thesis that is incompatible with the legal constructivism presented here. According to constructivism (legal positivism), that way of understanding legal

100 CHAPTER II

propositions would convert them all into false propositions. For a legal positivist, the propositions asserting that certain persons have certain legal rights or duties presuppose that certain normative consequences belong to a legal system, that is, they presuppose the existence of certain complex social facts of norm creation and elimination.

III. LEGAL DYNAMICS

1. Introduction: Legal Order and Legal System

In the two previous chapters we have looked at sets of legal norms as sets that are struc­tured by the relation of deductibility. Thus, we have considered legal systems as static systems, that is, systems containing all their logical consequences. It is a notorious fact, however, that the law changes over time, that there are acts of norm-creation (enacting) and acts of norm-elimination (derogation). Because of this, it is said that legal systems are dynamic. 1 The criterion habitually used to account for the dynamic structure of legal systems is the so-called genetic or legality criterion (Raz 1970, 183-185; Caracciolo 1988,57 f.). According to this criterion, a norm N belongs to a legal system LS if there is another norm N' in LS authorizing some organ 0 to enact N, and if 0 enacts N; simi­larly, a norm N ceases to belong to a legal system LS if there is another norm N' in LS authorizing some organ 0 to derogate N, and 0 derogates N.

Obviously, the genetic structure of a legal system depends on the fact that certain power-conferring norms or norms of competence belong to the system. In legal-philo­sophical scholarship, the nature of such norms has been extensively discussed. One of the central points of that discussion is the question whether or not those norms are to be regarded as prescriptive or as conceptual rules. I will not explore this question here.2

For my purpose, it suffices to accept that in many cases whether a particular norm be­longs to a particular legal system depends on whether another norm belongs to that same system which confers on some organ the power to enact that first norm. For the notion of a chain of norms, which will be needed here, it does not matter whether such a norm is regarded as a permission (authorizing the performance of some normative act), as a conceptual rule (identifying as valid the norms created in the exercise of the cor­responding competence), or as something else (e. g., as obligating one to behave in ac­cordance with the norms created in the exercise of that competence).

The genetic criterion enables us to account for the introduction and elimination of legal norms through specific normative acts. Enacting and derogating norms brings about a change from a system LSI to another system LSi. And since the identity of a set is a function of the identity of its content, every time the content is modified, we have a different set. This approach to the identity of a system, however, is inconsistent with an­other widely shared intuition that can be expressed as follows: Although the content of the law changes as a consequence of the enactment or derogation of norms, the legal system - e. g., the Spanish law - maintains its identity over time.

In order to solve this problem, I will introduce the distinction between a legal system and a legal order (Alchourr6n1Bulygin 1971, 1981; also Caracciolo 1988; More-

1 For the distinction between static and dynamic systems of norms, cf., for example, Kelsen 1945, 112 f.

2 On this point, cf., for example, Bentham 1970, 27f.; Hart 1961,78 f.; Kelsen 1945, 143; 1960, 57f.; 1973, 83,210; Raz 1970, 166-168; Ross 1958,32; 1968, 113; von Wright 1963a, 192; AtienzalRuiz Manero 1994, 1995; Bulygin 1991b; Caracciolo 1995a; Hernandez Marin 1984, 33-43; 1989, 153-166; Mendonca 1992; MendoncaIMoresolNavarro 1995; Paulson 1988; Spaak 1994.

102 CHAPTER III

so/Navarro 1993a, 1993b). Legal systems are momentary normative systems; every time a norm is enacted or derogated by a competent authority, the result is a different system. Legal orders are sequences of legal systems. The identification of a legal order LO conceptually presupposes the identification of the first system of the sequence (in what follows, the originating legal system LSD); further systems LSI' LS2 , LSJ, ••• , LS. be­long to LO if, and only if, they satisfy certain membership criteria, such as the legality of the normative acts.

The concepts of 'legal system' and 'legal order' can be understood as a new way of expressing the distinction introduced by Raz (1970, 34 f.) between momentary and non-momentary legal systems. However, it is important to underscore three differences concerning (i) the membership of legal systems in legal orders, (ii) the judges' obliga­tion to apply the norms of a legal system, and (iii) the relationship between the notions of legal order and non-momentary legal system and the notion of national law.

(i) The logical relations between momentary and non-momentary systems are characterized in an ambiguous way by Raz (1970, 34 f.; cf. Bulygin 1982b, 80). On the one hand, Raz asserts that momentary systems belong to non-momentary systems. On the other, he also holds that momentary systems are subclasses of non-momentary sys­tems. Since the relations of membership and inclusion between sets are logically diffe­rent (e. g., while inclusion is transitive, membership is not),3 with Raz's distinction it is not clear whether non-momentary systems are to be regarded as systems of norms, or rather as sets of systems of norms. According to the definition of 'legal order' and 'le­gal system', a legal system is a set of norms, and a legal order is a set of systems. Since the relation of membership is intransitive, legal norms do not belong to a legal order; the only elements of legal orders are (momentary) legal systems.

(ii) According to Raz (1980, 211), a momentary system only contains such norms which the judges have the obligation to apply. The distinction between system and order, in contrast, does not presuppose that all norms of the system are applicable at a certain time t. As I will later show in detail, a norm may belong to a system and still be inapplicable (i. e., the judges have the obligation not to apply it). That is the case, for instance, during the time of vacatio legis of a norm, when it belongs to the system of time t, but is not yet applicable in t.4

(iii) Raz's non-momentary systems seem to coincide with the law of states, e. g. Spanish law, French law, German law, etc. The definition of a legal order given above, however, does not commit us to see it this way. If legal orders are regarded as sequen­ces of legal systems, it is an open question whether or not they coincide with the law of some state. For example, in many cases when a political system changes, the legal sys­tems arising from a new constitution do not belong to the same legal order as the old ones. Still, they may all be systems of the law of the same state. For instance, if one ac-

3 In MoresolNavarro 1993b, we have analyzed the distinction and its consequences for the relationships be­tween momentary and non-momentary systems in more detail.

4 At the time of writing (January 1996), this was the situation of the new Spanish Criminal Code, enacted in 1995, which undoubtedly belonged to Spanish law at that time (as proven by the fact that it could be dero­gated), but was not yet applicable.

LEGAL DYNAMICS 103

cepts (and I think there are good reasons for doing so) that the Spanish legal order ori­ginated from the Constitution of 1978 is a new legal order, then the systems prior to 1978 do not belong to the current Spanish legal order; but, obviously, that does not mean that Spanish law begins with the Spanish Constitution of 1978.

2. Identifying Legal Orders

The notion of a legal order, together with the criteria of deductibility and legality, is useful for the task of identifying a particular legal order. The conceptual rule satisfying that function I will call the rule of identification (RI). RI serves to assess whether or not two systems belong to the same sequence, i. e., the same legal order. It also serves to as­sess what norms belong to a certain system of the sequence. For this, one needs to iden­tify, flrst of all, the flrst system of the sequence (the originating system) and then the subsequent systems.

Since I am not interested in any speciflc legal order, the RI I will present here is only a definition pattern. For a deflnition of a particular legal order, the norms of the originating system would have to be spelled out one by one. The general structure of RI can be given in the following way (cf. Bulygin 1991a, 263 f.):

(1) Independent norms: Norms N/, N2, ••• , N. are valid in the legal system LSo' LSo is the originating system of the legal order LO/; norms N/, N2 , ••• , N. are the first constitution of LOr

(2) Introduction of norms. If a norm ~ is valid in a system LS, which belongs to LO/, and ~ empowers authority x to enact norm Nt' and x enacts Nt at time t, then Nt is valid in system LSI+/ (i. e., in the system corresponding to the moment following t), and LS,+/ belongs to LOr

(3) Elimination of norms. If a norm ~ is valid in a system LS, which belongs to LOp and ~ empowers authority x to derogate norm Nt which is valid in LS, and x derogates Nt at time t, then Nt is not valid in system LS,+/ (i. e., in the system cor­responding to the moment following t) that belongs to LOr

(4) Persistence of norms. The valid norms of a system LS" belonging to an LO/, which have not been derogated at time t are valid in system LSI+/ of LO/ (corres­ponding to the moment following t).

(5) Derived norms. All logical consequences of valid norms of a system LS, be­longing to an LO / are also valid in LS"

This rule calls for some comments: (a) Independent norms are identified extensionally, and their being valid does

not depend on the satisfaction of any systematic relationship with other norms. They are similar to what has sometimes been called sovereign norms (cf. von Wright 1963a, 197-204). Dependent norms are identified intensionally, on the basis of the satisfaction of certain relationships, namely, deductibility and legality.

104 CHAPTER III

(b) Independent norms, i. e., the first constitution, must be distinguished from RI, since RI is not a norm of any system. RI gives a recursive definition of a valid norm in a legal system LS which is a member of some legal order LO. In contrast, the first consti­tution is a set of valid norms. It makes no sense to predicate validity or invalidity of RI, as validity is equivalent to membership in a legal system.

(c) Legal systems are static systems of norms which are closed under the notion of logical consequence. When the constitution of LSo is identified, the content of the ori­ginating system is given by the logical consequences of that first constitution.

(d) The clauses of introduction and elimination of norms provide a conceptual explication of the genetic relationships between legal norms. The dynamic nature of the law is reconstructed by the genetic relationships between the norms belonging to an ordered pair of adjacent systems in a legal order. The genetic relationships are, thus, in­tersystemic (Raz 1970, 184 f.; Caracciolo 1988,67-73).

(e) As has been said before, RI implies that while legal systems are sets of norms, legal orders are sets of legal systems. Since the relation of membership is in­transitive, legal norms do not belong to a legal order. The elements of legal orders are static systems of norms. Besides, the expression 'legal order' is not the same as 'law of a state'. Normally, the law of a state must be reconstructed as a set of successive legal orders.5

(f) The continuity of the law is usually regarded as an essential characteristic of legal dynamics (cf. Hart 1961, 60-64). Therefore, RI also includes a clause - clause (4) - about the persistence of norms.

(g) A rule of identification like RI must not be confused with a social rule, ad­dressed to the judges, prescribing the application of norms that are identified by some criterion. Since Hart, such a social norm is usually known as a 'rule of recognition' (for the idea that the rule of recognition imposes obligations on judges, cf. Raz 1970, 197-200). RI is nothing but the conceptual rule enabling us to identify valid norms, and is not of a prescriptive nature (cf. Bulygin 1976).

Now, I wish to underscore that the identity of a legal order depends on its originating system (the norms of its first constitution), on the one hand, and on the criteria of de­ductibility (reflected in clause (5» and legality (reflected in clauses (2), (3) and (4», on the other. In that sense, we can say that the originating system has a special primacy, conferred on it by the structure of RI.6

It should be clear that RI provides a possible rational reconstruction7 of the no­tion of a legal order. That notion attempts to explicate in an adequate way our concep-

5 A. Merk! (1968, 1269) has expressed this idea as follows: ,,[Ejinem Staate im historisch-politischen Sinne [entspriichenj unter Umstiinden mehrere Staaten im Rechtssinne ... ; als Bruchstellen der Rechtskontinuitiit und Staatsidentitiit habe ich die Revolutionen gekennzeichnet." Cf. also Caracciolo 1988, 19 ff.; Bulygin 1991a; Mo­reso 1994b; MoresolNavarrolRedondo 1995.

6 In the following chapter, I will present this idea of the primacy of the originating system and its conse­quences in more detail.

7 For an explication of the rational reconstruction of a concept, cf. Alchourr6n1Bulygin 1971,7-9.

LEGAL DYNAMICS 105

tual intuitions about different questions, namely, the membership of norms in legal sys­tems, the identity of legal orders, the normative changes correlated with the perform­ance of certain normative acts of norm creation and elimination, etc. That conceptual reconstruction assumes certain constraints for our linguistic usage and, in that sense, de­limits the sphere in which our conceptual intuitions will be accepted.

Usually, the validity of norms is associated not only with their membership in certain systems, but also with their obligating force. One of the aspects in which this obligating force is manifested is the capacity of valid norms to serve as premises in the justification of binding legal decisions. From this point of view, it should be noted that validity as membership in some system SJ of time tJ in the sequence of systems of a legal order is not the same as validity as obligating force, which I will call applicability. That a norm belongs to some system Sj in time tj does not mean that is is applicable in tj'

Also, there are norms that do not belong to system Sj and which, nevertheless, are ap­plicable in tj'

In the following sections of this chapter, I will try to show in what way the dis­tinction between validity as membership and validity as applicability helps one under­stand how the dynamics of legal systems affect the justification of decisions, i. e., the resolution of cases. Moreover, this will also show how a case can be hard not only because of problems of indeterminacy, as analyzed in Chapter II, but also because of problems caused by the applicability of norms.8

3. The Applicability of Legal Norms

It is a widely shared opinion among jurists that legal norms are applicable when normative authorities, e. g., judges, ought to apply them. In that sense, the statement 'Norm N is applicable' usually refers to institutional duties of normative authorities. The institutional nature of the law is intrinsically connected with its social and normative dimensions. Some specific institutional practices, e. g., the primary organs' practices in the recognition of norms, provide the criteria for the existence and identity of legal systems. Thus, it seems possible to conclude that the identification of applicable norms - those determining insti­tutional duties - is one of the main topics of any legal theory.

The description of these institutional obligations, however, does not exhaust the theoretical relevance of the concept of applicability. At least two other important ques­tions must be mentioned.

On the one hand, the notion of applicability is linked to the identification of the truth-conditions of legal propositions. As we already know, a proposition about rights, prohibitions, obligations, etc., is a legal normative proposition. Such propositions refer to the deontic qualification of some action and, thus, offer relevant information about the legal status of the actions of individuals. The truth of a normative proposition is al­ways relative to a norm that is applicable at a certain time t. For instance, if a norm N

8 MacConnick (1978, 70-72; cf. also Atienza 1991, 138-140) has given a fourfold classification of hard cases in which he includes hard cases because of problems of relevance, i. e., problems in detennining the nonn that is applicable.

106 CHAPTER III

prescribes action p, but is not yet applicable at time t, individuals have no legal obliga­tion to perform p in t in accordance with N. The criteria of applicability of a system Sj enable all individuals (and not only the authorities) to know the deontic qualification of an action. Therefore, when one asserts that a norm N is applicable, often what is inten­ded is only to point out that individuals do not have the duty to behave in accordance withN.

On the other hand, the criteria of applicability of legal norms not only serve to identify the legal rights and obligations at a time t, but also to describe the relationships between the actions of authorities and of individuals. Authorities and individuals are rarely interested in the identification of applicable norms as a conceptual problem. Nor­mally, authorities and individuals wish to identify the norms that are applicable to spe­cific cases that affect them, i. e., those norms that are applicable in some litigation or conflict of interests. Normative authorities, especially judges, solve conflicts of interests through the application of general norms. Their decisions must be justified, prima jacie, by norms that are applicable to the facts.

An adequate reconstruction of the concept of applicability seems to be a useful analytic tool for understanding the way in which the law motivates and qualifies human behaviour. The motivation of behaviour is one of the main functions of the law. This function is fulfilled by applicable norms when they regulate actions that have not yet been performed. But the motivation of behaviour is not the only function of the law. Le­gal norms also serve as guidelines for the qualification of behaviour. For example, an action A performed at time t can be qualified in retrospect as legally obligatory, or pro­hibited, by a norm N enacted at a time t+n. Norms can be applicable to actions that were performed before their enactment, i. e., they can be retroactive; and although in that case obviously they cannot affect the behaviour of individuals, they offer a guide­line for determining the deontic status of those actions (Munzer 1977). As can easily be seen, the two functions of the law are intrinsically connected, at least for the following reason: Legal norms regulate judicial decisions, by prescribing the application of other norms, i. e., of norms that are applicable to certain cases; and the latter, in turn, can motivate and qualify the behaviour of individuals.9

Lawyers and legal philosophers often assert that a valid norm is a norm that has legal effects. In this context, the expression 'Norm N has legal effects' is equivalent to the expression 'Norm N is legally obligatory' or, in other words, 'Norm N is applicable' (cf. Guastini 1993,32 f.). This property of legal norms, as we have already seen, is in­timately related to the institutional nature of the law, because a norm can have legal ef­fects only if it is applicable by the judges and courts. Now, we need to mold a concept of applicability that enables us, on the one hand, to distinguish it from other notions, like that of membership, and, on the other, to show the conceptual relationships be­tween these two notions.

I will speak of applicability when the truth of the statement 'Norm N; is applica­ble to an individual case c' depends on the fact that another norm ~, which obliges to

9 On the relationships between norms addressed to judges and norms addressed to individuals, cf. Kelsen 1979, ch. 55.

LEGAL DYNAMICS 107

or permits the use of N j in c, belongs to some system Sj (Bulygin 1982b, 1991a). Ac­cording to this proposal, if it is true that a norm Nj is applicable to an individual case c at time t, then it is also true that another norm Nj belongs to the legal system of time t. It should be noted that there must not necessarily be a one-to-one relation between Nj and Nr Nj can also prescribe the application of other norms Nh , Nk, N" etc. Norm Nj offers one of the criteria of applicability of norms in a legal system Sr The normative subjects of the norms stipulating the criteria of applicability in Sj are normative authorities of Sj' e. g., the judges.

The actions of individuals and authorities are thus normatively linked through a network of applicable norms. We can, therefore, define the concept of applicability as follows:

DF 1. A norm N, is applicable at time t to an individual case c, which is an instance of a generic case C, if, and only if, another norm ~ which belongs to the system Sj of time t prescribes (obliges to, or permits) the application of N, to individual cases that are instances of C.

Legal systems are institutional systems of norms. The main characteristic of such sys­tems is the existence of norm-applying institutions. Thus, only if there are norm-apply­ing organs does it make sense to predicate the applicability of norms.

Although there are many interesting relationships between the applicability and the membership of norms in a legal system, the applicability of a norm does not imply its membership in the system of a certain time t, nor does membership imply applica­bility. On the one hand, not all norms that are applicable at time t to an individual case c belong to the system Sj of t. In cases regulated by international private law, a norm N of a legal system of state Ej may be applicable in another state Ej although N does not be­long to the legal system of state Er In that sense, legal systems can be characterized as open normative systems (Raz 1979, 116). On the other hand, not all norms that belong to a system Sj are applicable. This is the case, for example, when a norm Nj belonging to Sj is declared inapplicable to certain cases by another norm Nj • For instance, in many countries certain constitutional guarantees and rights can be declared temporarily inap­plicable, in accordance with the exercise of exceptional powers, e. g., exceptional par­liamentary powers.

Now that membership and applicability have been clearly distinguished, we can say that normative authorities habe the obligation to apply certain norms that are not identified by RI as belonging to the legal system of the time of application, and that they have the obligation not to apply certain other norms that can be identified as be­longing to the legal system in question. That is why the open nature of legal systems is an essential feature of the institutional nature of the law.

It must be noted, however, that RI conceptually still has a fundamental role in the identification of applicable norms. It is because of a norm that does belong to the le­gal system of time t that the norm-applying institutions must (or may) apply norms that do not belong to that system, or must (or may) not apply norms that do belong to it.

As I have already said, some very common situations in the application of the law that have to do with the role of time in the law can be understood with the help of that distinction.

108 CHAPTER III

This applies not only to the cases regulated by international private law men­tioned above, but also, for instance, to cases in which a norm stipulates the obligation of applying the criminal norm most favourable for the defendant, of those valid between the moment in which the crime was committed and the time of trial. Thus, it may be the case that a judge must ground her decision on a norm that has already been derogated, but that is more favourable, and which perhaps belongs neither to the legal system of the time when the crime was committed nor to the legal system of the time in which it is tried (Bulygin 1982b).

It may also be the case that a norm that belongs to the legal system is not yet ap­plicable because it is still in the period of vacatio legis (cf. n. 4 to this chapter).

The importance of the notion of applicability deserves to be emphasized, since the truth-conditions of legal propositions depend on the content of certain applicable norms. Hence, the characterization of the truth-conditions of legal propositions given in Chapter II must be complemented with what has been said about applicability now. But the theoretical fruitfulness of the concept of applicability does not stop here. The con­cept can also be useful in analyzing several problematic questions in legal theory.

As RI has been stipulated, some norms habitually used in justifying legal deci­sions cannot be identified as belonging to the legal system of a certain time. On the other hand, norms identified as belonging to the legal system of a certain time seem to be unfit candidates for appearing as justifications of judicial decisions.

The first situation is that, for example, of customary norms that do not result from a normative act of law-creation by a normative authority; or of some standards of morality, implied principles, etc. that cannot be identified through the clauses of RI; or of irregular norms (unconstitutional norms, illegal regulations, etc.) which, since they were issued by incompetent organs, do not belong to any legal system of the sequence; or of received norms, i. e., norms that belonged to the legal systems prior to the consti­tution of the legal order OJ and which are still applicable in OJ' but which no clause of OJ identifies as belonging to any system of OJ"

In the group of norms identified as belonging to the system but which are unfit candidates for appearing as a normative premise of a judicial justification, we can point out those norms that, although they belong to a certain legal system, come into conflict with other, prevailing norms, on the one hand, and irrelevant logical consequences to which I have dedicated part of Chapter I, on the other.

In the sections that follow, I will try to show how the concept of applicability can overcome these difficulties. Here, the term 'validity' will be used as synonymous with 'membership in a legal system' , in order to distinguish it clearly from the term 'ap­plicability' .

4. Applicable But Invalid Norms

The paradigmatic case of applicable, but invalid norms, in the sense that they do not be­long to the legal system at the time of application, is perhaps that of the application of norms already derogated. It is the best-known case among jurists, and all other cases are analogous to it.

LEGAL DYNAMICS 109

As suggested before, this explication can be extended to the case of the norms of a legal system of another state which are applicable to certain cases according to the norms of conflict in international private law. For instance, if an Italian makes his will in Spain and there are doubts about the validity of the act because of certain problems concerning the testator's capacity, and in accordance with the norms of conflict the capacity for making a will follows the law of the testator's country - as is the case in Spanish law -, then that conflict should be resolved according to Italian law. That means that the Spanish judge ought to apply the corresponding Italian legal norms which, obviously (according to the notion of membership expressed in RI), do not be­long to the Spanish legal system of that time.

In what follows, I will try to broaden that explication still more, so that it will in­clude the cases of customary norms, standards of morality, irregular norms, and re­ceived norms.

a) Customary norms

RI does not enable us to account for customary norms. Therefore, when the notions of legal order and legal system are presented, one usually adds the reservation that these notions do not reconstruct how customary norms belong to the law (Alchourr6n1Buly­gin 1979, 18 f.). Nevertheless, the courts in most legal systems apply norms arisen from custom, because they recognize that unter certain conditions such norms possess bind­ing force.

If membership is not the ground of the binding force of customary norms, then what is it based on? A tentative explication could consider such norms, which do not belong to any legal system, applicable in accordance with norms stipulating under what conditions customs ought to be taken as a foundation for certain judicial decisions. Even in countries where the theory of the sources of the law concedes a privileged place to customs, some requirements customs must fulfill in order to be applicable are usually added (e. g., that they are not applicable in all legal areas, that they may not contradict the law, nor public order or morality, and that they must be well-established) (cf. Raz 1972a, 853).

I think this conception does not fall under the objection Hart (1961, 43-47) ex­pressed against those who ground the validity of customs in some act of tacit accept­ance by the authorities, nor does it assume that the applicability of customary norms de­rives from the fact that they are recognized by the courts. The courts apply them be­cause under certain conditions they have the obligation to do so. Now, once applicabil­ity and membership are clearly distinguished, there is no need to assert that customary norms belong to some legal system in order to recognize that they have binding force and ought to be applied by the courts.

b) Standards of morality

It is often argued that judges, in deciding a case, take into account not only legislated norms, but also certain standards of morality. It has even been held that disregard of this fact is one of the main shortcomings of legal positivism (Dworkin 1977c, 22-28). And

110 CHAPTER III

because of this, it is concluded, we should enrich our concept of law by including those standards.

Here, I do not wish to reproduce that controversy or its connections with the po­sitivist thesis of the separation between law and morality. 10 I only want to point out that the notion of applicability can help clarify part of that controversy.

In those instances where norms belonging to the law stipulate the obligation of the courts to take into account certain standards of morality, we can say that those stan­dards are applicable, although they do not belong to the law. Those standards then re­ceive the same consideration as norms of foreign law or customary norms, which some­times also must be applied by the courts.

I will not go into the question of how such standards are identified either. Whe­ther they must be found through an empirical investigation of a society'S positive mo­rality or through some kind of epistemic access to critical morality is, of course, an im­portant question; but whatever those standards may be, I think it is clear that in some cases they are applicable.

Thus, when the Spanish Civil Code (in the third paragraph of art. 1271) stipu­lates that 'All services that are not against the law or against good customs can become an object of contract', the appeal to good customs undoubtedly is an appeal to a stan­dard of morality. Should we interpret this appeal saying that it incorporates into the law those moral rules that stipulate what good customs are? I think not, and in that sense the thesis of the separation of law and morality is fully upheld. Still, those moral rules must be applied by the courts when they judge the validity of contracts.

In general, we can conclude that references of legal norms to certain standards of morality declare these to be applicable, rather than incorporating them into legal sys­tems as members'! I

c) Irregular norms

Irregular norms, i. e., unconstitutional norms, are invalid since they do not satisfy the required systematic relationships with member norms. RI thus guarantees that no irre­gular norm ever belongs to any system of the sequence of a legal order. Elsewhere (Mo­reso 1994b), I have said that this notion of legal order shapes what can be called an optimal legal world, i. e., that possible world where all the autorities creating norms (looking like legal norms) do so on the grounds of a previous norm authorizing them to do so. In that world, the legislative power issues laws only in accordance with the Con­stitution, the executive power only issues regulations authorized by law, and the judicial

10 Which is what has led Raz to defend a form of positivism called exclusive positivism, according to which the law is based on certain social sources, in contrast to an alleged inclusive positivism said to hold that the law also consists in certain standards of morality. Cf. Raz 1994, ch. 9; Waluchow 1994.

11 In Chapter V, we will see some important consequences of that distinction. I already wish to point out that one of them is that, whatever the force of the thesis that legal measures must be interpreted in accord­ance with the authority's intention (as will be determined in that chapter), that thesis does not apply to the rules stipulating standards of morality, since obviously such standards are not created by any authority.

LEGAL DYNAMICS 111

power (or the administration) only issues sentences (or administrative resolutions) in ac­cordance with valid laws.

Now, in the real world, authorities often issue irregular norms, and it may even happen that a court of last instance (e. g., a Constitutional Court) decides that an irre­gular (and, in that sense, invalid) norm is valid. As Hart (1961, 138-144) reminds us, this is possible if we distinguish the finality of a decision from its infallibility: a decision of last instance is final, but not infallible. But if a Constitutional Court stipulates that a particular invalid norm is in accordance with the constitution, then the courts must ap­ply it.

Here, again, it is useful to distinguish validity from applicability (Bulygin 1991a, 267). Irregular norms are invalid, and the fact that a court says that they are valid does not make them so, even though under certain circumstances (by decision of a court) they may become applicable for certain organs.

d) Received Norms

In legal theory, sometimes there are questions about the status of the norms that existed before the enactment of a new constitution - elaborated independently of the previous­ly existing rules - and which are still used afterwards by the courts, and by legal ope­rators in general, for attributing obligations and rights to individuals.12 For instance, even after revolutionary changes resulting in a new political system, great parts of the previous law (e. g., usually most of private law) are still applied. Traditional opinion is that, therefore, received norms (as they have been called) belong to the legal systems arising from a new constitution.

However, none of the clauses of RI permits such a conclusion. Therefore, in Mo­resolNavarro 1996b we have proposed once again to use the distinction between mem­bership and applicability. Although received norms do not belong to any legal system of the order resulting from a new constitution, they are applicable according to those clauses of the new constitution that say so.

At least on one occasion, Kelsen suggested that there is a similarity between the case of received norms and the case of applicable norms of some foreign law. But since he did not distinguish between validity as membership and validity as applicability (cf. Bulygin 1990), he did not draw the same conclusions as I do here. Kelsen (1945, 244) presented the case as follows:

"The rule obliging the courts of a State to apply nonns of a foreign law to certain cases has the effect of in­corporating the nonns of the foreign law into the law of this State. Such a rule has the same character as the provision of anew, revolution-established constitution stating that some statutes valid under the old, revo­lution-abolished constitution should continue to be in force under the new constitution. The contents of these statutes remain the same, but the reason for their validity is changed."

If one accepts Kelsen's suggestion that the conflict norms of international private law have the same status as those norms which convert norms of previous legal orders into

12 Cf. Kelsen 1945, 244; Fuller 1969,142; Finnis 1973; Hart 1983, 362 ff.; MoresolNavarrolRedondo 1995.

112 CHAPTER III

received norms, then the proposal to regard them as invalid, but applicable gains plausi­bility.I3

5. Valid But Inapplicable Norms

The paradigmatic case of norms that are inapplicable although they belong to a certain legal system at a certain point in time is probably that of enacted norms that have not yet come into force. In this paragraph, however, I want to consider two additional in­stances: that of norms whose applicability is cancelled by some criterion when they come into conflict with other norms (I will call them cancelled norms), and that of norms which are irrelevant logical consequences of other norms (they will be called irrelevant derived norms).

a) Cancelled norms

When a norm NJ is inconsistent with another norm N2, and both of them belong to the same legal system, then for logical reasons the courts cannot ground their decision in both norms at the same time if they want to motivate the addressees' behaviour. There­fore, legal systems usually contain so-called antinomy-resolving criteria which in the case of such an inconsistency permit one, so to speak, to cancel the application of one of the norms.

Take, for example, two norms NJ and N2 , both belonging to the same legal sys­tem, of the following content:

NJ : Killing another person will be punished with 10 to 15 years in prison.

N2: Killing another, terminally ill person with her consent, in order to deliver her from unbearable pain will not be punished.

Obviously, one cannot apply both norms to a case of euthanasia, because they reach mutually inconsistent conclusions. If the legal system contains the lex specialis rule, we can say that the applicability - though not the membership - of NJ is cancelled, and only N2 will be applicable to the case.

If, as I propose, the lex specialis rule is seen as a criterion of applicability, then criteria of applicability provide legal systems with a rank order. Because of such crite­ria of applicability for the norms belonging to one system, legal systems are not only sets of norms containing all their logical consequences, but hierarchically ordered sets of norms (cf. Alchourr6nIMakinson 1981).

13 Sometimes the problem of the reception of norms has been linked to the problem of the identity of the state, on the assumption that the legal order of a state owes its identity to the political identity of the state (cf. Hart 1983, 309-342; Honore 1967; Raz 1979,78-102; Finnis 1973; Eekelaar 1973; Harris 1971). What­ever the status of that assumption, it should be noted that the problem of the reception of norms is indepen­dent of the question of state identity. A state can very well receive norms of another, disappeared state, as is the case, for instance, when a state is divided (for example, the norms the Czech Republic has received from former Czechoslovakia).

LEGAL DYNAMICS 113

Now, of course, there are many criteria of applicability in legal systems that en­able one to establish a rank order, and unfortunately, they don't always lead to one and the same order (cf. Bobbio 1964). But, in any case, they give us an additional reason for distinguishing between the norms that belong to a legal system and the norms that are applicable in it.

b) Irrelevant derived norms

In Chapter I, I have presented a notion of irrelevant logical consequence that enables us to divide the set of derived norms into the two subsets of relevant and irrelevant derived norms. I will now suggest that irrelevant derived norms are not adequate candidates for justifying judicial decisions. That means that, although they belong to a legal system, ir­relevant derived norms are not applicable (cf. Moreso 1996). Thus, for example, al­though the norm 'Buyers must pay the price of the merchandise' leads to the irrelevant derived norm 'Buyers must pay the price of the merchandise or dance a waltz', a judi­cial decision grounded in this irrelevant derived norm that would stipulate 'Ticius, hav­ing bought a house from Sempronia, must pay her the price of the house or dance a waltz' would certainly not be considered a justified judicial decision. And the reason why it would not be justified is because it is grounded in an irrelevant derived norm, i. e., an inapplicable norm.

As already suggested in Chapter I, this idea could lead us to redefine the notion of a normative system as a set of statements (among which at least one must be a norm) containing all their relevant logical consequences. However, because of a problem re­lated with derogation and the dynamics of legal systems, this possibility is not advisa­ble. For example, take the following normative system NS:

N]:p~q.

N 2 : q ~ Or

Norm N] can be understood as a conceptual rule (a definition, for instance, 'Persons over 18 are of age'), and norm N2 as a genuine prescription (,Persons of age must vote'). From these two norms, we can derive norm NJ: 'p ~ Or'. This is a relevant con­clusion. Now, suppose a norm-authority derogates norm N4 : 'p ~ O(r v s)' (which could be done with a derogating measure like 'All norms prescribing persons over 18 to vote or to register are derogated'). N4 is an irrelevant consequence of system NS. If ir­relevant derived norms would not belong to NS, the authority's act of derogation would be an unhappy act, i. e., it would not succeed in eliminating any norm from the system. Now, if all derived norms belong to NS, then the elimination of N4 also implies the eli­mination of NJ (a relevant logical consequence) and - but that is another problem­leaves open the question of whether N] or N2 ought to be eliminated too (the thesis of the logical indeterminacy of the system, cf. Alchourr6n1Bulygin 1981).

Or take another case: A father orders his children: 'You must do your homework every afternoon' ('Op'). Later, he decides to soften the strictness of the norm, and per­mits his children not to do their homework on any of their birthdays. That second norm

114 CHAPTER III

can be seen as a derogation of the derived norm 'Even if it is one of your birthdays, you still must do your homework' ('q ~ Op'). That is a norm derived from 'Op' and, there­fore, the act of derogating it implies the elimination of 'Op', which is replaced by an­other norm, like '-,q ~ Op' ('If it is not one of your birthdays, you must do your home­work'). Now, if we would not accept that irrelevant derived norms belong to the norma­tive system, the father's act of permission would not succeed in eliminating any norm, since the derogated norm is an irrelevant derived norm (' q ~ Op', where variable q can be replaced by any other salve validitate).

If only for this reason, we must define the notion of a normative system as a set of statements containing all their logical consequences. But in such a set, it is important to distinguish between relevant and irrelevant logical consequences.

Moreover, legal dynamics (changes in the law as a result of the enactment or de­rogation of norms) can be seen as a special case of the dynamics of knowledge. It has been pointed out that in processes of knowledge change, previously irrelevant conse­quences can become relevant as a result of a contraction process. It is, therefore, neces­sary to distinguish between the set of all beliefs and the subset of the relevant parts of that set. Contractions and revisions must be applied to the set of all beliefs, rather than to its relevant parts only (Schur:ziLambert 1994, 94 f.). If one accepts that derogation (a kind of contraction and revision of normative systems) is similar to a kind of belief change (Alchourr6n1GardenforslMakinson 1985), then the conclusion sketched here, according to which derogation must be applied to all the logical consequences of a nor­mative system, rather than to its relevant derived norms only, is an application of the more general thesis about the dynamics of knowledge.

6. Conclusions

The introduction of elements of legal dynamics brings with it the need to mold a more complex concept of a legal system than the one presented in Chapter I. Through certain authorized acts of the enactment and derogation of norms, new legal systems can be generated. The distinction between a legal order, as a sequence of systems of norms, and a legal system, as a set of norms that is closed under the notion of logical conse­quence, adequately reproduces this dynamic feature of the law.

RI expresses a notion of 'legal order' and of 'legal system' that can, at the same time, account for the problem of the identity of the law (guaranteed by the identification of the independent norms of the originating system and the criteria of legality and de­ductibility) and the problem of changes in the law (represented by the succession of sys­tems in the legal order).

RI also provides a notion of the validity of norms, understood as membership of the norms in some system of the sequence. That notion must be clearly distinguished from the notion of applicability. Although in one sense applicability depends on mem­bership, since norms are applicable or inapplicable according to other norms that belong to the legal systems, there can be valid norms that are inapplicable, as well as invalid norms that are applicable. The notion of applicability enables us to underscore two im­portant characteristics of legal systems:

LEGAL DYNAMICS 115

(i) Legal systems are open systems, i. e., they contain norms prescribing the ap­plication of other norms that do not belong to the system in question.

(ii) Legal systems are hierarchically ordered systems, i. e., they contain norms stipulating that certain norms are inapplicable to certain cases although they belong to the system in question.

IV. THE PRIMACY OF THE CONSTITUTION

1. The Constitution as Precommitment

In Chapter III, I have suggested that the originating legal system of a given legal order may consist in the norms of a constitution. Now, since the proposed definition of 'legal order' actually was a definition pattern identifying a sequence of normative systems, beginning with some first system, no further arguments for justifying the special prima­cy of that originating system were given. Primacy resulted, so to speak, from the mo­del's defining characteristics.

When we look at actually existing legal orders, however, it is legitimate to ques­tion the legitimacy of that primacy. The explication of primacy I will refer to is subject to some constraints; e. g., it only refers to an explication for legal orders arising in the context of a representative democracy. This constraint is due to the fact that my interest in the present investigation is constitutional interpretation in the context of representa­tive democracy, as is the case, e. g., of the Spanish Constitution of 1978. I do not ex­clude that similar strategies of explication can be found, mutatis mutandis, for constitu­tions existing in non-democratic political regimes. But that possibility will not be ana­lyzed here.

In fact, representative democracy, understood as a method of decision making by majority vote, contradicts the idea of the primacy of the constitution, understood as a constraint on the set of decisions a majority may take. This has been called the paradox of democracy (Elster 1979,94; also Holmes 1988,222) according to which 'Each gene­ration wants to be free to bind its successors, while not being bound by its prede­cessors' .

Jon Elster has spelled out an analogy between certain mechanisms of what he calls imperfect rationality - of an individual kind - and the case of the paradox of de­mocracy. His example is that of Ulysses and the Sirens. As everybody knows, Ulysses knows that his will is too weak for him to behave rationally should he hear the Sirens sing, and since he wants to satisfy his desire to hear them sing, he devises a strategy: he has his sailors tie him to the mast and makes them plug their ears with wax. Elster then continues (1979, 36):

"Ulysses was not fully rational. for a rational creature would not have to resort to this device; nor was he simply the passive and irrational vehicle for his changing wants and desires. for he was capable of achieving by indirect means the same end as a rational person could have realised in a direct manner. His predicament - being weak and knowing it - points to the need for a theory of imperfect rationality that has been all but negelected by philosophers and social scientists."

Ulysses's mechanism is a mechanism of precommitment, of binding oneself and "To bind oneself is to carry out a certain decision at time t1 in order to increase the proba­bility that one will carry out another decision at time lz" (Elster 1979,39).

Precommitment mechanisms are used by human beings in many situations of akrasia, for instance, in strategies to quit smoking (e. g., spending some time at a place where there are no cigarettes to be had), to lose weight (e. g., not having in the house, or

118 CHAPTER IV

at the place where one spends most of one's time, the food one craves), etc. To bind oneself in that kind of situations is to exclude certain future decisions, in order to up­hold a past decision that is valued positively.

The analogy with mechanisms of collective decision consists in the suggestion that precommitment, i. e., eliminating the possibility of taking certain decisions in the future in order to preserve especially valuable contents, is applicable to collective deci­sions too. l

We can thus understand the distinction common in political theory, between con­stituent power and constituted power. To say it once again with Elster's words:

"A crucial notion in this connection is the function of the constituent assembly that lays down the ground rules to be followed by all later generations. Only the constituent assembly really is a political actor, in the strong sense of la politique politisante; all later generations are restricted to la politique politisee, or the day-to-day enactment of the ground rules. The nation can bind 'itself (a controversial notion) through the constituent assembly, by en­trusting certain powers of decision to the judiciary branch, by requiring that the ground rules can only be changed by a two-thirds or a three-fourths majority, and so on." (Elster 1979, 93)

The idea of precommitment is adequately expressed in the ideal of constitutional demo­cracy. Certain things (fundamental rights, the territorial structure of the state, the divi­sion of powers, etc.) are beyond the reach of the ordinary political agenda and, there­fore, of public and legislative debate, i. e., of majority rule, which applies only to the political agenda of other matters. 2

Note that this protection works properly with rigid constitutions, whereas flexi­ble constitutions only have a procedural kind of precommitment mechanism (cf. Guas­tini 1991, 14-17).

The mechanism of precommitment is not a conceptual explication of the prima­cy of the originating system; but it is what we can call a contextual explication of that primacy. It is one of the circumstances under which one can speak of the primacy of the constitution, just as, for instance, moderate scarcity is one of what can be called the cir­cumstances of justice (Hume 1740, III.II.ii; Rawls 1971, 126 f.). These circumstances are not part of a conceptual explication of the notion of justice, but they are part of the context in which such an explication can be given.

In the present investigation, which is basically a conceptual analysis, I will say little more about this point. But its importance is worth emphasizing, since it not only determines the context in which one can analyze the primacy of the constitution, but also is a circumstance of constitutional interpretation.

1 Elster (1979, 88 f.) mentions various examples from the institutions of classical Athens.

2 Apparently, this is a widely accepted idea in contemporary political theory. Cf. the interesting presenta­tions of it, e. g., in Nino (1992, 70 f.); Rawls (1993, 151 ff.); Ackerman (1984, 1988, 1989). Caracciolo (1991) has used a similar idea, applied to individual rights, to account for the so-called liberal paradox. Referring to the question of the conflict between the protection of rights and democracy, as a beautiful - al· though, in view of the subsequent historical development, unfortunate - historical example of precommit­ment, Garz6n Valdes (1994, 128) reminds us of the oath foreseen in the Constitution of Cadiz: 'And if with respect to what I have sworn, or part of it, I should do the contrary, I shall not be obeyed; instead, that in which it is contradictory shall be null and void'.

THE PRIMACY OF THE CONSTITUTION 119

2. Possible Constitutional Worlds

The originating system of a legal order determines the set of sequences of constitution­ally adequate possible legal systems. Not only must all norms be created or eliminated in accordance with the norms of the first system or with other norms created in accord­ance with that first system; moreover, no norm contradicting a norm of the constitution, i. e., a norm of the originating system or a norm of the constitution that has been reformed in one of the subsequent systems, can be validly created. In that sense, it should be noted that the term 'constitution' is ambiguous: it may refer only to the origi­nal constitution - the originating system - or to the on-going constitution, i. e., the constitution that is in force at any given point in time. For simplicity's sake, in this pre­sentation I will assume that the constitution is never reformed and that, therefore, 'con­stitution' always means the original one, i. e., the originating system of the legal order.3

Another way of presenting the same idea is the following: a special kind of legal propositions are those expressed in statements like the following:

(1) Constitutionally, x ought to (may not, may) do I/J.

The truth of (1) determines the truth of

(2) Legally, x ought to (may not, may) do I/J.

And the falsity of (1) determines the falsity of (2). Now, what happens if (1) lacks a truth-value? Then we know nothing about the truth-value of (2). (1) can lack a truth­value, as we know, because the constitution is silent, or because it speaks in many voi­ces, or because it speaks in an uncertain voice. In all three cases, one would need to dis­cover what other, inferior rules say, in order to assess the truth-value of legal proposi­tions that may themselves be true, false, or lack a truth-value. It is in that sense that we can understand the idea that the law has a hierarchy. The content of the norms of each system determines the possible content of the norms of the next system.

Thus, constitutional norms constitute what Kelsen (1960, 346-356) has called a framework within which there may be several possibilities. Starting from the originating system, one can generate various constitutionally adequate sequences of legal systems. The constitution does not determine one single linear sequence of legal systems:

Diagram 1

{So} {S t} {S2} {S3} 1 __ 1 __ 1 __ 1 __ -

{Sn}, ___ I

3 This also serves to avoid the question of whether or not the clauses of constitutional reform may themselves be reformed, following the procedure stipulated in them. I have treated this question in Moreso (1991). Cf. Schmitt 1928; Ross 1958, 1969; Hart 1983b; Raz 1972b; Hoerster 1972; Guastini 1982; Gui­bourg 1983; Bulygin 1984; Nino 1985; Suber 1990; Alarcon Cabrera 1995.

120 CHAPTER IV

but rather a tree-shaped sequence that can be represented as follows:

Diagram 2

I

\

{SI}

{S2}­I

I \ I {S2a}-

\ {S'2}-\ I {S'I}

\ {S'2a} -

Diagram 2 should be read as saying that starting with any system not only two, but many (possibly, an infinite number of) systems can be generated. It should also be clear that not any system whatsoever can belong to a sequence. The originating system con­strains the scope of systems that can be generated, but every choice (every act of nor­mative creation or elimination) constrains even more the set of sequences that can be generated.

Diagram 2 represents the set of possible legal orders starting with a given origi­nating system. However, if we take a present or past time t as our point of reference, then we can reconstruct the sequence of systems from the initial time to - correspond­ing to system So - up to t as a linear sequence.4 Thus, Diagram 1 represents what can be called the real legal order, whereas Diagram 2 represents only possible legal orders.

Adopting an idea from philosophical scholarship on possible worlds,S we can distinguish the real or actual legal world from possible legal worlds, and, moreover, de­fine a relation between worlds, the relation of accessibility, according to which, follow­ing Diagram 2, {SJ is accessible from {SI}' but not from {S'I}'

Actually, the idea is somewhat more complicated. Systems are not worlds, but rather sets of norms closed under the relation of logical consequence. The worlds cor-

4 But even then, there is one reservation. Sometimes the result of an act of derogation does not determine the set of nonns in which the elimination occurs and, therefore, it does not detennine the resulting set of nonns either. In that case too, one cannot reconstruct the legal order as a linear sequence of legal systems. We then have what is called the logical indeterminacy of legal systems (cf. Alchourr6n1Bulygin 1979, 1981). But I will not go into this question here.

S In view of the general approach taken in the present investigation, that analogy should not force us to ac­cept a realist ontology of possible worlds, as Lewis claims (1973, 88); one can think of an anti-realist ap­proach to possible worlds, a way of speaking about how we could see the world if it were different from what it is now (cf. Kripke 1980).

THE PRIMACY OF THE CONSTITUTION 121

responding to each system are those possible worlds that make the system effective. Here, we have nothing like the actual world, since that would not be likely to be a legal world (it would have to be a world that makes all the norms of a given system ef­fective). Now, the (possible) world serving as the point of reference is the world that makes the originating system of a given order effective. The relation of accessibility is constructed starting from that system. I will call worlds that can be generated from the constitution, i. e., from the originating system, constitutionally possible worlds.

The most obvious case of a constitutionally impossible world is one correspond­ing to a system that contains a norm which contradicts some constitutional norm.6 Thus, if the constitution contains a norm prohibiting the application of the death penalty as a penal sanction (like art. 15 of the Spanish Constitution),7 then any system that contains a norm prescribing or permitting the application of the death penalty corresponds to a constitutionally impossible world, i. e., that norm does not belong to any system of the sequence originating in that constitution.

That does not only exclude systems containing norms that are inconsistent with certain constitutional norms. It also excludes systems containing norms with a certain constellation of generic cases - it excludes certain combinations of properties. The next section will be dedicated to the explanation of that idea.

3. The Thesis of Constitutional Accessibility

I will begin with a case: the Spanish Constitutional Court, in STC 22211992, was given opportunity to rule on the constitutionality of art. 58.1 of the Spanish Landlord and Tenant Act (Ley de Arrendamientos Urbanos) valid at the time (Decree 410411964, of December 24). The article in question stipulated the right to subrogation in a lease of the lessee's surviving spouse 'with whom [the deceased tenant] had habitually cohabi­ted in the rented property'. However, this right was denied to the surviving partner of a de facto couple. The question was raised whether that regulation, by treating surviving spouses differently from surviving de facto partners, violated the principle of equality. The Court's answer was afflrmative, and it therefore proceeded to declare the rule under scrutiny unconstitutional insofar as it denied the benefit of subrogation in a lease to persons having cohabited in a stable, quasi-marital relationship with a deceased lessee.8

It should be noted that the Spanish Constitution does not seem to require that parliament concede the right of subrogation in such cases.9 That means that a legal

6 The notion of nonnative contradiction introduced in Chapter I will do.

7 With the exception of what martial law stipulates for times of war.

8 Cf. on this topic the instructive work of Ferreres 1994. Here, I will not discuss the ruling of the Constitu­tional Court (STC 18411990,20/1991,3011991,3111991,3511991,3811991,77/1991 and 2911992) accord­ing to which the decision of parliament to concede a widower's pension to the surviving spouse of a de­ceased person and to deny it to a surviving de facto partner is constitutional.

9 One could argue that this is required by the right to a decent home (art. 47), but undoubtedly such a right could be guaranteed in an indirect way. Parliament could decide that it is a desirable political objective not

122 CHAPTER IV

world in which there is no right of subrogation is constitutionally possible. What does not seem possible, according to the court's ruling, is a legal world in which there are different normative consequences for the two generic cases. Thus, what is excluded is a certain combination of relevant properties of the case.

Alchourr6n and Bulygin (1971, ch. VI) analyze several characteristics of uni­verses of cases which will be useful for our purpose.

In the universe of discourse (UD) constituted by those cases in which the lessee of a home dies, and in relation to the universe of action (UA) constituted by the single act of subrogation in the lease upon the death of the lessee, only two properties are con­sidered relevantlO by article 58.1 of the Spanish Landlord and Tenant Act (i. e., the cor­responding universe of properties - UP - has only two elements): marriage (M), and cohabitation more uxorio (C). Starting from this universe of relevant properties, a rele­vant universe of cases (UC) is generated. ll These (generic) cases must be characterized in such a way that certain individual cases can be instances of them. Therefore, the combinations of properties defining some generic case may be neither tautological nor contradictory. For this, the chosen properties (M and C, in our example) must fulfill certain conditions:

a) The elements of the universe of properties must be logically independent. Two properties are logically independent if, and only if, the presence of one of them in an individual case is compatible with the presence as well as with the absence of the other one in the same individual case.

In our example, both properties satisfy the requirement. Sempronia (the deceased lessee) may have been married to Ticius and may have lived with him more uxorio; she may have been married to him, and not have lived with him more uxorio; she may not have been married to him and have lived with him more uxorio; and finally, she may not have been married to him and not have lived with him more uxorio.

b) The elements of the universe of properties must be logically independent from the behaviours in the universe of actions.

In the case of our example, marriage and cohabitation are clearly independent from the faculty of subrogation in a lease.

c) Finally, the universe of properties and the universe of discourse must be cor­responding universes: each one of the elements (individual cases) of the universe of dis­course must be able to have each one of the properties of the universe of properties.

It is easily seen that this prerequisite too is fulfilled by the properties of our ex­ample.

to constrain the rental market too much, and it could also take other measures which guarantee that everyone has access to a decent home (through subsidies in certain cases, or in other ways).

10 A property is relevant for a given case C of a universe of cases UC in relation to a given normative sys­tem S and a given universe of actions UA if, and only if, case C and its complementary case relative to P (i. e., the case in which, all other properties being equal, property P is absent) in UC have a different normative status in relation to S and UA. Cf. AlchourronlBulygin 1971, 101 f.

II This is not the only way of generating a universe of cases (cf. AlchourronlBulygin 1971, 26 f.); but it will be used here because it is the most common one.

THE PRIMACY OF THE CONSTITUTION 123

Now, we can define the notion of an elementary case as a case characterized by the conjunction of all the properties of UP or their negations. We can then construct a universe of elementary cases from those properties in such a way that it is a partition (or division), i. e., that it satisfies the following three conditions: i) that none of the pro­perties defines an empty class; ii) that the classes defined by the properties are logically exclusive (that no individual case is an instance of more than one generic case); and iii) that the combinations of properties are logically disjunctive (that every individual case belongs to some generic case).

The UP of our example is {M, C}, and the universe of elementary cases is crea­ted as follows: if n is the number of properties (n = 2, in our example), 2n is the number of elementary cases (in our case, 22 = 4). The following are the four cases of our ex­ample:

Table 1

1) M /\ C 2) M /\-,C

3) --,M /\ C 4) --,M /\ -,C

M

+ +

C

+

+

In accordance with the Landlord and Tenant Act mentioned above, case 1) was correla­ted with the following normative solution: subrogation in the lease is facultative (Fs); cases 2), 3), and 4) were correlated with the solution that prohibited subrogation in the lease (Phs). That means:

Table 2

M C

1) M/\ C + + Fs 2) M /\-,C + Phs 3) --,M /\ C + Phs 4) --,M /\ -,C Phs

However, this is not the only universe of cases that can be constructed with the two pro­perties I have used here. Any number of properties and their negations can give rise to a division. Thus, we get different universes of cases, depending on the properties taken into account. In our case, we could take into account only one of the properties and its negation, or both. We, thus, get three possible universes of cases. First, the universe of cases generated from property M and its negation: UC(M). Second, the universe of cases generated from property C and its negation UC(C). And finally, the universe of cases generated from both properties and their negations: UC(M, C).

124 CHAPTER IV

Following Alchourr6n and Bulygin (1971, 96), I will call the finite number n of properties of a given UP the width of that universe of properties. Since every universe of cases is a function of a UP, n not only measures the width of the UP, but also the level of the corresponding uc. Following our example, it is interesting to note the rela­tionships between one of the UCs of levell, UC(C) , and one of the UCs of level 2, UC(M, C). We can show this with the following diagram:

Diagram 3

UC(M, C) M,C ~,C M,-,C ~,-,C

UC(C) C -,C

Diagram 3 shows the relationships of logical implications between the cases of the uni­verses of cases UC(M, C) and UC(C). UC(M, C), we can say, is afiner universe of cases than UC(C). Generally,

"A division D, is said to be finer than another division D, if and only if every element of D, logically implies some element of D, and at least one element of D, is not implied by any element of D,." (Alchourr6n1Bulygin 1971,97)

The relation of being finer follows the direction of the arrow, in the sense that the UC the arrow points away from corresponds to a finer UP than the UC the arrow points to. Alchourr6n and Bulygin (1971, 98-10 1) also habe shown certain logical relationships between universes of cases, depending on the fineness of the corresponding UPs. Thus, if a given normative system a is complete in relation to some UCj , then it is also com­plete in relation to all the UCs that are finer than UCj • In contrast, that a normative sys­tem a is complete in relation to a UCj does not imply that it is also complete in relation to UCs that are less fine than UCj •

On the other hand, if a normative system a is consistent in a UCj , then it is also consistent in UCs less fine than UCj • And if a normative system is inconsistent in a UCj ,

then it is also inconsistent in UCs finer than UCj •

Thus, while completeness is inherited upwards, from less fine to finer universes of cases, consistency is inherited downwards, from finer to less fine universes of cases.

Returning to our example, in the universe of cases UC(M, C), which is the one taken into account by the Landlord and Tenant Act, the relevant properties are proper­ties M and C. In relation to M as well as in relation to C, there are cases in which those

THE PRIMACY OF THE CONSTITUTION 125

properties are relevant: those cases and their complementary cases have different nor­mative status.

The proposition identifying the set of relevant properties in relation to a given normative system and a given universe of actions we can call (following Alchourr6n1 Bulygin 1971, 103) the thesis of relevance of that system and that universe of actions. 12

For the system represented in Table 2, the thesis of relevance identifies the following set of properties: {M, C}.

One way of understanding STC 222/1992 is as holding that the principle of equality requires the UC in relation to subrogation to be less fine than the one taken into account by the legislator in the Landlord and Tenant Act. Thus, although parliament may stipulate the normative solution it considers adequate for the fact of more uxorio cohabitation concerning the right of subrogation, what it cannot do (i. e., what it is in­competent to do, what is a constitutionally impossible world) is to make marriage a privileged property with respect to the right to subrogation.13

Therefore, the declaration of unconstitutionality of art. 58.1 of the former Land­lord and Tenant Act must be understood as a declaration to the effect that the universe of cases UC(M, C) is too fine, and that this is excluded by the Constitution. The less fine universe of cases UC(C) must be used instead. However, since completeness is not inherited down from finer to less fine cases (moreover, a consistent and complete sys­tem in relation to some relevant UC is necessarily incomplete in relation to a less fine UC; cf. Alchourr6n1Bulygin 1971, 104), there is a gap in UC(C). Case...,C (cf. Diagram 3) seems to be regulated with the normative solution 'subrogation is prohibited' (Phs) since M is irrelevant in case ...,C, i. e., case (M, ...,C) and case (-.M, ...,C) have the same normative solution. But case C now appears with a normative gap. M is relevant in rela­tion to C after all, since its presence or absence determines whether or not subrogation is permitted in UC(M, C). Therefore, in UC(C), case C has no normative solution. That is why the Constitutional Court goes on to fill the gap, and sentence STC 22211992 is a law-creating sentence: it correlates the normative solution 'subrogation is permitted' with the case of a deceased lessee and his or her de facto partner having cohabited more uxorio. 14

Thus, the constitution determines not only that certain normative solutions are excluded; it also stipulates that certain universes of cases (of a certain fineness) are ex­cluded.

12 It should be clear that the meaning of 'relevance' in this chapter has nothing to do with the concept ana­lyzed in Chapter I.

13 Legal Foundation No.6 says it in the following words: "The measure is so radical as to impair unreason­ably the autonomy of the will of a man and woman who decide to cohabit 'more uxorio'." In fact, the new Landlord and Tenant Act of 1994 (in its art. 16.1) concedes the right of subrogation in case of the lessee's death to the spouse that cohabited with the tenant as well as to "the person that has permanently cohabited with the tenant in a relationship analogous to that of a spouse .. , ",

14 In Italian constitutional doctrine, this kind of sentences is called 'sentenze additive'. Cf. Guastini 1993, 313. For a similar idea in the context of an analysis of this type of constitutional cases, cf. Guastini 1995, 157-176, and Parodi 1996,

126 CHAPTER IV

On the other hand, not only can the universes of cases constructed by the legis­lator be too fine, i. e., can the legislator have made too many distinctions; the universes of cases constructed by legislators can also be not fine enough: the legislator can have made too few distinctions. In order to analyze this new case, we must introduce still an­other sense in which one can say that a property is relevant. We can distinguish the de­scriptive sense of relevance from its prescriptive sense (Alchourr6n1Bulygin 1971, 103-106). To say that a property is relevant in the descriptive sense is to say that a certain state of affairs exists, that a certain case and its complementary have different norma­tive status. To say that a property is relevant in the prescriptive sense is to say that a cer­tain state of affairs ought to exist, that a case and its complementary ought to have dif­ferent normative status.

Just as the proposition identifying the set of relevant properties in relation to a universe of actions was called the thesis of relevance, we can call the proposition iden­tifying the set of properties that ought to be relevant for a universe of actions the hypo­thesis of relevance (Alchourr6nlBulygin 1971, 103). Since we know that completeness is inherited from finer to less fine universes of cases, if a relevant universe of properties is complete then all universes finer than this will also be complete. However, jurists also speak of gaps when the legislator has not taken into acount some property that should be relevant according to some hypothesis of relevance, i. e., when they think that a given solution is inadequate because it has not taken into account a distinction that should have been taken into account.

It should be noted that the hypothesis of relevance presupposes that one adopt some axiological criterion for establishing the relevance of certain properties. Here, I will not go into the question of whether such criteria are subjective or objective; in any case, constitutions usually postulate certain values from which one can extract criteria which make some theses of relevance used by the legislator inadequate.

In my view, this is the case in the Spanish Constitution (cf. Peces-Barba 1984, 1993) with the value of personal dignity (art. 10), the free development of the individu­al (art. 10), and the prohibition of inhuman and humiliating treatment (art. 15), in rela­tion to the regulation the Criminal Code in force until May 15, 1996 gave to cases of euthanasia in its art. 411. For that Criminal Code, the victim's consent given in the case of a terminal illness or severe permanent and unbearable suffering was irrelevant. Moreover, so-called assistance in the execution of suicide was punished with the same sanction as murder, which means that in cases of the killing of another person consent was always irrelevant.

Now, one could say that our constitutional values require the legislator in those cases to take consent into account. 15 If that requirement were accepted, then the UC of the old criminal code would not be fine enough in relation to the universe of discourse of causing another person's death. The Constitution would require a finer UC, like that

IS In fact, the new Criminal Code (Act 10/1995) punishes assistance in the execution of suicide with a lesser sentence (from six to ten years, art. 143.3) than murder (from ten to twenty years, arts. 138 and 139), and with a still lesser one (one or two degrees less than that of art 143.3; art. 143.4) someone "who causes ... the death of another upon that person's serious and unequivocal explicit request, in case the victim suffers of a severe illness that would inevitably lead to death or that produces permanent and unbearable severe pain ... ".

THE PRIMACY OF THE CONSTITUTION 127

of the Criminal Code recently enacted (cf. previous note). Had the Constitutional Court had the opportunity to rule on that question, and had it adopted the preceding reasoning, then it would have encountered a case of a legislative thesis of relevance that was not fine enough. Using a constitutionally adequate hypothesis of relevance, it would have had to go on to create law, giving a new normative solution (moderating or eliminating the punishment) to the case of euthanasia.

Although the legislator of the old criminal code regulated the case of euthanasia, from the constitutional point of view it regulated it inadequately. Therefore, here we do not face a case similar to that of the previous example, where we had a normative gap for the less fine, but constitutionally adequate UC; instead, the UC that is finer than the one used by the legislator does have a solution for the case, but that is an inadequate solution.

In that case, Alchourr6n and Bulygin (1971, 106-110) speak of an axiological gap, which they define as follows:

• .A case C, of a Universe of cases U~ is an axiological gap of the nonnative system IX in relation to a UA, = Of. the case CI is correlated by IX with a maximal solution of the USmax (corresponding to UA;) and there is a prop­

erty p such that p ought to be relevant for C, (according to a certain hypothesis of relevance) and p is irrelevant for IX in relation to UA,."

That means that a universe of cases UCI can be constitutionally inadequate because it is generated by a thesis of relevance that is too fine, or because it is generated by a thesis of relevance that is not fine enough. In the first case, there is a normative gap in the UC that is less fine than UCi' but constitutionally adequate. In the second case, there is an axiological gap in the UC that is finer than, but constitutionally adequate.

These two situations correspond to those described by Alchourr6n and Bulygin (1971, 109) as situations in which the thesis of relevance and the hypothesis of relevance are different, but comparable:

.. Situation I

The set of properties identified by the thesis of relevance is properly included in that of the hypothesis of re­levance. This means that there is at least one property such that it ought to be relevant but is not relevant for the system in question. It also implies that the Universe of Cases corresponding to the hypothesis of rele­vance ... is finer than the UCR. In this situation there is at least one case of axiological gap."

This, we have seen in the euthanasia example examined above .

.. Situation II

The set of properties identified by the hypothesis of relevance is included in that of the thesis. This means that there is at least one relevant property which ought not to be so and that the UCR is finer than the [UC corresponding to the hypothesis of relevance]. Here the legislator has made too many distinctions. The con­sequence is that there are some solutions which are unjust (according to the hypothesis of relevance). but there are no axiological gaps."

128 CHAPTER IV

That was the case in the example of the right of subrogation in the Landlord and Tenant Act.

Thus, we see that the constitution determines the width of the universes of cases the legislator can generate. For certain normative solutions, the constitution determines certain hypotheses of relevance. If the theses of relevance used by the legislator differ from the constitutionally adequate hypotheses, the legislator gives rise to constitutionally impossi­ble worlds.

The proposition identifying the set of properties that are constitutionally relevant in relation to some legal system LS for a given universe of actions VA; or, what amounts to the same, the proposition identifying the set of properties that ought to be relevant for the legislator in relation to that LS for that VA; I will call the thesis of constitutional accessi­bility of the LS for the VA;.

Satisfaction of the thesis of constitutional accessibility, however, is only a neces­sary condition of the constitutional adequacy of a legal system. A legal system may satisfy the thesis of constitutional accessibility and still be constitutionally impossible, since even if all the properties that ought to be relevant really are relevant, it may not correlate all cases with the constitutionally required solutions (as was the case with the death penalty mentioned in the previous section). In that case, the resulting world is constitutionally im­possible because there is a normative antinomy between the normative solutions foreseen by the constitution and the normative solutions stipulated by the subconstitutional legisla­tor.

Alchourron and Bulygin (1971, 106) generalize the two cases as follows:

"We may generalize these observations saying: a nonnative system may be regarded as axiologically inadequate (bad or unjust) for a UA for two different reasons: (I) because it does not satisfy the hypothesis of relevance (what is wrong then is the selection of the cases); (2) because it fails to correlate the cases with the right solution (what is wrong then is the solution given to the cases which have been selected correctly). As is obvious, these two defects are not incompatible: a system may solve wrongly cases wrongly selected."

Adapting these considerations to the relationships between the constitution and subcon­stitutionallegislation, we can say that a normative system can be constitutionally inade­quate in relation to a given VA for two reasons: 1) because it is not in accordance with the thesis of constitutional accessibility; and 2) because the normative solutions it as­signs to certain cases are inconsistent with the solutions determined by the constitution.

Therefore, we can say:

A world is constitutionally impossible in relation to some UA of that world if and only if: I) either its rele­vant universe of cases does not accord with the thesis of constitutional accessibility; or 2) some of its nonna­tive solutions are inconsistent with the solutions foreseen in the constitution.

4. Conclusions

(1) The constitution, i. e., the originating system of a legal order, determines a set of sequences of legal systems. It thus divides the possible legal systems into two sets: constitutionally possible legal systems, and constitutionally impossible or excluded legal systems.

THE PRIMACY OF THE CONSTITUTION 129

(2) A contextual explication of that special primacy of the originating system can be found, for constitutional democracies, in precommitment as a mechanism of collective rationality. The authority creating the constitution 'binds the hands' of subsequent authori­ties. The constitutional authority is like a sober person binding the future decisions of that same person when she is drunk (Holmes 1988, 197).

(3) It should be noted that legal worlds following the original constitution can not only be excluded for stipulating normative solutions for some cases that are inconsistent with the originating system, but also for choosing cases in an inadequate way, i. e., for not satisfying what I have called the thesis of constitutional accessibility.

V. CONSTITUTIONAL INTERPRETATION

1. Introduction

In the previous four chapters, I have attempted to present a view of the law according to which the constitution plays a decisive role in the determination of the truth-value of le­gal propositions. In particular, it may be the case that the norm expressed by a provision issued by a subconstitutional authority, and which stipulates that certain persons have certain rights or obligations, does not belong to any of the legal systems of that legal or­der, because, e. g., it has been issued by an incompetent authority. Thus, norms that have been enacted but whose origin cannot be traced back to other norms do not belong to any of the systems of the respective legal order. In addition to this, constitutions also determine a range of matters for which subconstitutional authorities are definitively in­competent: these are all matters regulated in a way that is inconsistent with the constitu­tion, as well as those that do not fit to what I have earlier called the thesis of constitu­tional accessibility.

But, how is one to determine the meaning of constitutional provisions? In fact, the extensional identification of independent norms is not an easy task. Usually, there is widespread agreement about the text that contains those norms, although sometimes even the identification of the text is questioned in some aspect or other. Sometimes there are several texts with slight differences that may create problems for the determi­nation of the authentic text. 1 Here, I will not go into these questions concerning what (following Moore 1989, 115) we can call the syntactic constitution, that is, the constitu­tion understood as a set of uninterpreted linguistic symbols.2 What interests me, instead, is the question of the meaning of those texts; and since throughout this book I have sub­scribed to a position according to which norms are abstract entities, what I am interested in is to determine what the norms of the constitution are, i. e., the question of the semantic constitution as a set of meanings. Only by considering the meaning of the con­stitutional text, the idea of the primacy of the constitution can be understood. This pri­macy is of a semantic nature, because it would be rather strange to say that a text of un­interpreted symbols has such primacy; the idea of syntactic primacy is unintelligible.3

Thus, the question of constitutional interpretation, of conferring a certain mean­ing to certain constitutional texts, can be approached from the point of view of asking

1 For two interesting contributions to such problems with texts cf. Amar (1987) concerning the Constitution of the United States, and Vallejo (1993-1994) concerning the Spanish Constitution of 1869.

2 In Spain, Hernandez Marin (1984, 11 ff., 1989,49 f.; cf. also Braybrooke 1989, 289) has defended a radi­cally inscriptionist conception of legal nonns, based on a nominalist philosophy of language. But even if one concedes that legal nonns are nothing but texts - token-sentences - what matters is how they should be interpreted. Unfortunately, whatever the result of that controversy about the ontology of language, there will still be the more important question of how to construct a theory of constitutional interpretation.

3 One could say that meta-languages have syntactic priority over their object-languages. But in order to dis­tinguish between language and meta-language, we need a minimum of interpretive activity which renders even that weak priority questionable.

132 CHAPTER V

about the truth-conditions of a special kind of legal propositions which I will call con­stitutional propositions. These are propositions expressed in sentences like:

(CS) Constitutionally, Fs ought to (must not, may) do Y.

Everything said in Chapter II about legal propositions applies here too. Now, in that chapter, I had assumed that the existence of norms guaranteed the truth or falsehood of propositions expressed by sentences like (CS). Moreover, when there are no such norms, it was said, such propositions have no truth-value. Now, in contrast, we must ask how certain normative texts are correlated with a certain meaning, since it is in relation with the meaning of normative texts that we can establish the truth-conditions of consti­tutional propositions.

Sentences like (CS) can be analyzed as

(CSj ) The normative consequences of the constitution contain a norm which sti­pulates: 'Constitutionally, Fs ought to (must not, may) do Y'.

The truth-conditions of that statement depend on the attribution of meaning to certain constitutional texts. They depend on the truth of interpretive statements like

(IS) Constitutional text Tmeans ...

But how does one attribute meaning to a constitutional text? That is a general question of legal philosophy which as such is, I think, of theoretical interest. But it is also of great practical interest, since many contemporary legal systems have courts whose func­tion it is precisely to decide whether or not certain provisions issued by subconstitu­tional authorities (mainly, the legislative power) are in accord with the constitution. Ob­viously, in order to fulfil this function the constitutional texts must necessarily be given a meaning. In view of the importance of the questions treated by such courts, theories of constitutional interpretation are usually seen to have great political relevance. One of the questions to be explored in this chapter is to what extent a theory of constitutional interpretation can be independent of political considerations.

Theories of constitutional interpretation do, in fact, presuppose some theory about the nature of the law. Thus, they make it necessary to go back to some of the considerations presented at the beginning of Chapter II. A realist conception of the law will lead to another theory of constitutional interpretation than a constructivist concep­tion. Besides, we must distinguish a particularly radical form of constructivism, ac­cording to which constitutional propositions referring to generic cases never have a truth-value. According to this position, which I will call skepticism, there is no way in which we could give unequivocal meaning to constitutional texts, and the constitution is nothing but what the courts say that it is.

In an illuminating article, Hart (1983a) describes the theory of law in the United States as being situated between two extremes that can be understood as a version of realism and a version of skepticism, respectively.

In Hart's own words:

CONSTITUTIONAL INTERPRET A TION 133

"I have portrayed American jusrisprudence as beset by two extremes, the Nightmare and the Noble Dream: the view that judges always make and never find the law they impose on litigants, and the opposed view that they never make it. Like any other nightmare and any other dream, these two are, in my view, illusions, though they have much of value to teach the jurist in his waking hours. The truth, perhaps unexciting, is that sometimes judges do one and sometimes the other. It is not of course a matter of indifference but of very great importance which they do and when and how they do it. That is a topic for another occasion." (Hart 1983a, 144)

Following Hart, I will call the conception that takes a realist metaphysical stand towards the interpretation of the constitution the 'Noble Dream'. This will enable me sometimes to avoid the word 'realism' which is so confusing in this context because what is known as '(American) legal realism' is, from the point of view adopted here, a skeptical posi­tion. The skeptical conception of constitutional interpretation, I will call the 'Night­mare'. The intermediate position, a kind of moderate constructivism which I will try to defend, can then be called the 'Vigil'.

To each of these positions, we can, thus, ascribe four theses about the status of constitutional propositions. These theses will later help me to structure the discussion.

The theses of the Noble Dream are the following:

(1) Metaphysical thesis: Independently of our knowledge, there is a legal world which makes our constitutional propositions true or false. (2) Semantic thesis: The meaning expressed by constitutional statements is de­termined by their truth conditions, independently of how we may be able to veri­fy them. (3) Logical thesis: The principle of bivalence is accepted. All constitutional pro­positions are either true or false. (4) Legal thesis: There always is exactly one right answer in constitutional cases. Courts always apply and never create law. Therefore, courts can err in establish­ing the constitutional rights and duties of citizens.

The theses that can be ascribed to the conception of the Nightmare are:

(la) Metaphysical thesis: Beyond our capacity for knowing the law, as construc­ted by human beings, there is no legal world that could make our constitutional propositions true or false. (2a) Semantic thesis: One cannot ascribe meaning to pure constitutional state­ments (i. e., statements the truth of which depends exclusively on the existence of certain constitutional norms), because it is impossible to give meaning to the norm-formulations in a constitution. (3a) Logical thesis: Pure constitutional propositions have no truth-value since they refer to the alleged existence of the meanings given to constitutional norm formulations, and there are no such meanings (norms). (4a) Legal thesis: There never is a right answer in constitutional cases. Courts always create and never apply law. Therefore, courts cannot err in establishing the constitutional rights and duties of citizens.

134 CHAPTER V

Finally, the theses that can be ascribed to the Vigil are:

(lb) Metaphysical thesis: Beyond our capacity for knowing the law, as construc­ted by human beings, there is no legal world that could make constitutional pro­positions true or false. (2b) Semantic thesis: The meaning of pure constitutional statements is deter­mined by their conditions of assertability, that is by the possibility to show that certain consequences follow from the original legal system, i. e. from the consti­tution. (3b) Logical thesis: Not all constitutional propositions are true or false. (4b) Legal thesis: Sometimes there is a right answer in a constitutional case, and sometimes there isn't. On some occasions, courts create law, on others they ap­ply it. Therefore, in those cases where they apply previously existing law, courts can err in establishing the constitutional rights and duties of citizens.

In hard cases, the controversy between those three conceptions becomes very important. For Noble Dream conceptions, cases can only be hard for epistemic reasons; under ideal epistemic circumstances, we would find solutions for all constitutional hard cases. For Nightmare conceptions, in contrast, all cases are hard cases. And for the Vigil, even un­der ideal epistemic circumstances there can be hard cases, though not all cases are hard.

Obviously, there are several ways of defending each one of those positions. The later sections of this chapter will try precisely to present and discuss some of the ways in which those conceptions are expressed. What I have called the legal thesis - in its forms of (4), (4a), and (4b) - will help us distinguish between different conceptions, since not all the conceptions I will present endorse the four theses just presented. In fact, in what follows, one of my topics will be the conceptual relationship between those theses. That relationship is one of the central questions in the philosophical con­troversy between realism and antirealism, of which the legal debate is only one of se­veral battlefields (Wright 1992b, 2).

2. The Noble Dream

The conceptions I include in this section hold that all constitutional cases, even the most difficult and controversial ones, have a unique right answer and that, therefore, in all cases presented to them, the courts have a unique applicable solution. Sometimes these conceptions have been known asJormalist.4 The conceptions presented here, however, are restatements of the thesis of the right answer taken from the more recent literature. More precisely, I will look at an especially robust form of metaphysical realism, linked to the natural-law tradition, which has recently been defended by M. Moore (1981, 1982, 1985, 1987, 1989, 1992a, 1992b), as well as by O. Brink (1988) and H. Hurd (1990). I will also take up the conception of legal interpretation held by R. Dworkin

4 Or as BegriJfsjurisprudenz, or mechanical jurisprudence, etc. Cf. Jhering 1854, Pound 1908. Interesting presentations of formalism can be found in Bobbio 1965, Schauer 1988.

CONSTITUTIONAL INTERPRETATION 135

(1977c, 1985, 1986, 1991), certainly the most distinguished advocate of the thesis of the one right answer. Again, I would like to underscore that what these conceptions have in common is their acceptance of thesis (4), whereas theses (1), (2) and (3) can be ascribed to the Noble Dream only for an idealized version which, as we will see, may fit only to metaphysical realism.

a) The Noble Dream o/metaphysical realism

Wright (1992b, 1 f.) recently characterized philosophical realism as a combination of modesty and presumption. Taking as an example the case of the philosophical position about the existence of an external world, modesty would be expressed in the notion of that external world's independence from us, i. e., from the conceptual scheme with which we think about it. Human thinking is, in Frege's words, only a map of the world, and the world does not in the least depend on the cartographical techniques used in re­presenting it. What is presumptious, in contrast, is the thought that we are able to con­ceive of the world in an adequate way and to know the truth about it, although whether or not our thinking fits to the world is determined independently of human cognitive activities.

Perhaps it is precisely this combination of modesty and presumption that makes philosophical realism attractive. Now, as Wright himself has recognized, the extent of that modesty depends on the propositions we refer to. To think that there are mathema­tical entities that render our mathematical propositions true or false is less modest than to think that there are ordinary objects (like tables and chairs) that make our proposi­tions about the ordinary world true or false. Similarly, modesty is weaker in the case of legal propositions: To think that there are legal duties and rights in the world that make our legal propositions true or false is not precisely a very modest philosophical position.

Thus, not all realisms are equally plausible. Realism in mathematics, ethics, aes­thetics, literary criticism, or law is a philosophical position that is more presumptious than modest. However, several philosophers recently constructed a realist theory for le­gal propositions that is worth to be analyzed.

Moore is probably the clearest supporter of philosophical realism as applied to legal propositions and, more specifically, to the question of the interpretation of the law. In his words (1985, 397), "our interpretive practices reveal us to be both metaphysical realists and (as a special case of that) natural lawyers". The following quote gives a good idea of his position:

"I call the thory of interpretation I wish to defend a natural law theory of interpretation because of two pro­positions that characterize it: I) that there is a right answer to moral questions, a moral reality if you like; and 2) that the interpretive premises necessary to decide any case can and should be derived in part by re­course to the dictates of that moral reality. In short, real morals, not just conventional morality or 'shared values', have a necessary place in the interpretation of any legal text." (Moore 1985, 286)

Moore (1985, 283 f.) regards the theory of legal interpretation as part of a more general theory of legal reasoning. In his view, the theory of legal reasoning has four parts: (i) a theory about the nature of law that enables us to identify the set of authoritative state­ments that belong to the law, i. e. a rule o/recognition, (ii) a theory about facts that en-

136 CHAPTER V

abIes us to determine which of the innumerable descriptions of the facts of a case should be used. i. e. that enables us to determine which is the relevant universe of cases (in the sense explained in Chapter IV). (iii) a theory of interpretation that enables judges to establish the interpretive premises which connect the facts of a case with the norms of the law. and (iv) a theory about logic and its place in legal reasoning.

Although. obviously. the parts are related. we can adopt Moore's classification and turn to the theory of interpretation.

According to Moore. a theory of interpretation presupposes a theory of the meaning of linguistic expressions in natural language. and such a theory is part of a broader theory of communication. Theories of meaning. in turn. can be divided into conventionalist and realist theories (1985. 291-301). Conventionalist theories see the relations between symbols and things as essentially arbitrary. as a mere matter of con­vention. Realist theories. in contrast. hold that the meaning of words is not determined by convention. Realist theories usually (Moore 1985. 321-328. Brink 1988. 116-119) rely on the so-called causal theory of reference defended in contemporary philosophy oflanguage by Kripke (1971. 1980) and Putnam (1973. 1975).

According to that causal theory of reference. the referents of proper names (e. g .• 'Aristotle'. 'Kelsen') as well as of certain predicates (such as 'water'. 'gold') determine their meanings. For Kripke. proper names are rigid designators. they name the same in­dividual object in all possible worlds: and according to Putnam. predicates like 'water' also are rigid designators. since their referent is what determines their meaning. Such rigid designators are regarded as the result of a kind of "baptizing act" ("naming cere­mony" or "act of introduction"; Putnam 1975; cf. also Evans 1982. 121 ff.) that is ac­cepted by some linguistic community and carried out in reference to certain objects of our experience. We may be ignorant or mistaken about the true nature of the properties of the object referred to (perhaps Aristotle wasn't the teacher of Alexander the Great or the student of Plato ...• or Kelsen was not born in Prague. or was not a member of the first Austrian Constitutional Court; perhaps the term 'water'. when it was first used. was erroneously associated with some mysterious structures other than Hp ... ). but the "baptizing act" has fixed the referent of the object once and for all. with complete in­dependence of our corresponding beliefs (a linguistic division of labor is the idea with which Putnam explains that the beliefs of some members of a community may be mis­taken).

Putnam (1975. 215-219) sees the causal theory of reference as a critique of two commonly accepted theses of the theory of meaning: a) that to know the meaning of a term consists in being in a certain state of mind ("meanings are in the head"), and b) that the meaning. in the sense of the intension of a term. determines its extension. With the causal theory. we can sustain. on the contrary. that it is the extension that determines the intension of a term. and that meaning therefore does not correspond to any state of mind at all.

Moore holds an unconventional theory of meaning according to which "to find the meaning of all such words is. accordingly. not a matter of finding some antecedently stipulated convention. the task is the more creative one of discovering how the world is constituted" (Moore 1985. 338). Moore distinguishes between nominal kinds (derived

CONSTITUTIONAL INTERPRETATION 137

by convention) and natural kinds (whose classification corresponds to the structure of the world).5 Many terms that appear in legal texts must therefore be interpreted ac­cording to the realist theory of meaning. Moore gives the example of the word 'death' and, regardless of whether there is some legislative definition of it, holds:

"The meaning of words like 'death', therefore, is not to be found in some set of conventions; meaning is neither a set of standard examples, nor a set of properties conventionally assigned to a symbol. The meaning of a word like 'death' is only to be found in the best scientific theory we can muster about the true nature of that kind of event. By assuming that there are such true natures of natural kinds of things, the theory of meaning presupposed by our usage is aptly termed a realist theory of meaning." (Moore 1985, 300)

Moore, who is also an advocate of moral realism (1982, 1992b), sustains that there are moral kinds (1985, 333-338). Thus, terms - sometimes known as evaluative terms -which appear in legal texts, as, for instance, the expression 'inhuman or humiliating punishment or treatment' in art. 15 of the Spanish Constitution of 1978, refer to the mo­ral kinds of the inhuman and the humiliating which await "discovery" in real life. Thus, what kind of punishment or treatment the Spanish Constitution prohibits is a question that does not depend on an evaluation or convention (nor on the intentions of the foun­ders of the Constitution, nor on social speech habits), but on certain characteristics of the world, independently of human beings.

Now, the question here is not to go into a discussion of the plausibility of moral realism,6 although later I will have to say something about the relevance of that thesis for the question we are interested in. Here, I will concentrate exclusively on how the position of Moore or Brink deals with the four theses I attributed to the Noble Dream.

First, the metaphysical thesis according to which there is a legal world which, in­dependently of our knowledge about it, can render our constitutional propositions true or false. According to metaphysical realism, whether or not a treatment is inhuman or humiliating does not depend on us (or our beliefs or attitudes), but on how the world is. Thus, Brink (1988, 120 f.) writes that 'just' or 'cruel' denote natural moral kinds, just as 'toxic' denotes a natural chemical kind. We may be wrong about whether or not certain behaviours are humiliating, but if we had adequate knowledge about the relevant facts we would have reasons for changing our judgment.7

In the case of the law, the realism such a position must endorse is especially ro­bust. It must assume that for each legal system there is a corresponding, completely de­termined possible world that can give each action a single deontic qualification.

Hurd (1990) has taken a step to strengthen this position - perhaps for the price of making it implausible. According to Hurd, legal norms should not be understood, as

5 In a recent article, Moore (l992a) has distinguished those kinds from the kinds now called functional (predicates referring to artificial objects created by man). Putnam (1975, 242-244) had extended his con­ception in order to include the names of artefacts created by human beings in his theory.

6 See the critique of moral realism in Mackie 1977a and Harman 1977 and the recent defence in Brink 1989 and Smith 1994.

7 Moore (1985, 312) defends a kind of coherentist epistemology for the justification of particular judgments. To him, such a theory seems compatible with a realist theory of truth (truth as correspondence).

138 CHAPTER V

would be the case in a communicative model of law, as prescriptions that guide our conduct, but as descriptive statements about what he calls "optimal legal arrangements". Therefore,

"the legislature has authority for us only if it functions as a heuristic guide to antecedently existing moral facts. To the extent that the legislature mistakenly decides or distorts the optimal state of affairs or its atten­dant obligations, the legislature will fail to have (theoretical) authority for us" (Hurd 1990, 1010).

On this view, there is a world of optimal legal arrangements independently of what le­gal norms stipulate. And legal norms are true or false, depending on whether or not they describe this world. This is a formulation of the metaphysical thesis in its pure state, so to speak.

In this way, the semantic thesis of metaphysical realism, according to which the meanings expressed in constitutional statements are determined by their truth condi­tions, independently of how we may be able to verify them, also is established. As Moore (1987, 455) says, "what is distinctive about realism is the belief that propositions can be true even if we, at present, have no rational grounds for believing or asserting that they are true". The metaphysical thesis does in fact imply this semantic thesis. If there is a world of optimal legal arrangements independently of what legal norms stipulate, then the propositions referring to legal duties and rights are true or false, depending on whe­ther or not they correspond to that world. As I have formulated it, the semantic thesis also implies the metaphysical thesis.

The logical thesis, i. e., the acceptance of bivalence, is implied by the metaphy­sical and the semantic thesis. We may have doubts about whether or not force-feeding prisoners who are conducting a hunger strike for the improvement of the living condi­tions in prison constitutes an inhuman or humiliating treatment. 8 But the corresponding constitutional proposition is either true or false, since "The realist justifies his bivalent semantics by sharing the language-users' faith that there is an underlying reality out there whose 'hidden nature' is determinative of truth-values one way or the other in cases we presently cannot resolve" (Moore 1987, 485).

According to Moore (1987, 506), the thesis of the one right answer can only be grounded in moral realism and in a natural-law theory of legal reasoning. Moore thus thinks that an institution like constitutional review is justified because the courts with the competence to review the constitutionality of other provisions only apply the law (the constitution) that stipulates rights and obligations which already exist, independent­ly of the rulings of those courts. Thus, Moore says (1985, 395): "The justification for judicial review is simply that people really have rights, and no consensus of the majority, even when embodied in a statute, should be allowed to trample on them. A realist can concede the antidemocratic nature of judicial review because he can justify it with higher values." Therefore, it is obvious that courts may be mistaken in stipulating certain rights and duties of citizens, just as our descriptive judgments about the world may be wrong.

8 Cf. the rulings of the Spanish Constitutional Court of June 27, 1990 and July 19, 1990 as well as Atienza 1993, chap. 4.

CONSTITUTIONAL INTERPRET A TION 139

Those theses constitute a forceful defence of metaphysical realism as applied to constitutional propositions, and, as we saw at the beginning, according to Moore they give meaning to the interpretive practices of our lawyers and courts. Such a robust form of metaphysical realism, however, requires a careful evaluation which I will present at the end of this section. But first I will look at a more modest way of defending the thesis of the one right answer, namely, the theory of Ronald Dworkin.

b) Dworkin: 'The Noblest Dreamer'9

In contemporary legal theory, the thesis of the one right answer is closely associated with Dworkin's legal theory. It is a sophisticated, complex theory that has developed over the years. lO But for the present purpose, I will concentrate on his defence of the thesis of the one right answer, as it can be found in Dworkin 1985. 11

Dworkin considers a version of legal positivism which he criticizes in all his works, using an analogy to a certain exercise of literary criticism. Dworkin invites us to assume that some Dickens scholars go about analyzing David Copperfield as if David were a real person. They want to assert certain propositions about David as true: that he attended Salem House, that he was an industrious student, etc. Dworkin suggests that these propositions are governed by the following rules (1985, 134):

,,(1) Any proposition about David may be asserted as 'true' if Dickens said it, or said something else such that it would have been inconsistent had Dickens denied it."

,,(2) Any proposition about David may be denied as 'false' if Dickens denied it, or said something else such that it would have been inconsistent had Dickens said it."

These rules do in fact stipulate truth-conditions for propositions about David similar to those formulated above, in Chapter II, for legal propositions. They amount to saying that a proposition about David is true if it describes a sentence that belongs to Dickens's novel, or if it describes sentences that are the logical consequence of propositions that belong to that novel; and that a proposition about David is false if it describes a sen­tence whose negation belongs to Dickens's novel, or if it describes sentences whose ne­gations are logical consequences of propositions that belong to that novel. That means that the truth-values of propositions about David presuppose the (explicit or implicit) existence of certain sentences in the novel. 12 Now, what happens if the novel does not

9 As Hart says (1983, 137): "He [Dworkin] is, if he and Shakespeare will allow me to say so, the noblest Dreamer of them all, with a wider and more expert philosophical base than his predecessors, and he con­centrates formidable powers of argument on defence of his theory."

lO For this evolution, cf. Raz 1986a.

II This is a revision of Dworkin 1977b.

12 As we will see, Dworkin thinks these rules are insufficient. But in a certain sense, they are excessive: ac­cording to them, many irrelevant consequences will be true. Let's assume the proposition expressed in the sentence 'David attended Salem House or visited Barcelona'. Since the negation of that proposition is in­consistent with something Dickens said - namely, that David did attend Salem House -, according to rule (I) it is true. Here too, the notion of relevant logical consequence, introduced in Chapter I, is useful to re-

140 CHAPTER V

say anything, neither explicitly nor implicitly, about some fact about David, e. g., that David had a homosexual relationship with Steerforth?'According to the rules proposed by Dworkin, such propositions would have no truth-value, since because of (1) they are not true, but because of (2), they are not false either.

According to Dworkin, this exercise in literary criticism suggests a special kind of legal positivism, according to which a legal proposition is true if some authority has, explicitly or implicitly, issued a norm of a certain kind, and it is false if some authority has issued, explicitly or implicitly, a norm in the opposite sense. If none of the two things has taken place, then the proposition would have no truth-value.

Dworkin adds (1985, 135) that this kind of legal positivism must be distingui­shed from another kind which holds that if there is no legal norm to determine the va­lidity or invalidity of, e. g., a contract, then the proposition according to which that con­tract is legally valid as well as the proposition according to which it is invalid are both false. In our literary exercise, this would be equivalent to saying that 'According to Dickens's novel, David had a homosexual relationship with Steerforth' is just as false as saying that 'According to Dickens's novel, David did not have a homosexual relation­ship with Steerforth'. As the analysis of the controversy between Russell and Strawson about predicative statements - treated in Chapter II - has shown, the two reconstruc­tions are equivalent. One only needs to distinguish - as I did in Chapter II - two con­cepts of falsehood. One is the sense of falsehood reconstructed by Dworkin's rule (2); on this notion of falsehood, the proposition referring to David's relationship with Steer­forth is neither true nor false. According to another notion of falsehood, however, a pro­position referring to David is false if it is not one of the propositions that can be shown to belong to Dickens's novel. In this second sense, the propositions referring to the rela­tionship between David and Steerforth are false. The advantages of distinguishing be­tween falsehood and non-truth are by now well-known, and I will not go into them here.

For the conception of law underlying this analogy, there are cases that have no right answer. Dworkin thinks, however, that neither in literary criticism nor in law there is any reason to assume such a meagre set of rules. According to Dworkin, a more ade­quate way of literary practice would be one where "a further proposition about David is assertable as true (or deniable as false) if that further proposition provides a better (or worse) fit than its negation with propositions already established, because it explains in a more satisfactory way why David was what he was, or said what he said, or did what he did, according to those already established propositions" (Dworkin 1985, 136). Analo­gously, he prefers a reconstruction of legal practice similar to that which arises from those new rules of literary criticism. He (1985b, 137) thinks that the most influential ar­gument in favour of the thesis against the right answer is the one he calls the thesis of demonstrability. It holds that if the truth of a proposition cannot be shown even if all the relevant 'hard facts' (i. e., physical facts and facts about the behaviour of people, in­cluding their thoughts and attitudes) are known or stipulated, then it cannot be true. Dworkin does not think that there are good reasons for the thesis of demonstrability. He

strict the exercise of literary criticism to those propositions stated by Dickens, and their relevant logical consequences.

CONSTITUTIONAL INTERPRETATION 141

suggests the possibility of an extention of our ontological baggage, and he adds that if moral facts existed, then the thesis of demonstrability would be false. If the fact that a certain behaviour (like force-feeding prisoners on a hunger strike) is humiliating would depend on the existence of certain moral facts, then the corresponding legal proposition could be true even if, for some reason, we would not know it. Dworkin adds that though he will not try to make the existence of moral facts plausible, he will maintain the existence of some kind of facts beyond hard facts. Returning to the literary analogy, he believes that the best form of literary practice is the one that postulates the existence of certain facts of narrative consistency. Such facts do not presuppose that David is a real person, but they offer arguments that enable us to look at the novel in its best recon­struction. The existence of that kind of facts, according to Dworkin, is given as internal to literary practice, and it is immune against any skeptical attack, because it does not as­sume that there is any metaphysical literary world outside of the practice to which we can turn in order to find out whether or not David did have a homosexual relationship with Steerforth. The question is answered using the technique of considering whether the truth of that proposition would allow us to understand the novel any better. There may, of course, be unanswerable questions, but Dworkin thinks that that is very rarely the case.

Dworkin invites us to carry that literary practice over to our theory of legal pro­positions. A legal proposition would thus be true if it formed part of the best justifica­tion that can be given of the legal system in question. In view of the legal material at hand, such a justification must be given along the lines of two dimensions: the dimen­sion offit and the dimension of political morality. 13

The dimension of fit is related to the idea that a political theory is a best justifica­tion of existing legal material if it accounts for the legislative and jurisdictional history of some legal order better than other theories. It may happen that several theories offer different justifications that adequately fit the legal material; in that case, the dimension of political morality is decisive for the best justification: it is that political or moral theory that comes closest to including the rights people de facto possess. In Dworkin's view, this means that practically in all cases precisely one right answer will emerge from the filter of the two dimensions.

These two dimensions obviously constitute a particular conception of the nature of law which Dworkin has been constructing on the basis of his critique of Hart's legal positivism. I will only mention a few steps in the course of this construction:

(i) the law is not only a set of rules, but also of principles of political morality that cannot be identified with the help of the rule of recognition (Dworkin 1977c, ch. 2);

(U) the law is of an interpretive nature - the only nature that can adequately ex­plain the existence of disagreements among jurists (Dworkin 1986, chap. 2); for in­stance, when jurists disagree about whether or not a certain treatment is humiliating, that disagreement is not about the linguistic conventions underlying the meaning of 'hu-

13 Cf. Dworkin 1977c. ch. 4. The importance of those two dimensions has been reaffirmed. while the theory has been developed further in Dworkin 1986.

142 CHAPTER V

miliating treatment', but about the different conceptions given by different theories of the concept of humiliating treatment; and

(iii) the best reconstruction of this interpretive notion of law can be found neither in conventionalism, according to which judges discover and apply certain conventions of the past, nor in pragmatism, according to which judges must construct future law without being bound by conventions, but in the law as integrity which offers the best in­terpretation of our legal practice (Dworkin 1986, ch. 7).

None of these points, which constitute sharp and brilliant arguments, will be ana­lyzed here. My purpose is more modest. It only consists in showing some of the grounds on which Dworkin's conception of law, that leaves no room for indeterminacies, is based.

From what has been said, it already follows that Dworkin defends the logical thesis - that of bivalence - and the legal thesis - that of the one right answer. As I said before, he is the most brilliant advocate of these theses. Now, it is interesting to note that Dworkin does not think it necessary to base them on a defence of the meta­physical and the semantic thesis. He thinks that the truth of the thesis of the right an­swer (i. e., of determinacy and objectivity in the law) is independent of the status of the metaphysical and the semantic thesis. This is shown by his position on skepticism.

Dworkin (1986, 76-85)14 distinguishes two kinds of skepticism: external and in­ternal skepticism. External skepticism is a thesis about legal practice as interpretive practice. On this conception, which is a metaphysical theory, moral values (like aesthe­tic ones) are not part of the fabric of the universe. When someone says that slavery is unjust, he is not stating something about reality, but projecting his own opinions on the world. 15 According to Dworkin, this kind of skepticism is no challenge to the objecti­vity of value judgments of any kind. Whatever the status of this metaphysical theory, the relevant skepticism is internal skepticism, according to which, in moral discussion, 'slavery is unjust' does not have better arguments on its side than 'slavery is just'. Dworkin holds this position to be untenable. In any interpretive practice (literary, mo­ral, legal), participants are able to give arguments in favour of a position, and those ar­guments can be evaluated within the practice. Thus, internal skepticism is highly im­plausible.

It actually looks as if Dworkin's interpretive conception of the law is construc­tivist with respect to the existence of a legal world that is independent of our knowl­edge, and of a legal truth that is independent of our capacity to justify, under ideal epis­temic conditions, the truth of our legal propositions.

Now, is it plausible to maintain this constructivist approach on the metaphysical and semantic level while holding the thesis of bivalence and the thesis of the one right answer, which I have attributed to metaphysical realism? The critique of Dworkin that I will present later will turn on the possibility of making the two conceptions compatible. Here, I want to point out that Dworkin's conception is perhaps compatible with the one

14 A recent reconsideration of these arguments, which will not be discussed here, can be found in Dworkin 1996a,1996b.

IS For a projectivist conception of moral judgments cf. Blackburn 1984, 197-202.

CONSTITUTIONAL INTERPRETATION 143

Putnam (1981, 1983, 1987, 1990, 1995) has called internal realism. Against a metaphy­sical realism that accepts the metaphysical and the semantic thesis and assumes an ex­ternalist perspective (the point of view of God's Eye), Putnam sets internal realism, ac­cording to which the question 'What objects does the world consist of?' only makes sense within some conceptual scheme. For this conception, truth is a kind of (idealized) rational acceptability - a kind of ideal coherence of our beliefs with each other and with our experiences, as represented in our belief system - and not a correspondence with states of affairs that are independent of the mind or of discourse.

Putnam's internal realism makes the metaphysical and semantic considerations underlying Dworkin's theory plausible; but the question remains whether the logical thesis of bivalence and the legal thesis of the one right answer can be justified with the arguments of internal realism. I will come back to it in the critical part of this section.

c) Critique

The Noble Dream of metaphysical realism is based on the truth of moral realism. For this conception, the truth of legal propositions depends on the existence in the world, independently of human activities, of moral facts those propositions may correspond to. Actually, as Hurd's conception shows, it is not only that legal propositions could then be true or false; according to this moral universe, legal norms themselves would be directly true or false. Thus, the most straightforward criticism against this position is to question moral realism.

Moral realism has been criticized with - in my opinion - convincing argu­ments (Mackie 1977a, Harman 1977). According to Moore, however, moral realism is the conception that best reconstructs our evaluative practices and our moral discourse. When two persons disagree, for example, about whether or not it is cruel to hold some­one in solitary confinement for ten days, they disagree about some (perhaps superveni­ent) characteristics of the world. Skeptical (emotivist, prescriptivist, projectivist, quasi­realist, etc.) conceptions of moral judgments cannot account for this objectivity require­ment for moral judgments. Moral realism has problems when it comes to explaining the motivating capacity, the practical force attributed moral judgments are assumed to pos­sess. Sometimes, realists reply that knowing what is good does not have any motivating capacity (Moore 1982, 1122 f., Brink 1989, 37-50). By this, they try to avoid Mackie's argument of the 'queerness' of moral judgments which describe objective characteris­tics of the world and, at the same time, prescribe - constitute reasons for - how we should behave. In my view, however, the skeptics' reasons about moral realism, i. e., the moral antirealists' reasons, are convincing. When we genuinely disagree about mo­ral questions, i. e., when our disagreements are not based on a lack of information about non-moral facts, we do not disagree in our beliefs about, but in our attitudes towards the world.

Yet, I do not want to argue here in favour of moral antirealism, in favour of the inexistence of moral facts in the fabric of the universe. I only wish to point out two problems which the conception of metaphysical realism about legal propositions, in my view, cannot solve:

144 CHAPTER V

(i) The causal theory of reference, even if considered true, is no solid argument for metaphysical realism.

(ii) Moral realism does not succeed in showing its relevance for the analysis of legal propositions.16

(ad i) Before commenting on the relations between the causal theory of reference and realism, it should be said that the causal theory of reference confronts severe prob­lems. Kripke's and Putnam's examples are always of indexical expressions like proper names, e. g. 'Kelsen', or of general natural kinds, like 'water' or 'gold'. But even in these cases, it is highly questionable whether the referent can be held stable by way of a naming ceremony. Evans (1985, 10-11) has presented the following counterexample: Apparently, 'Madagascar' was the name of a place on the African continent; and be­cause of a misunderstanding, Marco Polo converted it into the name of that big African island. Now, does that mean that 'Madagascar', in spite of actual linguistic usage, is not the name of that big African island, since the chain of reference cannot be kept intact? Thus, one can argue that not even for proper names the causal theory of reference is adequate.

On the other hand, the link between the causal theory of reference, and the ana­lysis in terms of natural kinds, and metaphysical realism also is questionable. More specifically, Putnam, the main advocate of that analysis, rejects (1981, 22-48) metaphy­sical realism (cf. Bix 1993, 167). What's more, it has even been suggested that some as­pects of the analysis in terms of natural kinds are compatible with an antirealist concep­tion and with a theory of linguistic meaning as use (a theory Moore holds to be com­pletely wrong). Marmor (1992, 138-146) suggests to pay attention to Wittgenstein's distinction between criteria and symptoms (Wittgenstein 1953, sect. 354). Take the case of the meaning of 'rain'. We know that, when the barometer falls, it rains; but we also recognize rain through certain sensory experiences like humidity and coldness. While the fall of the barometer is a symptom of rain and, in that sense, is not part of the meaning of 'rain', our sensations of humidity and coldness - remember that for Wittgenstein all criteria can be revoked - are criteria of the use of 'rain', and they define, as a matter of linguistic convention, what 'rain' means. According to Marmor, we can draw two conclusions from the distinction between criteria and symptoms:

"First, that the most plausible examples of indexical predicates the theory of whose reference can change without causing any changes in meaning, do not constitute counter-examples to Wittgenstein's analysis at all, as they involve a revision of symptoms rather than criteria. Second, that according to Wittgenstein, a change in criteria carries with it a respective change of meaning." (Marmor 1992, 143)

(ad ii) Moral realists often argue that the obvious fact that there is moral disagreement is no argument for moral skepticism. We also have important disagreements in other areas (e. g., about the origin of the universe, or of certain diseases, or about the true

16 I will not go into another way of looking at Moore's theory here: Schauer (1991,215-218; cf. Bix 1993, 154-157) has suggested an interpretation of Moore's conception not as a metaphysical and semantic theory, but as a normative theory of rule-application according to which rules should be applied in accordance with their purpose (a theory similar to that of Fuller 1969) rather than in accordance with their meaning.

CONSTITUTIONAL INTERPRETATION 145

identity of Shakespeare), but that does not lead us to confuse an intersubjective agree­ment with objectivity (Moore 1982, 1089 f.). Now, to clarify the difference between an intersubjective agreement and objectivity does not imply that moral disagreements too can be resolved through some kind of objectivity. In the natural sciences and the social sciences, there are epistemic procedures that enable us to resolve our disagreements and to revise our beliefs. Obviously, there is no absolute agreement about these procedures and methods. But in moral matters, there isn't even anything remotely comparable to those procedures (Waldron 1992, 170-176). If convergence is a mark of truth (Wiggins 1987, 147-151), and the best way to explain a broad convergence in some belief is that there is something that makes it true, then it is more difficult in moral matters to speak of true or false moral judgments.

When two persons disagree about whether or not it is humiliating to forcefeed hunger-striking prisoners (or whether the death penalty is cruel; cf. Richards 1983 for the discussion about the U. S. Constitution), once all non-moral aspects of the matter have been settled it is hard to imagine what procedure could make them change their views. In the writings of moral realists, there is nothing to be found on this epistemic procedure besides a general suggestion to 'look and see' and, in most cases, the adop­tion of a consistency-oriented epistemology for moral judgments. Consistency, how­ever, is a necessary, but not a sufficient condition for truth. Astrology, witchcraft, and many other superstitions can be presented in the form of perfectly consistent systems, but that does not make them true.

The problem with disagreements about legal propositions that contain some kind of moral evaluation is to find a procedure to resolve them. Since moral realists have little to say about this, the truth of moral realism is irrelevant for analyzing legal propo­sitions. Besides, there is no reason why moral realism should be the best reconstruction of our judicial practice. If judges do not have an epistemic instrument enabling them to know that moral reality, then its existence cannot be of any help for them to overcome their doubts. Thus, since metaphysical realists lack good arguments for defending either the metaphysical or the semantic thesis, the logical thesis of the bivalence of legal pro­positions and the legal thesis of the one right answer are left unsupported too.

That criticism, however, does not apply to Dworkin's conception, since he tries to justify the logical and the legal thesis without adopting the metaphysical and the se­mantic thesis.

Now, if Dworkin's metaphysics and semantics are constructivist, then how does he justify the logical and the legal thesis? Dworkin seems to argue that his special kind of internal realism is sufficient for the truth of the logical and the legal thesis. Putnam, in contrast, does not think that internal realism can justify bivalence for all propositions. The notion of truth of internal realism is that of justification under ideal epistemic cir­cumstances. And according to Putnam,

"the two ideas of the idealization theory of truth are (1) that truth is independent of justification here and now, but not independent of all justification. To claim a statement is true is to claim it could be justified. (2) truth is expected to be stable or 'convergent'; if both a statement and its negation could be 'justified', even if conditions were as ideal as one could hope to make them, there is no sense in thinking of the statement as having a truth­value" (Putnam 1981,57).

146 CHAPTER V

Putnam (1995, 6) therefore refers to Dworkin's doctrine of the one right answer as 'fan­tastic'. Since the notion of truth as justification does not guarantee bivalence for any field of knowledge, internal realism permits us to speak of the truth of legal proposi­tions and even of moral propositions, but it does not permit us to say that all legal and all moral propositions have a truth-value.

Hart (1983, 139 f.; cf. also 1994) already felt that the thesis of the one right an­swer was to become the most disputed point in Dworkin's work. For Hart, the claim that whoever wants to give an answer to an evaluative question must assume that there is one single, objective right answer in all cases, if those questions are to make any sense at all, is implausible. This is just the same in the law.

As Greenawalt observed,

"discretion exists as long as no practical procedure exists for detennining if a result is correct, informed lawyers disagree about the proper result, and a judge's decision either way will not widely be considered a failure to perfonn his judicial responsibilities" (Greenawalt 1975, 386).

Let us go back to the literary exercise proposed by Dworkin: If, from Flaubert's novel, there is no way to determine the colour of Madame Bovary' s eyes, then it does not help to assume that there is a better way of reconstructing the novel that would give a truth­value to the proposition about the colour of Madame Bovary's eyes.17 Since there are various ways of reconstructing the novel with respect to this, it is better to say that pro­positions about the colour of Madame Bovary's eyes have no truth-value.18 The same can be said with respect to legal propositions: if a case does not have a unique norma­tive solution and can be resolved in different ways, then it does not help to insist that there must be a reconstruction that is better than all others.

The conclusion on this point is that internal realism is insufficient for justifying the thesis of bivalence and the thesis of the one right answer (cf. Munzer 1985, Moore 1987, and a defence of Dworkin in this point by Presby 1994). If Dworkin wants to jus­tify the thesis of bivalence for legal propositions and the thesis of the one right answer, he must adopt metaphysical realism. Thus, Dworkin faces the following dilemma: Ei-

17 For a funny and instructive discussion of this matter, cf. the novel by Julian Barnes, Flaubert's Parrot, London: Jonathan Cape 1984, ch. 6.

18 J. L. Borges has given a brilliant representation of this conception of literary criticism in his 'Nueve en­sayos dantescos' (1982), in: Obras completas, vol. IV, Barcelona: Cfrculo de Lectores 1993, 253-255. In discussing the interpretation of verse 75 of the penultimate canto of Hell in the Divina Commedia, where Ugolino de Pisa, after speaking about the death of his sons in the Prison of Hunger, says 'Poscia, phi che'l dolor, pote il digiuno', Borges admonishes us not to confuse art with reality. It is a historical question -now impossible to ascertain - whether Ugolino ate his sons' flesh; and the interpretation of Dante's Ugo­lino can remain indeterminate on this point (i. e., indetenninate between the possibility that he did eat his sons' flesh or that, while the pain did not kill him, hunger did), and according to Borges, that is the best in­terpretation of this verse of Dante's. In Borges's own words (ibid., p. 235): "In real time, in history, when­ever a man has several alternatives, he chooses one and eliminates and loses the others; this is not so in the ambiguous time of art which resembles that of hope and that of oblivion ... In the darkness of his Tower of Hunger, Ugolino does, and does not, devore the beloved corpses, and this oscillating imprecision, this un­certainty, is the strange material it is made of."

CONSTITUTIONAL INTERPRETATION 147

ther he adopts metaphysical realism, and thus has bivalence and the right answer, or he adopts internal realism, and thus has legal propositions without a truth-value and no right answer, i. e., the law is - sometimes - undetermined, and judges therefore have discretion in some cases. Internal realism is, so to speak, too weak a nail for carrying the weight of the thesis of the one right answer. 19

On balance, then, the Noble Dream comes out as follows: If metaphysical realism were true, the thesis of the one right answer would also

be true. That means that the metaphysical and the semantic thesis of the Noble Dream imply the logical and the legal thesis. Metaphysical realism, however, is - to use Wright's adjective - much too presumptuous to be an acceptable doctrine. Now, if we stick to more modest internal realism, like Dworkin, then we cannot deduce the logical and the legal thesis from the metaphysical and the semantic thesis anymore. Therefore, the Noble Dream is not an adequate conception of legal propositions and constitutional interpretation. In its most presumptuous version, that of metaphysical realism, it presup­poses an excessively robust Platonic metaphysics; in its more modest version, its basis is too weak a foundation for what characterizes the Noble Dream, i. e., the thesis that the law determines all behaviour and that, therefore, judges, instead of ever creating the law they apply in solving cases, can always discover it.

3. The Nightmare

The Nightmare is a conception of the law according to which judges never apply pre­existing law to the cases they solve, but always create the law. The Nightmare is com­monly linked to American legal realism the best-known versions of which are the most extreme ones. Here, I will present the conception of Frank (1930) as the supreme repre­sentative of the Nightmare. Recently, however, the authors of the movement known as Critical Legal Studies have presented a more sophisticated version of that radically skeptical position, heavily relying on certain interpretations from the philosophy of lan­guage of the second Wittgenstein (about the idea of rule-following). I will also look at a certain tradition in continental legal thinking that subscribes to that skeptical version of the law.

The skeptical conception of the Nightmare insists that, if at all, only applicative legal propositions have a truth-value, and that that truth-value does not depend on what certain general norms may say, but only on what certain judicial decisions (individual norms) stipulate. Thus, in order to answer the question of whether some individual -say, Gaius - has the obligation to pay for the thing he bought from Ticius one must wait until there is a judicial decision about the case. As long as there is no such decision, the proposition expressed in the statement that 'Legally, Gaius has the obligation to pay for the thing he bought from Ticius' has no truth-value. Likewise, the

19 Obviously, there are many other interesting theses in Dworkin that deserve discussion. I will mention only two: the strange nature of Dworkin's legal propositions which, at the same time, describe what the law stipulates and prescribe how we ought to behave (cf. Bulygin 1982a); and his questionable insistence on the virtue of consistency that leaves no room in his theory for the incommensurability of values (cf. Mackie 1977b, Finnis 1987, Raz 1992, and a defence parallel to that of Dworkin in Hurley 1989, 193-212).

148 CHAPTER V

proposition expressed in the statement 'Constitutionally, Claudia has the right not to reveal her religious beliefs' (cf. art. 16.2 of the Spanish Constitution) has no truth-value as long as that right has not been stipulated by some court.

The law, then, is nothing but the set of individual norms (judicial decisions) which establish the rights and duties of citizens. Before there is a judicial decision, legal propositions are only a kind of predictions, propositions about the future, about the decisions the judges will make. 20 Thus, since, according to this view, what establishes the truth-value of legal propositions is the authoritative decisions of judges, judges can­not err when they stipulate the rights of the parties in a lawsuit (cf. Woozley 1984, Bix 1993,78-88).

a) Skepticism as therapy: Jerome Frank

Although the conception of the Nightmare can be attributed to many authors of Ameri­can legal realism21 and can also be found, in particular, in the work of K. N. Llewellyn (1930, 1962; cf. Twining 1973), here I will present its most extreme version, from the work of Jerome Frank.

Frank's conception of the law, in his own words, is the following:

"We may now venture a rough definition of law from the point of view of the average man: for any particular lay person, the law, with respect to any particular set of facts, is a decision of a court with respect to those facts so far as that decision affects the particular person. Until a court has passed on those facts no law on that subject is yet in existence. Prior to such a decision, the only law available is the opinion of lawyers as to the law relating to that person and to those facts. Such opinion is not actually law but only a guess as to what a court will decide." (Frank 1930, 46)

This passage in Frank shows that he accepts the Nightmare's legal thesis 4a) according to which the courts always create and never apply law. The opposite position, i. e., that of the Noble Dream, Frank holds to be irrational, "an illusion or a myth" (Frank 1930, 13) and "a partial substitute for the Father-as-Infallible-Judge" (Frank 1930, 19). The ex ante determination of all behaviour by the law is an illusion of which mature persons should free themselves, just as, according to Freudian psychology, they should free themselves of a certain image of the father as an infallible authority (Frank 1930, 125, 264, 277). Frank urges us to recognize such a conception as a myth, and, therefore, to give it up and to accept reality: the law is created by the judges through their decisions.

Thus, legal propositions about cases that have not yet been judicially decided cannot be true or false, they are mere conjectures.

But, what is the philosophical conception of language underlying these skeptical conclusions by Frank? Why should we abandon the image of a previously established law that determines the rights and duties of citizens? There are not very many argu-

20 This is how one can understand Holmes's famous phrase (Holmes 1920, 173): "The prophecies of what the courts will do in fact, and nothing more pretentious, are what I mean by the law."

21 Cf. the presentations of American legal realism in Tarello 1962, Summers 1982.

CONSTITUTIONAL INTERPRET A TION 149

ments for it to be found in our author's work. Sometimes, however, he suggests posi­tions that come close to an antirealist philosophical conception. Thus, he says:

"Words ... become our masters because the very nature of language fosters a belief in the independent reality of what are merely verbal contrivances." (Frank 1930, 91)

"The cause of certainty-illusion and other obstructions to realistic thinking about law is the crystallization of primitive attitudes in the language which children learn and grown men employ." (Frank 1930, 94)

The philosophical foundation of skepticism about the capacity of general rules to de­termine behaviour has, however, been enriched by certain interpretations of the philo­sophy of the second Wittgenstein. The heirs of American legal realism, the representa­tives of Critical Legal Studies,22 have expressed this new foundation. Therefore, I now go on to consider their conception.

b) Skepticism in Critical Legal Studies: "Law is Politics"

The legal theory of Critical Legal Studies has underscored the dependence between a certain image of the law and its foundations in the theory of political liberalism (cf. Un­ger 1975). However, I will not go into this aspect of the movement, which insists on the so-called fundamental contradictions of liberalism23 and asserts that the rule of law is a liberal myth that should be given up in favour of the recognition that there is no pos­sible distinction between law and politics.

Rather, I will try to analyze the thesis, often called the thesis of radical indeter­minacy, according to which it is never possible to give meaning to general rules, be­cause general rules are only "empty vessels" (cf. Altman 1990,91). The thesis of radi­cal indeterminacy is usually grounded in a certain interpretation of the second Wittgen­stein's reflections about rule-following.24 The interpretation of Wittgenstein as a skep­tic goes back to Kripke (1982) and is based on the following paradox:

"This was our paradox: no course of action could be determined by a rule, because every course of action can be made out to accord with the rule. The answer was: if everything can be made out to accord with the rule, then it can also be made out to conflict with it. And so there would be neither accord nor conflict here." (Wittgenstein 1953, par. 201)

According to Kripke (1982, 7), this is perhaps the central problem of the Philosophical Investigations. It is a new skeptical challenge that affects the philosophy of mathematics and logics as well as the philosophy of language and of meaning.

22 For the relationship between American legal realism and Critical Legal Studies, cf. Casebeer 1977, Schlegel 1979, Gordon 1982, Livingston 1982, Tushnet 1986, Altman 1986, P6rez L1ed6 1996, ch. IV.

23 Cf., for example, Kennedy 1979, Kayris 1982, Unger 1983, Kelman 1987, CaisamigJia 1992, Carrino 1992, as well as a defence of liberalism against these critiques in Altman 1990.

24 Cf., for example, Caracciolo 1982 - who at an early time already pointed out some of the skeptical conclusions from Wittgenstein's reflections on rule-following for legal theory -, Tushnet 1983, Yablon 1987, Langille 1988, Patterson 1992a.

150 CHAPTER V

Kripke (1982, 8 f.) offers the following example of a mathematical rule (exam­ples of this kind had been used by Wittgenstein 1953, pars. 185-187): someone who knows the rules of addition knows that the expression '68 + 57' has the result '125' (even if he has never before solved that equation). This is the right answer in an arith­metical sense, since the metalinguistic meaning of '+' (,plus') denotes a function that, when applied to the numbers '68' and '57', leads to that result. The challenge of the skeptic, now, is precisely that he questions that metalinguistic meaning. Perhaps '+' de­notes another function, of which we were not aware in our past usage; perhaps it recon­structed our usage only for numerical expressions smaller than 57, and the result of adding 68 and 57 could be 5.25 Perhaps in the past we used 'plus' and '+' in order to denote a function we can call 'quus' and symbolize by $. Let's define it as follows:

x $ y = x+y, ifx, Y < 57 = 5 in all other cases.

According to Kripke, there are reasons for taking the challenge of the skeptic seriously. And, obviously, this does not only affect mathematics but also applies to all meaning­related uses of language, and to all instances of rule-following, whether mathematical, linguistic or legal. Kripke compares this form of skepticism with Hume's skeptical argument about causality (Hume 1740, 269):

"Both develop a sceptical paradox, based on questioning a certain nexus from past to future. Wittgenstein questions the nexus between past 'intention' or 'meanings' and present practice: for example, between my past 'intentions' with regard to 'plus' and my present computation '68+75=125'. Hume questions two other nexus, related to each other: the causal nexus whereby a past event necessitates a future one, and the inductive inferential nexus from the past to the future." (Kripke 1982,62 f.)

The conclusion of Kripke's Wittgenstein has devastating consequences for the law. General rules do not determine behaviour at all, since any behaviour can be performed in accordance with any rule. That is, there is no way in which to attribute a unique meaning to a norm formulation. It is here where the skeptical theses about legal propo­sitions have a more solid foundation - although, as I will try to show below, not a completely solid one - than in the work of Frank. Since norm formulations have no meaning, the metaphysical and the semantic thesis of the Nightmare become plausible. Objective knowledge of norms turns out to be impossible (Caracciolo 1982) and, there­fore, our legal propositions (which refer to norms) have no truth-value (the logical the­sis of skepticism). And, what's more, there is never a right answer for legal cases (nei-

25 Cf. also another example of a representative of Critical Legal Studies: Tushnet (1983, 822 - taken from the Wittgensteinian Winch 1958, 29-32): "Consider the following multiple choice question: 'Which pair of numbers come next in the series I, 3, 5, 7? (a) 9, II; (b) II, 13; (c) 25, 18.' It is easy to show that any of the answers is correct. The first is correct if the rule generating the series is 'list the odd numbers'; the second is correct if the rule is 'list the odd prime numbers', and the third is correct if a more complexe rule generates the series. Thus, if asked to follow the underlying rule - the 'principle' of the series - we can justify a tremendous range of divergent answers by constructing the rule so that it generates the answer that we want. As the realists showed, this result obtains for legal as well as mathematical rules."

CONSTITUTIONAL INTERPRETATION 151

ther for easy nor for hard ones) and, therefore, the law is always created by the judges (or, perhaps, by the organs entrusted with the execution of the law), in assigning rights and obligations on an individual basis. In the semantic sense, the constitution does not exist.

For the skeptic, there are no clear legal cases, and this conception is used in the context of Critical Legal Studies to maintain that the idea that the law normatively de­termines behaviour is an illusion and that, therefore, all judicial decisions obey non­neutral, political criteria, i. e., criteria that presuppose the use of value judgments on the part of the judges.

c) Interpretation as creation

In the legal literature of continental Europe, the skeptical conception of rules is usually regarded as foreign to the legal tradition. I will try to show, however, that that concep­tion, although minoritarian, has also been present there.

Thus, for example, it was defended by Ascarelli (1959, 140). For Ascarelli, the object of interpretation is not a norm, but only a text; a norm exists only when it is ap­plied.

In the words of Satta:26

"Ascarelli's solution ... is simply the following: that the nonn, as it is understood by jurists and even in or­dinary language, does not exist. What does exist is what he calls a text (or a behaviour): the nonn is created in the very moment of its interpretation; but that happens only in one almost indescribable moment, since immediately thereafter, the nonn suddenly becomes a text again - a text that will need another interpreta­tion, in relation with another particular case, in order to become again, and only for a short moment, a nonn ('another nonn' , to be precise), and so on ad infinitum." (Satta 1968, 496 f.)

This conception has been presented in great clarity, and apparently adopted, by Guastini (1992, 109 f.; 1993, 336-338). In his view, the skeptical theory holds that interpretation is an activity, not of acquiring knowledge, but of evaluation and decision. This is so be­cause there is no such thing as the proper meaning of words, since every word can have a number of different meanings. Thus:

"It follows that interpretive statements ('text T means M') are neither true nor false. Such statements have the same deep structure as so-called stipulative definitions, that is, those definitions that do not describe the actual use of a tenn or expression, but propose to give a tenn or expression a certain meaning rather than another one. That stipulations are neither true nor false is uncontroversial. Thus, it is understandable that, from this point of view, legal norms do not exist before interpretation, but are its result." (Guastini 1992, 109)

26 Satta advocated a conception of lawsuits known as a monistic conception, according to which the law is stipulated by judicial decisions and does not exist prior to them. Thus, he says (1968, 23): " That French writer who said that the law is what the judges say it is thought to be contributing a skeptical note, but in­stead he expressed a profound truth, perhaps the only truth that can be expressed about the definition of law." The quoted sentence is similar to that passage of Bishop Hoadly so often quoted by American realists (cf. Gray 1902, 102, 125, 172): ,,Nay whoever hath an absolute authority to interpret any written or spoken laws it is he who is the lawgiver to all intents and purposes and not the person who first wrote or spake them."

152 CHAPTER V

And in a recent paper on norm propositions in which an interesting form of skepticism is defended, Mazzarese writes:27

"Truth-values do not seem to apply to the statements of an interpretive set. That is so since the function of an interpretive statement - because of the variety of criteria of legal interpretation together with their particular features - does not amount to a description but rather to a proposal and/or to an ascription. Furthermore if moral and/or social and/or political evaluations are conceived as significant elements which either have to interplay, or even, indeed, do interplay with legal interpretation, then there are no theoretical grounds for applying truth-values to interpretive statements." (Mazzarese 1991, 55 f.; cf. also Vemengo 1993)

In summary, then, since the results of interpretation are never propositions that can be true or false, and since legal propositions presuppose an interpretive activity - for the attribution of meaning to certain norm formulations -, the logical thesis of skepticism is adequate, and this implies the legal thesis: there is never a right answer for any case, that is, judges always create the law, and never apply it to particular cases.

d) Critique

Before taking on the core of the skeptical challenge, i. e., the paradox of rule-following, it seems useful to make some remarks about three aspects skeptics tend to overlook:

(i) an almost obsessive (cf. Hart 1983, 123), and unjustified, emphasis on law­suits; (ii) the distinction between the context of justification and the context of dis­covery of judicial decisions; and (iii) the distinction between the definiteness and the infallibility of judicial deci­sions.

(ad i) The temptation to turn legal theory into a theory of the application of the law dis­torts a great part of North American legal theory. Legal norms regulate many aspects of our daily life, and only a very small number of them ever become legal cases. Most of us, e. g., comply with the duty to pay taxes, and when we are asked to justify that be­haviour, we do so by pointing out that there are certain legal norms requiring us to do so. Excessive attention to the problems of the application of the law can lead skeptics to think that all cases are controversial (because, by definition, all cases of lawsuits are) and that, therefore, the law is not about being governed by rules (Schauer 1991, 181-196). But of this distortion, Hart had already warned:

"The rule-sceptic is sometimes a disappointed absolutist; he has found that rules are not all they would be in a formalist's heaven, or in a world where men were like gods and could anticipate all possible combinations of fact, so that open texture was not a necessary feature of rules. The sceptic's conception of what it is for a rule to exist, may thus be an unattainable ideal, and when he discovers that is is not attained by what are called rules, he expresses his disappointment by the denial that there are, or can be, any rules." (H. L. A. Hart 1961, 135)

27 For a critique of this kind of skepticism, cf. Mendonca 1996.

CONSTITUTIONAL INTERPRET A TION 153

On the other hand (as Hart 1961, 133, and Schauer 1991, 169-171, have pointed out), the claim that the assumption that human behaviour is governed by legal norms is a myth is inconsistent with the assertion that there is such a thing as judicial decisions. The existence of judicial decisions presupposes the existence of some kind of norms (secondary rules of adjudication, for Hart; jurisdictional rules, for Schauer) which confer jurisdictional competence on some set of individuals and, thus, give their deci­sions legal authority. As Hart (1961, 133) has added: "In a community of people who understood the notions of a decision and a prediction of a decision, but not the notion of a rule, the idea of an authoritative decision would be lacking and with it the idea of a court."

(ad ii) Often, an argument realists tend to use in favour of skepticism is the plurality of the facts that lead a judge to take a certain decision in a particular case. The judges' prejudices, their education, their social class, their reactions to witnesses, etc. (cf. Frank 1930, 114 f.) determine the decision at least as much as their beliefs about what is required by legal norms. That argument, however, can be countered with the distinction between the context of discovery and the context of justification. The distinction comes from philosophy of science (Reichenbach 1951,7) where the activity consisting in the discovery or expression of a scientific theory - an activity that must be studied by historians and sociologists of science, in order to show what factors playa role in the development of scientific knowledge (the system of the selection of scien­tists, their process of education, the structure of scientific communities, etc.) - is dis­tinguished from the obviously different activity consisting in the justification of such a scientific theory - an activity where logic and the rules of scientific methodology play the most important role. That distinction can be transferred to judicial decisions (Was­serstrom 1961, Golding 1987, Atienza 1991,22 f.): the procedure by which a decision is reached is one thing, and the procedure by which such a decision is justified is quite another. Many of the arguments of American realists are useful in the sphere of the con­text of discovery of judicial decisions. It is, indeed, important to know the motives that can explain the decision taken by a judge: her social class, her ideological prejudices, her legal education, etc. Now, the reasons that justify (cf. Nino 1985, 126, for the distinction between explanatory and justificatory reasons) a decision are the reasons the judge appeals to for grounding the decision in the law. If one takes this distinction into account, it becomes clear that an appeal to the plurality of reasons that explain judicial decisions is no argument for skepticism.28

(ad iii) In Hart's opinion (1961, 137-141), the most interesting form of skepti­cism is the one that argues that, since under certain circumstances the decisions of the courts are irrevocable, to say in those cases that the court erred has no consequences in the legal system, because it does not change the rights or duties of anyone. If the Spanish Constitutional Court, for example, decides that a particular opinion expressed by a particular person is not protected by the right to freedom of expression (for instance, because it violates someone else's right to privacy), then the skeptic would say

28 Some skeptics reject that distinction; e. g .• that is how Golding (1987. 135) argues with respect to Frank. and how Mazzarese (1996) argues in general.

154 CHAPTER V

that nothing is gained by saying that the court erred. And insofar, the skeptic would conclude, the constitution is what the Constitutional Court says it is. Hart, however, has clearly distinguished the definiteness from the infallibility of judicial decisions: That the decision of a court is definite does not mean that it is infallible. This is just as in tennis, where the decision of the umpire about whether or not a service was good is definite, without there fore being infallible (Greenawalt 1992, 45). The game of tennis is not a game of the umpire's arbitrary will, but a game governed by the rules of tennis. Similar­ly, the law is not what the courts say it is, although what the courts decide is definite (cf. also Moreso 1994b).

Here, it should be recalled that a wrong decision of a court cannot make an un­constitutional (and, therefore, invalid) norm constitutional (or valid). Rather, such deci­sions can, at best, make such norms applicable (cf. Chapter III on this point).29

The skeptics' most important argument, however, is Kripke's argument about rule-following. Now, Kripke's interpretation of Wittgenstein is far from uncontrover­sial, and it has been extensively criticized.30

The following passage in Wittgenstein (1953, par. 201), where he explains the paradox of rule-following, says:

"What this shows is that there is a way of grasping a rule which is not an interpretation, but which is ex­hibited in what we call 'obeying the rule' and 'going against it' in actual cases. Hence there is an inclination to say: every action according to the rule is an interpretation. But we ought to restrict the term 'interpretation' to the substitution of one expression of the rule for another."

Obviously, this passage requires a detailed analysis that I cannot deliver here. But we can at least extract some ideas from it that will turn us away from the paradox, and thus from skepticism.

Wittgenstein does not question that the idea of rightness is applicable to rule­following. What Wittgenstein questions is that there must be some mediating entity between our understanding of a rule and our following it. Platon's (mechanical, as an extreme formalist would assume) image of rules as rails is what is in question here. But Wittgenstein's position can be understood as intermediate one between such a hyposta­zation of rules and the negation of their existence (Wright 1989, 297; Bix 1993, 41). Wittgenstein does not deny that rules can determine their correct application, as indeed they do in cases that are clear. Rather, according to Wittgenstein, the concepts of same (gleich) and agreement (Ubereinstimmung) depend on the notion of rule-following (Wittgenstein 1953, pars. 224 and 225):

"The word 'agreement' and the word 'rule' are related to one another, they are cousins. If I teach anyone the use of the one word, he learns the use of the other with it."

29 Although it has been said (Schauer 1991,119) that validity is a necessary condition for applicability, in Chapter III I have argued that validity is neither necessary nor sufficient for applicability.

30 By philosophers (Blackburn 1981; McDowell 1981; McGinn 1984; BackerlHacker 1985; Wright 1989, 1993) and legal theorists (Bjarup 1988; Schauer 1991, 1992a; Bix 1992, 1993, 1995; Marmor 1992; Radin 1992; Smith 1992; Endicott 1996).

CONSTITUTIONAL INTERPRI;:T ATION 155

"The use of the word 'rule' and the use of the word 'same' are interwoven. (As are the use of 'proposition' and the use of 'true')."

For Wittgenstein, certainty in clear cases derives from certain fundamental facts con­cerning agreement in judgments, agreement in social context and stability of the world:

"If the language is to be a means of communication there must be agreement not only in definitions but also (queer as this may sound) in judgements. This seems to abolish logic, but does not do so. - It is one thing to

describe methods of measurement, and another to obtain and state results of measurement. But what we call 'measuring' is partly determined by a certain constancy in results of measurements." (Wittgenstein 1953, par. 242)

According to Wittgenstein, the agreement that exists in clear cases of rule application does not depend on an agreement in opinion, but on an agreement in the form of life (Wittgenstein 1953, par. 241). I cannot analyze here the complexities of the notion of 'form of life' in Wittgenstein's work (for its use in legal theory, cf. Aarnio 1986 and 1987, ch. 4; Niiniluoto 1981). Suffice it to say that it refers to the common background that makes human communication possible.

Thus, to interpret a rule is only to substitute one symbol for another. Without the support of a set of shared practices, this does not guarantee that the new symbol will be understood and will make rule-following possible. That a rule is understood is shown in the acts of following it (cf. Kress 1989). Therefore, one should be cautious about the position of Schauer (1991, 64-68) according to which following formulated rules (as in the case of legislated law) must be distinguished from following rules that are not cano­nically formulated (as in the case of judges' law or precedent). For Schauer, the skep­tics' considerations have force in the latter, but not in the former case. However, even in the case of following formulated rules we need the notion of semantic rules that stipu­late the meaning of the formulated terms; and, therefore, even in this case agreement in the forms oflife is necessary. In Marmor's words:

"A rule, in other words, is a sign and its meaning cannot be determined by another sign; the meanings of rules, like those of all symbols, must be determined by the actions themselves, that is, by the way the rules are used." (Marmor 1992, 149)

The conclusion from all this is that Wittgenstein' s notion of rule-following is not a good foundation for skepticism. The metaphysical and the semantic thesis of the Nightmare do not seem well-founded, and therefore we should also abandon the logical and the le­gal thesis. In particular, there is no reason for maintaining that in all the cases they de­cide judges create the law, and that the application of the law is always a political ques­tion. Actually, as Radin (1930, 866) - an author sometimes associated with American legal realism - said: "Words are not crystals, [ ... ], but they are after all not portmanteaus. We can not quite put anything we like into them."

I do, however, wish to underscore three lessons to be drawn from this analysis of rule-following in Wittgenstein:

156 CHAPTER V

(i) First, the question of legal interpretation - of the convenience of trying to substitute one symbol for another - arises when we have doubts about the application of some rule to a particular case (Wroblewski 1985, 22, 26, 35, and 1992, 91 f.; Mar­mor 1992, 153; Schauer 1992b, 740, 1993, xii; Endicott 1994). This takes us back to the old legal saying that interpretatio cessat in claris (or in claris non fit interpretatio). In clear cases, there is basically an understanding of the rule manifested in the acts of fol­lowing it.

(ii) Meanings are not platonic entities, existing independently of our linguistic usage. Nor are they mental entities that reside 'in our heads'. Rather, on the contrary, meanings are, or perhaps one should better say: supervene (Williamson 1994,205-209), our linguistic uses.

(iii) Finally, cases can also be controversial because of a lack of agreement in judgment, reflecting a lack of agreement in the most fundamental forms of life. That is the reason why the distinction between clear cases and hard cases is not always comple­tely determined.

4. The Vigil

The Vigil represents a moderate constructivist position like that of Hart (1961, ch. VII), according to which, although there is no legal world independently of our ability to know the law as it is constructed by human beings (metaphysical thesis), we can deter­mine the meaning of constitutional propositions by showing that certain consequences follow from the constitution (semantic thesis), and although this does not guarantee that all constitutional propositions are true or false (logical thesis), sometimes there is a right answer for constitutional cases and, therefore, sometimes the courts have a right to, and legally ought to, apply it, although they can, of course, err in doing so (legal thesis).

This is a position that can explain decisions like that contained in the statement of the Spanish Constitutional Court STC 34111993 of November 18, 1993, on the con­stitutionality of art. 21.2 of Organic Law 111992, on the protection of public safety, 31 in relation with art. 18.2 of the Spanish Constitution which reads: "Private homes are in­violable. They may not be entered or searched without the consent of the dweller or a judicial warrant, unless a crime is in the act of being committed." The right of inviola­bility of private homes was discussed only for the three cases mentioned in art. 18. The meaning of the phrase "in the act" was questioned. That is, it was questioned whether

31 Here is the text of the first paragraphs of art. 21 of the law in question:

"I. The agents of the Security Forces may enter and search a private home only in the cases permitted by the Constitution, and within the terms stipulated by the law.

2. To the effect of the provisions of the previous paragraph, it will be a legitimate reason for entering and searching a private home on the grounds that a crime is in the act of being committed, if the Public Security Forces have well-founded information leading to the certain belief that one of the crimes sanctioned in the Criminal Code concerning toxic or hallucinogenic drugs or psychotropic substances is being, or has just been, committed, on the condition that the immediate intervention of the agents is necessary for preventing the crime, the escape of the perpetrator, or the disappearance of the effects or instruments of the crime."

CONSTITUTIONAL INTERPRET A TION 157

the members of the Public Security Forces are constitutionally authorized to enter and search private homes when they have information that leads them to believe with cer­tainty that a crime related to drug-trafficking is being, or has been committed, if and when their immediate intervention is required for preventing the crime, the escape of the perpetrator, or the disappearance of evidence. Art. 21.1 of the implicated Organic Law was declared unconstitutional by the Constitutional Court.

Here, a short analysis of the reasons given by the Court for the unconstitutional­ity of the provision, insofar as they refer to the question of whether or not the legisla­tor's notion of 'in the act' is in accord with the constitutional concept, can be of interest. The text of the Constitution contains no definition of 'in the act', but as the Court ob­serves, "the Constitution surely does not arise from a legal vacuum, but out of a legally organized society" (legal foundation 8° and STC 1111981), and the concept of 'in the act', like other concepts, is "rooted in a legal culture in which the Constitution is em­bedded and [in the context of which it] must, therefore, be identified, without forgetting what this Court has earlier called the 'generalized ideas and beliefs generally accepted among jurists, judges and legal experts in general' (STC 11 11981)". For the Court, the core of the concept of 'in the act' consists of two aspects: evidence or sensorial percep­tion of the crime, and the urgency of police action. But while the urgency of police action is provided for in the article under scrutiny, well-founded information and certain belief are not exactly the same as the knowledge based on evidence or perception that, according to the Court, forms an integral part of the notion of 'in the act'. 32 As the Court asserts:

"If the language of the Constitution is to keep its meaning - which is the finn premise of any interpretation - one cannot but recognize that these connotations of 'in the act' (evidence of the crime and urgency of police action) fonn part of the concept mentioned in art. 18.2 of the Basic Nonn, which, using the traditional notion, has drawn the limits of a fundamental right, and at the same time, of the infringement of that right by the public powers."

Thus, the unconstitutionality of the provision is grounded in the fact that the language of the Constitution is meaningful, and that the meaning of 'in the act', in the context of the practice of interpreting and applying the law, excludes some of the cases included by the legislator. Here, it is interesting to note that the Court uses, if I may say so, a Wittgensteinian notion of meaning; the meaning of 'in the act' is a function of the use - in this case, since it is a terminus technicus, the legal use - of that expression.

Although this was a controversial issue (that has provoked heated debates not only in the legal community, but in Spanish society at large), it is not clear whether it

32 In epistemology, sentences like 'x knows (that) p' are usually analyzed as follows:

1) x believes that p. 2) x is justified in believing that p. 3) p is true.

The requirement contained in 3) is perhaps excessive for the purpose of defining 'in the act', and what really matters here is justified belief. The difficulty is that the justification of the belief in the case of 'in the act' must be based on sensorial data, and belief can obviously be based on other kinds of evidence.

158 CHAPTER V

was, in fact, a hard case. Jurists actually agreed that the use of the expression had been followed in the past with sufficient regularity for one to say that it was a reasonably clear case. And that is also what the Court says - in my view, correctly. Therefore, one can say that, according to the point of view of the Vigil, this is a case where things are sufficiently objective and determined, i. e., where one can show that the norm deriving from the Constitution contradicts the norm expressed in art. 21.2 of the Organic Law under scrutiny: the norm deriving from the Constitution is that in cases where there is well-founded information by other means than sensorial perception, the agents of the Public Security Forces are not authorized to enter and search a private home.

In order to argue in favour of this conclusion, we do not need any of the premi­ses of the Noble Dream: neither that 'in the act' designates a natural kind, nor that there is a reconstruction of the notion of 'in the act' that best reconstructs the corresponding legal practice and solves all doubtful cases. The notion of 'in the act' solves this (as can be seen: generic) case, but that does not mean that there could not be doubtful cases, lo­cated in the twilight zone of the concept. The Nightmare, in contrast, would leave the Court with the responsibility of having stipulated a meaning of 'in the act' as the result of an evaluation. It would not be the Constitution that would have determined this re­sult; the Constitution would not determine any result at all.

I think that none of these versions adequately reflects the habitual interpretive practice of the courts and that one can say that in this case, the answer of the Court was right, according to well-established communicative practice. But that does not mean that there will always be a right answer.

In that sense, the Vigil maintains a conception of the interpretation of the law -and, in fact, of interpretation tout court, since the interpretation of the law does not pre­sent too many particular problems (cf. Bulygin 1992,29) - that is grounded in the ob­vious fact that the language of the law lives in the web of rules, conventions and prac­tices natural languages consist of. The only form of giving sense to norm formulations is to attribute it to them in the sphere of this web. And since this web is not completely stable, because of the open texture of the expressions used, their ambiguity, etc., doubts about the application of these norm formulations arise. Thus, the organs of application that must justify their decisions with such norm formulations do have discretion in re­solving cases.33

A possible explanation for the instability of the web supporting human commu­nication can be given in accordance with ch. V of Lewis' influential book Convention (1969). According to Lewis, our natural languages actually are only a hybrid resonance of the possible languages of which they are part (169, 201). That means that our com­munication does not take place in one single language, with a set of interpreted expres­sions, but in a number of overlapping languages that, thus, allow for stable communica­tion as well as for flexibility:

"The different languages of the cluster may have different virtues and vices, and hence may be differently suited to individual opinions, tastes, and conversational purposes. If everyone can pick from the cluster,

33 Kelsen (1960,349-356), Ross (1958, ch. IV), Hart (1961, ch VII) or Bulygin (1992) adopt this point of view.

CONSTITUTIONAL INTERPRETATION 159

incompatible preferences among languages may be all satisfied. Moreover, by not committing ourselves to a single language, we avoid the risk of committing ourselves to a single language that will tum out to be in­convenient in the light of new discoveries and theories; we allow ourselves some flexibility without change of convention." (Lewis 1969, 202)

Based on these ideas, the constitution can be regarded as a convention representing not a single interpretation, but rather a number of interpretations of the text of the constitu­tion (Bobbit 1991, 32). If that is so, then we can understand controversies about inter­pretation (cf. Dworkin 1986, 43-46) as well as, even more importantly, agreements based on a broad overlap of different interpretations (cf. Bay6n 1991, 712-715; Kutz 1994).

We can, then, also understand a number of platitudes about constitutional inter­pretation that would otherwise seem mutually incompatible. Campos (1993) recently tried to show the inconsistencies arising from the following three assertions:

1. The meaning of the text of the constitution has changed and keeps changing. 2. These changes are caused by controversies about the meaning of the text. 3. Such controversies are desirable because judges sometimes make mistakes in their interpretations, and controversies help to distinguish wrong interpretations from correct ones.

For Campos, the first assertion is comprehensible as long as the second is understood in the sense that the meaning of the text is stipulated by each interpretation (as a skeptic would say). But then, the third assertion becomes untenable. If all meaning is stipulated by interpreters, then it certainly makes no sense to speak of the correctness of an inter­pretation.

However, if we keep in mind that the constitution is only a hybrid resonance of several possible interpretations of the text, then we can understand the three assertions without any inconsistency. That the text of the constitution changes only means that in different historical stages, different ones of the possible interpretations are more privi­leged than the others, and these changes are caused by controversies over the meaning of the text, in the sense that not all constitutional actors have used the same interpreta­tion out of the group of relevant interpretations. Thus, judges can make mistakes, for example when they use interpretations that are clearly outside of the range of admitted interpretations, i. e., that are not really interpretations of the constitution.

If our natural languages do not have a, so to speak, stable semantics it is not sur­prising that the constitution, which is part of a natural language, lacks it too. But that it has no stable semantics does not mean that is has no semantics. There may even be doubts about whether or not some particular interpretation belongs to the range of ad­missible interpretations of the text; but the fact that sometimes we have this doubt does not mean that we always have it.

The Vigil represents that moderately objective conception of legal interpretation (Fiss 1982, ColemanlLeiter 1995) according to which, even though the law does not firmly determine all human behaviour, we can formulate a conception of legal proposi­tions that makes it possible to assign a truth-value to them on many occasions.

160 CHAPTER V

In that sense, the Vigil is compatible with many philosophical assumptions, with­out thereby implying severe ontological or semantic commitments for their adherents. As I said in Chapter II, only certain extreme versions of metaphysical realism and of skepticism are excluded. As we have seen, the Vigil can also account for many of our platitudes about the law: that the normative qualification of actions depends on actions of the authorities (a thesis hardly compatible with metaphysical realism) or that judges can be mistaken in establishing the rights and duties of persons (a thesis incompatible with skepticism).

However, there are two very relevant questions in the recent literature on consti­tutional interpretation that deserve a more detailed analysis. They are the role of the au­thorities' intentions in interpretation, and the justification of judicial review of the con­stitutionality of the laws. The next two sections will be dedicated to the discussion of these points.

5. Intention in Constitutional Interpretation

One of the most controversial questions about legal interpretation is the role of inten­tions in interpretation. As is well-known, the traditional distinction was that between subjective theories that attribute an important role to the legislator's intention, and ob­jective theories emphasizing the so-called 'will of the law'. Besides, in the area of con­stitutional interpretation, a version of the subjective theory called 'originalism' has re­cently been defended, according to which the text of the constitution should be interpre­ted in accordance with the intentions of the text's authors (cf. Brest 1980).

In the United States, this conception is accompanied by a particular justification of the procedure of judicial review. In the originalists' s view, what justifies judicial re­view of constitutionality is the appeal to the intentions of the authors of the Constitu­tion; which means that in this view the only restriction imposed on democratic legis­lators is that which unequivocally arises from the intentions of the authors of the con­stitutional text; with respect to matters that do not arise from those intentions, the courts must defer to the legislator. Any conception that legitimates some other kind of restric­tions on the democratic legislator is said to be unjustified (Bork 1971, Scalia 1989). I am not interested here in the politico-theoretical aspects of that conception, but only in the fact that it emphasizes that a constitution can only acquire meaning through the in­tentions of its authors.

The most radical version of this thesis about intention - the Radical Intention Thesis (cf. Raz 1996a) - is the one that says: "An interpretation is correct in law if and only if it reflects the author's intention."

In Chapter I, when I discussed intentionalist theories of meaning (like that of Grice), I already questioned the scope of a theory of the meaning of language that relies on the notion of intention. Many statements have never been uttered by anyone, so if meaning would depend on the intentions of the persons uttering them, such statements would be meaningless. Something similar applies to constitutional interpretation: A constitution has normative consequences (derived norms) that have never been explicit­ly prescribed by an authority and which, therefore, are not backed by any intention.

CONSTITUTIONAL INTERPRETATION 161

Thus, at least for such derived norms, the thesis of intention is inadequate. That alone is reason enough for restricting the thesis as, for example, Raz (1996a) has done by re­placing it by what he calls the Authoritative Intention Thesis:

"To the extent that the law derives from deliberate law-making, its interpretation should reflect the intentions of its law-maker."

This thesis - which in the case of Raz (1986b, chs. 2 and 3; cf. also Marmor 1992, ch. 8; Marmor 1995; Alexander 1995; for an exploration of the thesis with respect to con­stitutional interpretation, cf. Feldman 1992) depends on an elaborate theory of authority which I cannot go into here - strictly limits the scope of intentionalist theories. In the case of legal interpretation, not only do derived norms not depend on deliberate pro­mulgation by their authors; there are also other norms not covered by the thesis of in­tention: It does not apply to common law, nor - it seems - to foreign law, because al­though statutes of a foreign state A may be applicable in another state B, the authors of the legislation of state A are not authorities of state B. 34 Something similar happens with norms that are prior to the constitution (which I have called 'received norms' in Chapter III): for such norms, the clause that they should be applied in accordance with the constitution must be understood to mean that, since the authorities that issued them are not authorities of the new legal order that arose with the new constitution, the inten­tion of those authorities must be regarded as irrelevant; they must be interpreted as if they had been issued after the constitution became valid.35

But the weak thesis of intention, i. e., the thesis that interpretation consists basi­cally in retrieving the intentions of the author of the interpreted object (for a discussion of interpretation as retrieval, cf. Raz 1995 and 1996b), also is plagued by a number of problems.36

I will mention only four of them:

34 This idea can explain why those norms, in order to be applicable, must pass some filter, as, for example, the so-called public-policy clause.

35 This is perhaps what can be replied to assertions like the following (Hernandez Marin 1989,471): "Ac­cording to the philosophy of language, the meaning of an expression depends on the context in which it is formulated, not on any other context. However, E. Garda de Entema, and with him many other jurists, hold that in order to interpret a legal sentence created 100 years ago (for example, some of our Civil Code) one must take into consideration the context given by the Constitution, which was created only 10 years ago." The assertion that some provision of the Civil Code should be interpreted in accordance with the Constitution must be understood in the following sense: The courts which apply that provision (and, there­fore, justify their decisions with it) must, in accordance with the Constitution, disregard the intentions of its authors and take it into account only if one of its meanings is compatible with the Constitution (otherwise, the provision would be derogated by the Constitution). There is nothing strange here, just as it is not strange at all to adopt a sentence pronounced by someone else, regardless of the intentions that other person had in pronouncing it.

36 This thesis has been widely criticized: cf. Radin 1930; MacCallum 1968; Tarello 1980,364-369; Moore 1985; Dworkin 1986, 53-76; Hurd 1990; Posner 1990, 261-269; Schauer 1991, 218-221; Freeman 1992; Shiner 1992, 35-38; Bix 1993, 183-190; Lyons 1993, 141-163; Nino 1994, 88-92; Endicott 1994; Waldron 1995.

162 CHAPTER V

(i) The first problem concerns the question whose authors' intentions are the re­levant ones. (ii) Secondly, we must find out which intentions are relevant. (iii) The third problem has to do with the question to what extent counterfactual circumstances should be taken into account when determining the relevant inten­tions. (iv) And, finally, we must determine the level of abstraction that should be used in describing the intentions of norm-authorities.

(ad i) Legal texts, and constitutions in particular, usually are not the work of one single author, but of a collective organ (a constitutional convention, or a constitutional as­sembly, as in the case of the Spanish Constitution of 1978). Now, in the case of an as­sembly it is much less clear what its intentions are than in the case of an individual ut­tering a statement. If we want to refer to the intentions of a collective organ without committing ourselves to the questionable metaphysical position implied in speaking of such an organ as if it had a mind where its intentions are located, we must refer to the intentions of the majority of the members, i. e., to the intentions of those members who voted for the corresponding text. In order to do this, we must assume that some inten­tions can be shared (a common assumption in law, where we often refer to the will ex­pressed by a company or by a jury, for example) and that in this sense there are col­lective intentions; but that brings us to another problem: not that of whose intentions, but that of what intentions.

(ad ii) Now, assume that during dinner I say to a friend 'Could you please give me some more wine?' What are the intentions relevant for attributing meaning to that sen­tence? It seems that, besides a knowledge of English, all my friend needs to understand is that I am making a request. Of course, in order to understand that sentence correctly, participation in a set of practices is required (we must, to say it somewhat pedantically, share aform of life). Usually, throwing the bottle of wine across the table or pouring the wine directly into my mouth would not be considered correct ways of reacting to my request; they would not show an adequate understanding. Other intentions I may have are irrelevant for the meaning of the sentence; e. g., I may wish to have more wine in order to relax and ignore the boring conversation one of the dinner guests is engaging me in; or I may wish to be served the wine that is left in the bottle in order to avoid that another friend, who is driving, will drink it, or to avoid that someone else whom I don't like very much will get it, etc. None of these intentions is relevant for giving meaning to my sentence.

In this example, there are two aspects that are relevant for the intention behind legislated texts. On the one hand, there is the distinction between intention and purpose (MacCallum 1968, 240 f.). The (intentional) purposes of legislators (or members of a constituent body) are often irrelevant for determining the meaning of those texts. Some­times, we know nothing about the purposes of the legislators; and those purposes also may not even be the same for all those who voted in favour of the respective text. Nor­mally, legislators issue certain texts with the purpose of causing certain changes in soci-

CONSTITUTIONAL INTERPRETATION 163

ety. But the meaning of the texts must be understood independently of the purposes, be­cause unless we know the meaning, we will not be able to assess their capacity for caus­ing those changes. On the other hand, intention is relevant for determining what kind of speech act I am performing, and that depends on the context in which a sentence is ut­tered, and on certain linguistic conventions.

In the previous example, although the sentence superficially looks like a ques­tion, an affIrmative answer without further action would usually mean that the sentence was misunderstood. The intention is part of the force in the theory of meaning; and in the above case, the intention that can be attributed to the speaker is what converts this speech utterance into a request. Similarly, the texts issued by legislators are given in certain contexts, and are subject to certain linguistic and legal conventions (determining the required size of a majority, etc.). Thus, the intention required of legislators is mini­mal: all that is required is that in legislating one intents, under some description, to ef­fect that that which is being legislated becomes law (Raz 1996a). A peiformative kind of utterance is issued, certain contents are converted into law through certain acts. In­tentions do not serve to determine what the meaning of the text is, for the obvious rea­son that only sometimes (some) legislators really know well the text they are voting for. Not even in the case of the text of a constitution is it worth trying to fInd out what in­tentions the legislators had in issuing it, because surely even in that case not all of them had the same understanding of the texts they voted for (as was already said earlier, the constitutional text actually contains not a single, but several constitutions).

As Raz says:

"A person is legislating (voting for a Bill, etc.) by expressing an intention that the text of the Bill on which he is voting will - when understood as such texts, when promulgated in the circumstances in which this one is promulgated, are understood in the legal culture of this country- be law." (Raz 1996a, 267)

This minimal thesis of intention is, I think, all we can concede to intentionalist theories. But it is important to note that from this intention (which can plausibly be assumed to be shared by all those who voted for some text, as, e. g., the text of a constitution) of the norm-authorities it depends that certain contents are regarded as law in a society. Whe­ther this minimal thesis is extended to include other additional intentions clearly expres­sed in declarations of motives, preambles, or parliamentary debates is a contingent question for every legal order and will, therefore, not be analyzed here (but cf. Ezquiaga 1987, ch. 7, on this question in the doctrine of the Spanish Constitutional Court).

(ad iii) Because of the empirical and other difficulties in discovering the intentions of others, it is sometimes said that, actually, the notion of 'author' is a fictitious notion, and that we do not refer to the intention of an actual, historical author, but to the intention some author might have had under certain, counterfactual circumstances. Thus, for instance, ~armorthinksthat

"meaning is assigned through a counter-factual statement. [ ... J an interpretive statement is either a statement on the communication intentions of the actual speaker, or else it must be a counter-factual statement, characterizing the communication intentions of a fictitious speaker" (Marmor 1992, 31).

164 CHAPTER V

The intentions of a fictitious speaker, however, are of no help in interpretation (Endicott 1994, Raz 1996a). There are no criteria for guiding the construction of such a fictitious speaker. Thus, in this way, Hamlet could be interpreted as if it had not been written by Shakespeare but, say, by Woody Allen, or by one of its readers. That idea makes the thesis of the author's intention (intentio auctoris) collapse with the thesis that privileges the reader's intention (intentio lectoris) (cf. Eco 1995, Levinson 1982). And although that is a thesis cherished by the advocates of radical indeterminacy, it is unjustified. As Kelsen (1979, n. 170) wrote:

"Diese Identifizierung des Gesetzes mit seiner Interpretation ist unhaltbar. Aueh die Bibel oder Shake­speare's Hamlet wird interpretiert; aber niemandem wird es einfallen zu behaupten, die Bibel sei von ihrem Interpreten oder Hamlet nieht von Shakespeare, sondem von seinem Interpreten gesehrieben. "

When, sometimes, we have doubts about the intention of an author, these doubts do not go away by inventing a fictitious author and making certain counterfactual assertions about his fictitious intentions. That may be an attractive task for certain kinds of literary criticism, but it is not very helpful in the sphere of legal interpretation.

(ad iv) Finally, we must turn to the degree of abstraction the intentions of the authorities must be described with. This is especially important in the case of constitutional texts, since they usually contain terms whose content is rather dense. I am referring here to terms like 'human dignity', 'liberty', 'equality', 'inhuman and humiliating treatment', etc. What happens with such terms is that linguistic conventions tend to leave them more indeterminate than other terms whose content is not as dense. The meaning of those terms depends on people's moral-political attitudes. Now, does that mean that those terms must be understood as they were understood by the authors? To begin with, we must remember how difficult it is to find out how the authors did understand these clauses. Even among those who voted in favour of a constitutional text, some may have thought that certain treatments are humiliating, while others thought that they are not. Also, some kinds of treatments they possibly never thought of at all (e. g., force-feeding prisoners on a hunger strike), or didn't know what to think of, etc. But even assuming that all agreed about some particular point, e. g., that two weeks of solitary confinement is not a humiliating treatment, we must ask what consequences this has.

Advocates of originalism maintain that especially in those cases, the constitution must be interpreted and applied by restricting these clauses to the meaning they had for those who formulated them. Thus, they believe that it is applicable only to the kinds of cases (paradigms) the authors included in those clauses. That is rather strange: Accord­ing to it, if the authors never thought that (through the use of a new technology) a new method of torture could be invented, then such a new method of torture would not be a humiliating treatment. Why should the language of the constitution be interpreted in this way? Why should those clauses be interpreted at such a low level of abstraction? When we use terms of this kind, we expect to be understood in a wide, flexible sense. Thus, if I invite a friend to dinner, together with other guests on whom I would like to make a good impression, and I say to him 'Please, watch your behaviour', I hope he understands that request in a flexible way. I do not ask him to guess my intentions,

CONSTITUTIONAL INTERPRETATION 165

which may be vague if I do not know my guests' customs; rather, I ask him to develop his own understanding of what it means to behave properly in that new situation.37 For example, I may not know whether the wife of one of my guests is a Moslem, or a Jew, or a vegetarian, etc. I hope that my friend understands my request in a rather abstract way, and not that he follow exactly the good-conduct rules we were taught as children in Catholic school. Another interesting example was presented by Dworkin:

.. Suppose that my mother, now dead for many years, told me when I was young never to do anything unfair in business, and that I am eager to keep faith with her instruction. Suppose I know that she herself, managing her own business, drove small competitors to the wall by undercutting them and then raising prices when they were gone, a practice cornmon in her day and widely thought fair. Suppose I myself, like most people today, find this an unfair business practice. What shall I do? How would it help if someone told me that I should give full effect to her intentions as well as to what she actually said? She had at least two relevant convictions: the first was her desire that her son should do nothing that really is unfair in the conduct of his business; the second was her belief that ruining competitors by temporarily driving prices below cost is not unfair. I now find that I cannot both satisfy the desire and accept the belief, and so I cannot reach a decision simply by resolving to follow her convictions. I must decide which of my mother's convictions - the more abstract or the more detailed - is the right one to follow." (Dworkin 1993, 136)

Obviously, Dworkin leans towards the more abstract interpretation, and he grounds this inclination on his distinction between concept and conceptions (Dworkin 1986,70-73) according to which a concept like justice (Hart 1961, 155-159; Rawls 1971, 5) can be understood as that which different conceptions of justice, different sets of principles, have in common. Here, I will not discuss whether the distinction really holds (cf., e. g., MunzerlNickel 1977), i. e., whether different conceptions do not, in fact, reflect diffe­rent concepts. But I want to underscore that the constitutional clauses under scrutiny do entail a number of possible interpretations, and that the fact that such abstract clauses were used is sufficient reason for not trying to interpret them in their more detailed ver­sion.

Changes in the world may bring about new types of actions that constitute humiliating treatments. But since the concept of humiliation is an evaluative concept, i. e., presupposes the use of moral rules in order to be applied, a change in our moral theories (which for moral realists means truer theories about moral reality, whereas for moral constructivists it only means a change in our attitudes, projected on the world) also can affect our concept of humiliating treatment. Thus, the constitutional clauses in question must be interpreted in their more abstract form, in order to allow for such changes of meaning.

By way of conclusion, we can say that the role of the authors' intentions in the interpretation of the law is limited. Intentions should be integrated in a global theory of linguistic meaning, but they should be located in the theory of force rather than in the theory of sense. There is hardly anything to be added to this; but perhaps one should keep in mind a restriction of interpretation that comes from the so-called principle of

37 Wittgenstein (1953, 33) writes: .. Someone says to me: 'Show the children a game'. I teach them gaming with dice, and the other says 'I didn't mean that sort of game'. Must the exclusion of the game with dice have come before his mind when he gave me the order?"

166 CHAPTER V

charity (Quine 1960,59; Davidson 1984, 136 f., 152 f., 168 f., 196 f., 200 f.; Dworkin 1986, 53), according to which an adequate theory of interpretation maximizes agree­ment, i. e., that one should select, from among the different possible interpretive hypo­theses, that which produces the attribution of the most adequate meaning in the direction word-world (since in the case of prescriptive language, we have no notion of truth, we cannot say that a hypothesis produces a greater number of true statements). That means, e. g., that one should discard as much as possible interpretive hypotheses that would lead us to qualify as obligatory things which are impossible (for logical or for empirical reasons).

Before ending this section, I want to make a comment - partly linked to the re­levance given to authors' intentions in interpretation - about a question that refers not to the interpretation of the law, but to that of certain texts regarded as sacred by certain religions. Because we do speak of interpretation not only in the case of legal texts. For instance, we speak of the interpretation of conversation when, in daily life, we interpret what others tell us. We also use the expression 'interpretation' in order to refer to the in­terpretation of literary, religious or other texts. We even speak of the 'interpretation' of a play or a symphony, where we mean the performance of the respective work. And we say that data are 'interpreted' (scientific interpretation), or historical or social facts (interpretation in the social sciences), or paintings and sculptures (interpretation of art) (cf., e. g., Betti 1990). This shows that 'interpretation' actually does not mean one con­cept, but a family of concepts. And perhaps the family resemblance with theological in­terpretation can show some interesting similarities with the interpretation of the law.

Levinson (1988; cf. also Grey 1984) has pointed out an analogy between consti­tutions and sacred texts, as, e. g., the bible for Christians. In both cases, Levinson (1988,5) suggests, those texts are seen as constituting the identity of a community - a political community in the first, a religious community in the second case. Now, he says, there are different concepts of constitutionalism, just as there are different con­cepts of Christianity (as well as of other religions). These different concepts can be clas­sified according to two criteria: (1) Should 'constitution' be understood as meaning only the constitutional text, understood in the detailed version that reflects the intention of its authors, or should it be understood as the text plus the different interpretive tradi­tions that arise in the course of its application? (2) Is the authority to interpret the Con­stitution, in the Spanish case, conferred exclusively on the Constitutional Court which, therefore, is infallible, or can all of us to some extent take part in that interpretive prac­tice?

If these questions are applied to the bible, according to Levinson (1988, 29) we obtain positions that neatly divide Protestants and Catholics. Protestants reply to the first question with the textualist answer: sola scriptura, whereas Catholics reply dif­ferently: scriptura et traditio. And Catholics answer the second question by attributing infallibility in the interpretation of the scriptures to the schoolmen and, in the last in­stance, to the Pope. Protestants, in contrast, reply with the notion of the right to private judgment according to which all Christians can interpret the holy scriptures.

If we apply these criteria to the interpretation of the constitutional text, we get four possible models of constitutionalism:

(1) (2) (3) (4)

Protestant -1 Protestant -1 Catholic-l Catholic-l

CONSTITUTIONAL INTERPRETATION

Protestant-2 Catholic-2 Protestant-2 Catholic-2

167

I will leave to others the task of fitting the different conceptions of constitutionalism into those classes.38 But I do wish to point out that the peculiar version of the Vigil de­fended above corresponds to position (3) - Catholic-I, Protestant-2. That means that, while it holds that the application of the constitution should take into account the prac­tices and interpretive conventions added to the text (Grey 1975, MunzerlNickel 1977), rather than only the text with the intentions of its authors, since it is those practices and conventions which delimit the set of admissible interpretations that de facto make up the meaningful content of the constitutional text, it does not hold that the decisions of any court are infallible; rather, the meaning of the constitutional text in many cases de­termines the right answer to constitutional cases; the Constitutional Court is subject to the Constitution and is, therefore, itself fallible in the determination of constitutional rights and duties.

6. Judicial Review of the Constitutionality of Laws

The question of the justification of the control of constitutionality has become a topic of political theory and, more precisely, democratic theory. Since many democratic states have judicial organs with the competence to examine the constitutionality of provisions enacted by democratically elected legislatures, the foundation for this counter-majori­tarian power is often questioned. Some say that judicial organs ought to be limited ex­clusively to declaring unconstitutional provisions whose unconstitutionality is plainly visible, and that in all other cases they ought to defer to the legislator (that is perhaps what Bork 1990 had in mind). Some degree of deference to the legislator can also be found in other theories of judicial review, like those of Wechsler (1959), Bickel (1962), Ely (1980) or Choper (1980), where it is grounded in the protection of basic values of constitutional democracy, or in reinforcing the representation of minorities and the transparency of the procedures of representation. Still other theories advocate a more substantive theory of rights that leaves more room for the activity of jurisdictional organs (Tribe 1985, Michelman 1988), or that can produce a constructive conception of constitutional interpretation in which to ground even the most innovative decisions of the courts (Dworkin 1977c, 1986), or that distinguishes constitutional opportunities from normal political opportunities for changing one's positions (Ackerman 1991), or that holds that any theory of judicial review is either impossible (because disagreements about existing conventions are so radical that the latter never succeed in restricting ju-

38 Some authors have taken an explicit stand, though. Thus, Dworkin (1986, 413) has defined his attitude towards the interpretation of the constitution as a Protestant one: "It is a protestant attitude that makes each citizen responsible for imagining what society's public commitments to principle are, and what these com­mitments require in new circumstances."

168 CHAPTER V

dicial activity) or unnecessary (because if there is agreement, then no control is needed) (Tushnet 1988), and so on.39

Unfortunately, the Vigil is not a normative theory of the application of the con­stitution, and still less a normative theory of democracy. In fact, the explanation of con­stitutional primacy given in the present study is compatible with the absence of me­chanisms for the control of constitutionality. It could well be that the legislator has the last word about whether or not some law is in accordance with the constitution. Ac­cording to the Vigil, the legislator too could err, even if his word is decisive. Some ad­vanced democracies have lived, and go on living, without a control of the constitution­ality of their laws. This is a contingent characteristic of legal orders, and the Vigil is supposed to be a conceptual analysis of a more general kind, such that it gives account of all legal orders molded by the idea of the primacy of the constitution. Also, besides the control of the constitutionality of the laws by some judicial organ, there could be an­other means for trying to guarantee the constitutionality of the laws, namely, the per­sonal responsibility of the organ that enacted an inconstitutional norm (Kelsen 1942, 194). The two most common systems in current democracies, however, are (a) diffuse control of constitutionality, as it has been established in the United States, since the famous Marbury v. Madison case, as well as in many other countries; and (b) concen­trated control of constitutionality, as inaugurated by the Austrian Constitution of 1920 and instituted in many countries in Continental Europe. Despite the differences between the two models (cf. Kelsen 1942), it is important to note that both prescribe that some jurisdictional organ is competent to declare laws unconstitutional.

For the Noble Dream, this is no problem, because there is always a right answer, and the jurisdictional organ only faces the epistemic difficulty of finding it. In the case of the Nightmare, in contrast, jurisdictional organs determine the content of the consti­tution, which is nothing but the set of their particular decisions. I have tried to show, however, that both conceptions are inadequate. So now we must look at the position of the Vigil on this point.

For the Vigil, jurisdictional organs will sometimes confront cases that are clear, because there is no indeterminacy once meaning has been assigned and because con­ventions for the attribution of meaning are sufficiently solid not to throw doubts on the meaning of the constitutional text (as in the example of the expression of a 'crime in the act of being committed'). Thus, in such cases there is a right answer for the courts, although they offer a wrong one. If the foundation of constitutional primacy is con­sidered normatively adequate, then judicial review in those cases will be normatively justified.

But they may also be confronted with cases that are hard, because they are parti­cular instances of hard generic cases (due to lacunae, antinomia, generic subsumption, or what in Chapter IV I have called problems concerning the thesis of constitutional ac­cessibility) (cf. Navarro 1993), or because they are hard individual cases (because of problems of individual subsumption), or because there are difficulties in determining

39 For an excellent overview over the controversy in North American theory, and an attempt to apply these conceptions to the Spanish case, cf. Ferreres (1996); cf. also Alonso Garcia 1984.

CONSTITUTIONAL INTERPRET A TION 169

the meaning of the norm under scrutiny or of the applicable constitutional norm (per­haps there are several meanings in the admissible set and it is difficult to know which one is the most adequate). In these cases, according to the Vigil, jurisdictional organs have discretion. Their decision cannot be qualified as right or wrong in accordance with the constitution. This does not mean that judicial organs in that case act illegitimately; instead, what they do is create rather than apply law (cf. Prieto Sanchis 1987, 118-127). As Hart writes:

"That judges should be entrusted with law-making powers to deal with disputes which the law fails to regu­late may be regarded as a necessary price to pay for avoiding the inconvenience of alternative methods of regulating them such as reference to the legislature; and the price may seem small if judges are constrained in the exercise of these powers and cannot fashion codes or wide reforms but only rule to deal with the specifics thrown up by particular cases." (Hart 1994, 275)

In the case of the control of constitutionality, the courts' decisions can have especially serious consequences and can affect far-reaching reforms of legislation; but it seems that the idea of constitutional primacy is important enough to make up for this price.

It must be underscored that the Vigil should not be regarded as a normative theory of the application of the constitution. It says nothing about what judges should do, all things considered. Should they in some cases depart from the applicable norms, all things considered (cf. Schauer 1991, Bay6n 1994)? The Vigil doesn't say anything about it. Thus, it only offers a conceptual analysis of what judges legally ought to do. But not even for easy cases does it offer a theory of what judges should do, all things considered. That depends on an ethical-political theory in which questions like the justification of representative democracy, the obligation of loyalty arising from the acceptance of an office, and the construction of a set of principles of public morality must playa central role. I am thinking here of theories like, e. g., those of Rawls (1971, 1993) or Habermas (1996). The Vigil cannot provide any of that. But to know what judges should do, according to the law, and when their obligations seem determined by the law (clear cases) and when they have discretion (hard cases) is relevant for any legal theory and also, I think, for any normative theory of how and when the law ought to be applied. Unfortunately, however, legal theory cannot ease the burden of anyone who must take decisions of moral relevance.4o

7. Conclusions: Modest Objectivity and Limited Indeterminacy

We have, in the end, not made much progress from our starting point. Hart has left us in a vigil, beleaguered by two dreams of great explicatory force. Extremes are always at­tractive. If the Noble Dream, if metaphysical realism were true, we would not only have

40 Similarly, unfortunately, for those of us who are non-cognitivists in moral matters, that fact does not ease our burden when we must take decisions of moral relevance. As Leff (1979, 1249) said, perhaps ethics is an "unspeakable" notion: "Nevertheless: I Napalming babies is bad. I Starving the poor is wicked. I Buying and selling each other is depraved. I Those who stood up and died resisting Hider, Stalin, Amin, and I Pol Pot -and general Custer too - have earned salvation. I Those who acquiesced deserve to be damned. I There is in the world such a thing as evil. I [All together now:] Sez who? I God help us."

170 CHAPTER V

an elegant legal theory but also an answer to our normative questions. What judges should do, all things considered, would coincide with what they should do from the le­gal point of view. Knowledge of moral reality would also be part of our legal knowl­edge. Skepticism about that strong vision of the law leads us to the Nightmare in which judges must always decide without any help of previously existing rules, day after day they must invent the law they apply to their cases. If the existence of meaningful rules were a myth, skeptics would be right. In both cases, legal reasoning - and interpreta­tion, as part of it - becomes a matter of moral decision. The attempt of the law to insu­late a zone of reasoning fails.

Unfortunately, both conceptions are inadequate. The Vigil teaches us that legal reasoning is partly insulated (Raz 1993) and that we can, as legal positivism has always maintained, conceptually isolate the law from morality; and thus, that our moral duties may not coincide with our legal duties. This does not mean that legal reasoning does not sometimes lead us to moral reasoning, e. g., because the law decided to adopt certain moral standards (Laporta 1993, 60-63), and still less that it exempts us from moral rea­soning in our practical tasks as jurists. Where the law is not enough, jurists should show the ultimate foundations of their proposals of application, in order to allow open discus­sion, in order not to disguise moral decisions under a legal cloak. Now, the modest ob­jectivity of the Vigil can perhaps serve to avoid useless controversy. Perhaps it can also help that the objectives stipulated in democratic constitutions and which many of us consider valuable can be achieved, through the relative insularity, in the more or less extensive zone, immune to indeterminacy, that is provided by the instrument we know by the name of 'law'.

The theses of the Vigil are compatible with the three theses J:Iart (1980) attributes to legal positivism; and insofar, they can be seen as a conception of interpretation in accord with Hart's legal positivism.

(1) The conceptual separation of law and morality: Despite the many contingent connections between law and morality, the Vigil does not establish a conceptual con­nection between them. As I said, it is possible that a court morally ought to do some­thing that does not coincide with what it legally ought to do. The semantic thesis of the Vigil, which says that constitutional propositions are true if, and only if, it can be shown that certain consequences derive from the constitution makes it possible that some of these consequences conflict with the requirements of morality. That possibility of con­flict cannot be eliminated.

(2) The thesis of the social sources of the law: According to that thesis, the exist­ence of the law depends exclusively on the existence of certain social practices which determine the ultimate criteria of membership in the law. This thesis is equivalent to the metaphysical thesis of the Vigil according to which there is no law other than the set of social practices that can make our constitutional propositions true.

(3) The thesis of judicial discretion: ,,[I]n any legal system there will always be certain unregulated cases in which on some point no decision either way is dictated by law and the law is accordingly partly indeterminate or incomplete" (Hart 1994, 272). This thesis is compatible with the logical thesis - the rejection of bivalence - and with the legal thesis - the assertion that sometimes judges create and sometimes they

CONSTITUTIONAL INTERPRETATION 171

only apply law, and that in the latter case, they are fallible (theses 3b and 4b) - of the Vigil.

The Vigil is, thus, a reconstruction of that moderately objective conception of legal and, more specifically, constitutional interpretation which assumes a limited dose of indeterminacy.

BIBLIOGRAPHY

AARNIO, Aulis (1986): On changes in the systematics of law, in: T. Eckhoff, L. M. Friedman, J. Uusitalo (eds.), Vernunft und Erfahrung im Rechtsdenken der Ge­genwart (= Rechtstheorie 10), Berlin: Duncker & Humblot, 161-170.

AARNIO, Aulis (1987): The Rational as Reasonable. A Treatise on Legal Justification, Dordrecht: Reidel.

ACKERMAN, Bruce (1984): The Storrs Lectures: Discovering the constitution, in: Yale Law Journal, 93, 1013-1072.

ACKERMAN, Bruce (1988): Neo-Federalism?, in: Elster/Slagsted (1988), 153-194.

ACKERMAN, Bruce (1989): Constitutional politics/constitutional law, in: Yale Law Journal, 99, 453-547.

ACKERMAN, Bruce (1991): We the People. Foundations, Cambridge, Mass.: Harvard University Press.

ALARC6N CABRERA, Carlos (1995): Derogaci6n en de6ntica, in: id. and A. G. Conte, De6ntica de la Validez, Madrid: Tecnos, 129-148.

ALCHOURR6N, Carlos (1969): Logic of norms and logic of normative propositions, in: Logique et Analyse, 12, 242-268.

ALCHOURR6N, Carlos (1991): l,Hay realmente un desacuerdo entre Strawson y Russell respecto de las descripciones definidas?, in: id.lBulygin (1991), 535-544.

ALCHOURR6N, Carlos (1993): Concepciones de la 16gica, in: id.lMendezlOrayen (1993), 11-48.

ALCHOURR6N, Carlos and BULYGIN, Eugenio (1971): Normative Systems, Vienna and New York: Springer.

ALCHOURR6N, Carlos and BULYGIN, Eugenio (1975): Introducci6n ala Metodologfa de las Ciencias Jurtdicas y Sociales, Buenos Aires: Astrea.

ALCHOURR6N, Carlos and BULYGIN, Eugenio (1979): Sobre la Existencia de las Nor­mas Juridicas, Valencia, Venezuela: Universidad de Carabobo.

ALCHOURR6N, Carlos and BULYGIN, Eugenio (1981): The expressive conception of norms, in: R. Hilpinen (ed.), New Studies in Deontic Logic, Dordrecht: Reidel, 95-121.

ALCHOURR6N, Carlos and BULYGIN, Eugenio (1984): Permission and permissive norms, in: W. Krawietz, G. Winkler, A. Schramm (eds.), Theorie der Normen. Festgabefur Ota Weinberger, Berlin: Duncker & Humblot, 435-469.

ALCHOURR6N, Carlos and BULYGIN, Eugenio (1989): Von Wright on deontic logic and philosophy of law, in: Schilpp (1989).

174 BmLIOGRAPHY

ALCHOURR6N, Carlos and BULYGIN, Eugenio (1991): Andlisis L6gico y Derecho, Madrid: Centro de Estudios Constitucionales.

ALCHOURR6N, Carlos and BULYGIN, Eugenio (1991a): Los lfmites de la 16gica y el ra­zonamiento jurfdico, in: id.lid. (1991), 309-313.

ALCHOURR6N, Carlos, GARDENFORS, Peter and MAKINSON, David (1985): On the logic of theory change: Partial meet contraction and revision functions, in: Journal of Symbolic Logic, 50, 510-530.

ALCHOURR6N, Carlos and MAKINSON, David (1981): Hierarchies of regulations and their logic, in: R. Hilpinen (ed.), New Studies in Deontic Logic, Dordrecht: Rei­del,125-148.

ALCHOURR6N, Carlos and MARTINO, Antonio A. (1987-8): L6gica sin verdad, in: Theoria, 7-8, 7-43.

ALCHOURR6N, C., MENDEZ, J. M., ORAYEN, R. (eds.) (1993), Enciclopedia Ibero­americana de Filosofia, Vol. 7: L6gica, Madrid: Trotta and Consejo Superior de Investigaciones Cientificas.

ALEXANDER, Larry (1995): All or nothing? The intentions of authorities and the au­thorities of intentions, in: Marmor (1995), 357-404.

ALEXY, Robert (1978): Theorie der juristischen Argumentation, Frankfurt a. M.: Suhr­kamp.

ALoNSO, Juan Pablo (1995): Un caso dificil en el C6digo Civil Espanol, in: Doxa (Ma­dridlAlicante), 17-18,403-432.

ALONSO GARciA, Enrique (1984): La Interpretacion de la Constitucion, Madrid: Centro de Estudios Constitucionales.

ALTMAN, Andrew (1986): Legal realism, Critical Legal Studies and Dworkin, in: Phi­losophy & Public Affairs, 15,205-235.

ALTMAN, Andrew (1990): Critical Legal Studies. A Liberal Critique, Princeton, NJ: Princeton University Press.

AMAR, Akhil Reed (1987): Our forgotten Constitution: A bicentennial comment, in: Yale Law Journal, 97, 281-297.

ANDERSON, Alan Ross and BELNAP, Nuel D. (1975): Entailment. The Logic of Rele­vance and Necessity, Princeton, NJ: Princeton University Press.

ANSCOMBE, G. E. M. (1957): Intentions, Oxford: Basil Blackwell.

ASCARELLI, Tullio (1959): Interpretazione e giurisprudenza costituzionale, in: Problemi Giuridici, Milan: Giuffre, 137-162.

ATIENZA, Manuel (1991): Las Razones del Derecho, Madrid: Centro de Estudios Constitucionales.

BIBLIOGRAPHY 175

ATIENZA, Manuel (1992): Sobre los lfmites del analisis 16gico en el Derecho, in: Theo­ria, 7 (16-17-18) 1007-1018.

ATIENZA, Manuel (1993): Tras la Justicia, Barcelona: Ariel.

ATIENZA, Manuel and RUIZ MANERO, Juan (1994): Sulle regole che conferiscono pote­ri, in: ComanduccilGuastini (1994), 55-84.

ATIENZA, Manuel and RUIZ MANERO, Juan (1995): Adversus Pompeyanos. A proposito d'alcune critiche alIa nostra concezione delle regole che conferiscono poteri, in: ComanduccilGuastini (1995), 237-252.

AUSTIN, J. L. (1962): How to Do Things with Words, Oxford: Oxford University Press.

BACKER, G. P. and HACKER, P. M. S. (1985): Wittgenstein: Rules, Grammar and Necessity, Oxford: Basil Blackwell.

BAy6N MOHINO, Juan Carlos (1991): La Normatividad del Derecho, Madrid: Centro de Estudios Constitucionales [English transl.: The Normativity of Law, Dordrecht: Kluwer, in preparation].

BAy6N MOHINO, Juan Carlos (1994): Reasons for Action, Rule-Making and Rule­Fol/owing, paper prepared for the III International Seminary on Legal Theory, Alicante, Spain, November 1994 (unpublished manuscript).

BENTHAM, Jeremy (1970): Of Laws in General, ed. H. L. A. Hart, London: Athlone.

BETTI, Emilio (1990): Teoria Generale della Interpretazione, Milan: Giuffre.

BICKEL, Alexander M. (1962): The Least Dangerous Branch, New Haven, NJ: Yale University Press.

BIX, Brian (1992): The application (and mis-application) of Wittgenstein's rule-fol­lowing considerations to legal theory, in: Patterson (1992), 209-224.

BIX, Brian (1993): Law, Language, and Legal Determinacy, Oxford: Oxford University Press.

BIX, Brian (1995): Questions in legal interpretation, in: Marmor (1995), 137-154.

BJARUP, Jes (1988): Kripke's case, in: Rechtstheorie, 19, 39-49.

BLACKBURN, Simon (1981): Reply: Rule-following and moral realism, in: Holtzman! Leich (1981),163-190.

BLACKBURN, Simon (1984): Spreading the Word, Oxford: Oxford University Press.

BOBBIO, Norberto (1964): Des criteres pour resoudre antinomies, in: Dialectica, 18,237-258.

BOBBIO, Norberto (1965): Giusnaturalismo e Positivismo Giuridico, Milan: Commu­nita.

BOBBIT, Philip (1991): Constitutional Interpretation, Oxford: Basil Blackwell.

176 BIBLIOGRAPHY

BORK, Robert H. (1971): Neutral principles and some First Amendment problems, in: Indiana Law Journal, 47, 1-35.

BORK, Robert (1990): The Tempting of America. The Political Seduction of the Law, New York: Free Press.

BRAYBROOKE, David (1989): The conditions on which rules exist, in: Schilpp (1989).

BREST, Paul (1980): The misconceived quest for the original understanding, in: Boston University Law Review, 60, 204-238.

BRINK, David (1988): Legal theory, legal interpretation, and judicial review, in: Philo­sophy & Public Affairs, 17, 105-148.

BRINK, David (1989): Moral Realism and the Foundations of Ethics, Cambridge: Cambridge University Press.

BULYGIN, Eugenio (1976): Sobre la regIa de reconocimiento, in: Derecho, Filosofia y Lenguaje. Homenaje a Ambrosio L. Gioja, Buenos Aires: Astrea, 31-39.

BULYGIN, Eugenio (1982a): Norms, normative propositions and legal statements, in: G. Fl0istad (ed.), Contemporary Philosophy. A New Survey. Vol 3: Philosophy of Action, The Hague: Martinus Nijhoff, 107-125.

BULYGIN, Eugenio (1982b): Time and validity, in: A. A. Martino (ed.), Deontic Logic, Computational Linguistics and Legal Information Systems, Amsterdam: North Holland, 65-82.

BULYGIN, Eugenio (1984): La paradoja de la reform a constitucional, in: Alf Ross. Estu­dios en su Homenaje (Revista de Ciencias Sociales. Universidad de Valparaiso, Chile), 329-336.

BULYGIN, Eugenio (1990): An antinomy in Kelsen's legal theory, in: Ratio Iuris, 3, 29-45.

BULYGIN, Eugenio (1991a): Algunas consideraciones sobre los sistemas juridicos, in: Doxa, 9, 257-279.

BULYGIN, Eugenio (1991b): Sobre las normas de competencia, in: Alchourr6n/id. (1991),485-498.

BULYGIN, Eugenio (1992): Sull'interpretazione giuridica, in: Comanducci/Guastini (1992), 11-30.

BULYGIN, Eugenio (1993): L6gica De6ntica, in: Alchourr6n1MendeziOrayen (1993), 129-142.

BURNS, Linda C. (1991): Vagueness, Dordrecht: Kluwer.

CALSAMIGLIA, Albert (1992): La ret6rica de Critical Legal Studies. Impresiones de un lector espanol, in: Doxa, 11,295-310.

BmLIOGRAPHY 177

CAMPOS, Paul (1993): Three mistakes about interpretation, in: Michigan Law Journal, 92,388-397.

CARACCIOLO, Ricardo (1982): Conocimiento de Normas. Paper prepared for the Inter­national Congress of Philosophy of Law, La Plata (Argentina), October 1982 (un­published manuscript).

CARACCIOLO, Ricardo (1988): Sistema Juridico. Problemas Actuales, Madrid: Centro de Estudios Constitucionales.

CARACCIOLO, Ricardo (1991): Derecho y decisiones colectivas, in: Revista del Centro de Estudios Constitucionales, 10,25-48.

CARACCIOLO, Ricardo (1995): Due tipi di potere normativi, in: ComanduccilGuastini (1995), 199-218.

CARACCIOLO, Ricardo (1996): Esistenza di norme e di sistemi normativi, in: Coman­duccilGuastini (1996), 223-242.

CARNAP, Rudolf (1955): Meaning and synonymy in natural languages, in: Philosophi­cal Studies, 6, 33-47.

CARNAP, Rudolf (1963): Abraham Kaplan on value judgements, in: P. A. Schilpp (ed.), The Philosophy of Rudolf Carnap. The Library of Living Philosophers, La Salle, Ill: Open Court 1963, 999-1013.

CARRINO, Agostino (1992): Ideologia e Coscienza. Critical Legal Studies, Naples: Edizione Scientifiche Italiane.

CARRI6, Genaro R. (1965): Notas sobre Derecho y Lenguaje, Buenos Aires: Abeledo­Perrot.

CASEBEER, Kenneth M. (1977): Escape from liberalism: Fact and value in Karl N. Lle­wellyn, in: Duke Law Journal, 671-703.

CHOPER, Jesse (1980): Judicial Review in the National Political Process: A Functional Reconsideration of the Role of the Supreme Court, Chicago: University of Chi­cago Press.

COLEMAN, Jules L. and LEITER, Brian (1995): Determinacy, Objectivity, and Authority, in: Marmor (1995), 279-328.

COMANDUCCI, Paolo and GUASTINI, Riccardo (eds.) (1992): Analisi e Diritto 1992. Ricerche di Giurisprudenza Analitica, Torino: Giappichelli.

COMANDUCCI, Paolo and GUASTINI, Riccardo (eds.) (1994): Analisi e Diritto 1994. Ricerche di Giurisprudenza Analitica, Torino: Giappichelli.

COMANDUCCI, Paolo and GUASTINI, Riccardo (eds.) (1995): Analisi e Diritto 1995. Ricerche di Giurisprudenza Analitica, Torino: Giappichelli.

178 BIBLIOGRAPHY

COMANDUCCI, Paolo and GUASTINI, Riccardo (eds.) (1996): Struttura e dinamica dei sistemi giuridici, Torino: Giappichelli.

DAVIDSON, Donald (1967): Truth and meaning, in: id. (1984), 17-36.

DAVIDSON, Donald (1973): Radical interpretation, in: id. (1984), 125-140.

DAVIDSON, Donald (1979): Moods and performances, in: id. (1984),109-122.

DAVIDSON, Donald (1984): Inquiries into Truth and Interpretation, Oxford: Oxford University Press.

DAVIDSON, Donald (1990): The structure and content of truth, in: Journal of Philoso­phy, 87, 279-326.

DAY, Timothy J. (1992): Excluded middle and bivalence, in: Erkenntnis, 37, 93-97.

DUMMETI, Michael (1978): Truth and Other Enigmas, Cambridge, Mass.: Harvard University Press.

DUMMETI, Michael (1991): The Logical Basis of Metaphysics, Cambridge, Mass.: Harvard University Press.

DWORKIN, Ronald (1977a): Introduction, in: id. (ed.), The Philosophy of Law, Oxford: Oxford University Press 1977, 1-16.

DWORKIN, Ronald (1977b): No right answer?, in: P. M. S. Hacker and J. Raz (eds.), Law, Morality and Society. Essays in Honour of H. L. A. Hart, Oxford: Oxford University Press, 58-84.

DWORKIN, Ronald (1977c): Taking Rights Seriously, London: Duckworth.

DWORKIN, Ronald (1985): A Matter of Principle, Cambridge, MA: Harvard University Press.

DWORKIN, Ronald (1986): Law's Empire, Cambridge, Mass.: Harvard University Press.

DWORKIN, Ronald (1989): La completude du droit, in: P. Amselek and C. Grzegowski (eds.), Controverses autour de [,Ontologie du Droit, Paris: Presses Universitaires de France, 127-134.

DWORKIN, Ronald (1991): Pragmatism, right answers, and true banality, in: M. Brint and W. Weaver (eds.), Pragmatism in Law and Society, Boulder, San Franciso, and Oxford: Westview, 359-387.

DWORKIN, Ronald (1993): Life'S Dominion, New York: Alfred A. Knopf.

DWORKIN, Ronald (1996a): Indeterminacy and law, in: S. Guest (ed.), Positivism To­day, Aldershot: Dartmouth, 1-9.

DWORKIN, Ronald (1996b): Objectivity and truth: You'd better believe it, in: Philoso­phy & Public Affairs, 25,87-139.

BIBLIOGRAPHY 179

Eco, Umberto (1995): Interpretacion y Sobreinterpretaci6n, transl. J. G. L6pez Guix, Cambridge: Cambridge University Press.

EDGINGTON, Dorothy (1993): Wright and Sainsbury on higher-order vagueness, in: Analysis, 53, 193-200.

EEKELAAR, J. M. (1973): Principles of revolutionary legality, in: A. W. B. Simpson (ed.), Oxford Essays in Jurisprudence. Second Series, Oxford: Oxford University Press, 22-43.

ELSTER, Jon (1979): Ulysses and the Sirens, Cambridge: Cambridge University Press.

ELSTER, Jon and SLAGSTED, Rune (eds.) (1988): Constitutionalism and Democracy, Cambridge: Cambridge University Press.

ELY, John Hart (1980): Democracy and Distrust. A Theory of Judicial Review, Cam­bridge, Mass.: Harvard University Press.

ENDERTON, Herbert B. (1972): A Mathematical Introduction to Logic, London: Aca­demic Press.

ENDICOTT, Timothy (1994): Putting interpretation in its place, in: Law and Philosophy, 13,451-479.

ENDICOTT, Timothy (1996): Linguistic indeterminacy, in: Oxford Journal of Legal Studies, 16,667-697.

ENDICOTT, Timothy (1997): Ronald Dworkin y las consecuencias de la vaguedad, transl. J. J. Moreso and P. Navarro, in: Doxa, 19,375-383.

ENGEL, Pascal (1991): The Norm of Truth. An Introduction to Philosophy of Logic, London: Harvester Wheatsheaf.

ENGEL, Pascal (1992): Les concepts vagues sont-ils sans frontieres?, in: Revue Intema­tiona Ie de Philosophie, 46, 527-538.

EVANS, Gareth (1982): The Varieties of Reference, Oxford: Oxford University Press.

EVANS, Gareth (1985): The causal theory of names, in: Collected Papers, Oxford: Ox­ford University Press, 1-24.

EVANS, Gareth and McDOWELL, John (eds.) (1976), Truth and Meaning. Essays in Semantics, Oxford: Oxford University Press.

EZQUIAGA, Francisco J. (1987): La Argumentaci6n en la Justicia Constitucional Espanola, Ofiati: HAAE-IV AP.

FELDMAN, Louis E. (1992): Originalism through Raz-colored glasses, in: University of Pennsylvania Law Review, 140, 1384-1428.

FERRERES COMELLA, Victor (1994): EI principio de igualdad y el 'Derecho a no casar­se' (a prop6sito de la STC 22211992), in: Revista Espanola de Derecho Consti­tucional,42, 163-196.

180 BIBLIOGRAPHY

FERRERES COMELLA, Victor (1996): Judicial Review and Democracy, JSD Diss., Yale Law School.

FINE, Kit (1975): Vagueness, truth and logic, in: Synthese, 30, 265-300.

FINNIS, John (1973): Revolutions and continuity of law, in: A. W. B. Simpson (ed.), Oxford Essays in Jurisprudence. Second Series, Oxford: Oxford University Press, 44-76.

FINNIS, John (1987): On reason and authority in Law's Empire, in: Law and Philosophy, 6,357-380.

FISS, Owen (1982): Objectivity and interpretation, in: Stanford Law Review, 34, 739-763.

F0LLESDAL, Dagfinn and HILPINEN, Risto (1971): Deontic logic: An introduction, in: R. Hilpinen (ed.), Deontic Logic: Introductory and Systematic Readings, Dordrecht: Reidel, 1-35.

FRANK, Jerome (1930): Law and the Modern Mind, Gloucester, Mass.: Peter Smith (reprint 1970).

FREGE, Gottlob (1879): Begriffsschrift, Jena: Nebert (English transl. by S. Bauer-Men­gelber, in: 1. van Heijenoort (ed.), From Frege to Godel, Cambridge, Mass.: Harvard University Press, 1967).

FREGE, Gottlob (1970): Selections from the Philosophical Writings of Gottlob Frege, transl. P. Geach and M. Black, Oxford: Basil Blackwell.

FREEMAN, Samuel (1992): Original meaning, democratic interpretation and the consti­tution, in: Philosophy & Public Affairs, 21, 3-42.

fuLLER, Lon L.(1958): Positivism and fidelity to law: A reply to Professor Hart, in: Harvard Law Journal, 71, 630-672.

FULLER, Lon L.(1969): The Morality of Law, 2nd rev. ed. New Haven: Yale University Press.

GARRIDO, Manuel (1981): LOgica Simb6lica, Madrid: Tecnos.

GARZ6N VALDES, Ernesto (1994): Instituciones suicidas, in: Isegorfa, 9, 64-128.

GENTZEN, Gerhard (1934): Untersuchungen tiber das logische Schliessen, in: Mathe-matische Zeitschrift, 39, 176-210 and 410-431 (English transl.: Investigation into logical deduction, in: M. E. Szabo (ed.), The Collected Papers of Gerhard Gentzen, Amsterdam: North-Holland 1969,68-131).

GOLDING, Martin P. (1987): A note on discovery and justification in science and law, in: Justification (= Nomos 28), 124-140.

GORDON, Robert w. (1982): Holmes' Common Law as legal and social science, in: Hofstra Law Review, 10,719-746.

BIBLIOGRAPHY 181

GRANT, C. K. (1968): Imperatives and meaning, in: The Human Agent. Royal Institute of-Philosophy Lectures. Volume I, London: Macmillan, 181-195.

GRAY, John Chipman (1902): The Nature and Sources of Law, Gloucester, Mass.: Peter Smith (reprint 1972).

GRAYLING, A. C. (1982): An Introduction to Philosophical Logic, Brighton: Harrester.

GREENAWALT, Kent (1975): Discretion and judicial decision:The elusive quest for the fetters that bind judges, in: Columbia Law Review, 75, 359-398.

GREENAWALT, Kent (1992): Law and Objectivity, Oxford: Oxford University Press.

GREY, Thomas C. (1975): Do we have an unwritten constitution?, in: Stanford Law Re­view, 27, 703-718.

GREY, Thomas C. (1984): The constitution as scripture, in: Stanford Law Review, 37, 1-25.

GRICE, Paul (1975): Logic and conversation, in: P. Cole and J. Morgan (eds.), Syntax and Semantics 3: Speech Acts, New York: Academic Press, 41-58.

GRICE, Paul (1989): Studies in the Way of Words, Cambridge, Mass.: Harvard Univer­sity Press.

GUASTINI, Riccardo (1982): Problemi d'analisi dellinguaggio normativo, in: A. Ross (ed')' Critica del Diritto e Analisi del Linguaggio, Bologna: Il Mulino, 47-69.

GUASTINI, Riccardo (1991): Quindici Lezione di Diritto Costituzionale, Torino: Giap-pichelli.

GUASTINI, Riccardo (1992): Daile Fonti aile Norme, Torino: Giappichelli.

GUASTINI, Riccardo (1993): Le Fonti del Diritto e l'Interpretazione, Milan: Giuffre.

GUASTINI, Riccardo (1995): II Giudice e la Legge, Torino: Giappichelli.

GUIBOURG, Ricardo (1983): La autorreferencia normativa y la continuidad constitucio­nal, in: E. Bulygin et al. (eds.), El Lenguaje del Derecho. Homenaje a G. R. Carri6, Buenos Aires: Abeledo-Perrot, 181-196.

HAACK, Susan (1978): Philosophy of Logics, Cambridge: Cambridge University Press.

HABERMAS, Jiirgen (1996): Between Facts and Norms. Contributions to a Discourse Theory of Law and Democracy, transl. W. Regh, Cambridge: Polity Press.

HARE, R. M. (1952): The Language of Morals, Oxford: Oxford University Press.

HARMAN, Gilbert (1977): The Nature of Morality, New York: Oxford University Press.

HARRIS, J. W. (1971): When and why does the Grundnorm change?, in: Cambridge Law Journal, 29, 103-133.

HART, H. L. A. (1961): The Concept of Law, Oxford: Oxford University Press.

182 BIBLIOGRAPHY

HART, H. L. A. (1980): EI nuevo desaffo al positivismo juridico, transl. F. Laporta, L. Hierro and 1. R. Paramo, in: Sistema, 36 (May), 3-19.

HART, H. L. A. (1983): Essays in Jurisprudence and Philosophy, Oxford: Oxford Uni­versity Press.

HART, H. L. A. (1983a): American jurisprudence through English eyes: the Nightmare and the Noble Dream, in: id. (1983),123-144.

HART, H. L. A. (1983b): Self-referring laws, in: id. (1983), 170-178.

HART, H. L. A. (1994): Postscript, in: The Concept of Law, Second Edition, Oxford: Oxford University Press, 238-276.

HECK, Richard G. (1993): A note on the logic of (higher-order) vagueness, in: Analysis, 53,201-208.

HERNANDEZ MARiN, Rafael (1984): El Derecho como Dogma, Madrid: Tecnos.

HERNANDEZ MARiN, Rafael (1989): Teoria General del Derecho y de la Ciencia Juridi­ca, Barcelona: Promoci6n de Publicaciones Universitarias.

HIERRO S. PESCADOR, Jose (1990): Significado y Verdad. Ensayos de Semantica Filosofica, Madrid: Alianza.

HINTIKKA, Jaakko (1962): Knowledge and Belief. An Introduction to the Logic of the Two Notions, Ithaca and London: Cornell University Press.

HOERSTER, Norbert (1972): On Alf Ross's alleged puzzle in Constitutional Law, in: Mind, 81, 422-426.

HOFSTADTER, A. and MCKINSEY, J. C. (1939): On the logic of imperatives, in: Philo­sophy of Science, 6, 446-457.

HOLMES, Oliver Wendell (1920): The path of the law, in: id., Collected Legal Papers, New York: Harcourt.

HOLMES, Stephen (1988): Precommitment and the paradox of democracy, in: Elsterl Slagsted (1988), 195-240.

HOLTZMAN, Steven H. and LEICH, Christopher M. (eds.) (1981): Wittgenstein: To Follow a Rule, London: Routledge & Kegan Paul.

HONORE, Anthony (1967): Reflections on revolutions, in: The Irish Jurist, 2, 268-278.

HORWICH, Paul (1990): Truth, Oxford: Basil Blackwell.

HUGLY, Philip and SAYWARD, Charles (1993): Theories and truth-value gaps, in: Lin­guistics and Philosophy, 16,551-559.

HUME, David (1740): A Treatise of Human Nature, L. A. Selby-Bigge (ed.), Oxford: Oxford University Press (reprint 1978).

BIBLIOGRAPHY 183

HUNTER, Geoffrey (1993): The meaning of 'if in conditional propositions, in: The Phi-losophical Quarterly, 43, 279-297.

HURD, Heidi (1990): Sovereignty in silence, in: Yale Law Journal, 99, 945-1027.

HURLEY, Susan (1989): Natural Reasons, Oxford: Oxford University Press.

HYDE, Dominic (1994): Why higher-order vagueness is a pseudo-problem, in: Mind, 103,35-41.

JHERING, Rudolf von (1854): Teona de la Tecnica Jurfdica, transl. S. Sanjose, in: P. Casanovas and J. J. Moreso (eds.), El ambito de lo jur(dico. Lecturas de Pensa­miento Jurfdico Contemporaneo, Barcelona: Crftica-Grijalbo, 60-108 [German original: Der Geist des romischen Rechts aUf den verschiedenen Stufen seiner Entwicklung. Teil2.2, Aalen: Scientia 1968].

JORGENSEN, JOrgen (1938): Imperatives and logic, in: Erkenntnis, 7, 288-296.

KALINOWSKI, Georges (1975): LOgica del Discurso Normativo, transl. J. R. Capella, Madrid: Tecnos [French original: La logique des normes, Paris: Presses Univer­sitaires de France 1972].

KAMP, Hans (1975): Two theories about adjectives, in: E. Keenan (ed.), Formal Se­mantics and Natural Language, Cambridge: Cambridge University Press, 123-155.

KAYRIS, David (1982): Legal Reasoning, in: id. (ed.), The Politics of Law. A Progres­sive Critique, New York: Pantheon, 11-17.

KELMAN, Mark (1987): A Guide to Critical Legal Studies, Cambridge, Mass.: Harvard University Press.

KELSEN, Hans (1942): Judicial review of legislation, in: Journal of Politics, 4, 182-200.

KELSEN, Hans (1945): General Theory of Law and State, Cambridge, Mass.: Harvard University Press.

KELSEN, Hans (1960): Reine Rechtslehre, 2nd ed. Vienna: Franz Deuticke.

KELSEN, Hans (1973): Essays in Legal and Moral Philosophy, transl. P. Heat, ed. O. Weinberger, Dordrecht: Reidel.

KELSEN, Hans (1979): Allgemeine Theorie der Normen, Vienna: Manz.

KENNEDY, Duncan (1979): The structure of Blackstone's commentaries, in: Buffalo Law Review, 28, 211-382.

KIM, Jaegwon (1993): Supervenience as a philosophical concept, in: Supervenience and Mind, Cambridge: Cambridge University Press, 131-160.

KNEALE, William and KNEALE, Martha (1962): The Development of Logic, Oxford: Oxford University Press.

184 BIBLIOGRAPHY

KRESS, Ken (1989): Legal indeterminacy, in: California Law Review, 77,243-337.

KRIPKE, Saul (1971): Identity and necessity, in: M. K. Munitz (ed.), Identity and Indi­viduation, New York: New York University Press, 135-164.

KRIPKE, Saul (1980): Naming and Necessity, Oxford: Basil Blackwell.

KRIPKE, Saul (1982): Wittgenstein on Rules and Private Language, Oxford: Basil Blackwell.

KUTZ, Christopher L. (1994): Just disagreement: Indeterminacy and rationality in the rule oflaw, in: Yale Law Journal, 103,997-1030.

LANGILLE, Brian (1988): Revolution without foundation: The grammar of scepticism and law, in: McGill Law Journal, 33,451-484.

LAPORTA, Francisco (1993): Entre el Derecho y la Moral, Mexico-City: Fontamara.

LEFF, Arthur Allen (1979): Unspeakable ethics, unnatural law, in: Duke Law Journal, 1229-1249.

LEVINSON, Sanford (1982): Law as literature, in: Texas Law Review, 60, 373-404.

LEVINSON, Sanford (1988): Constitutional Faith, Princeton, NJ: Princeton University Press.

LEWIS, C. I. (1912): Implication and the algebra of logic, in: Mind, 21, 522-531.

LEWIS, C. I. (1914): The calculus of strict implication, in: Mind, 23, 240-247.

LEWIS, C. I. (1918): A Survey of Symbolic Logic, Berkeley: University of California.

LEWIS, David (1969): Convention. A Philosophical Study, Cambridge, Mass.: Harvard University Press.

LEWIS, David (1973): Counteifactuals, Oxford: Basil Blackwell.

LEWIS, David (1983): General semantics, in: Philosophical Papers. Volume I, Oxford: Oxford University Press, 189-232.

LIVINGSTON, Debra (1982): Round and round the bramble bush: From legal realism to C. L. Scholarship, in: Harvard Law Review, 95, 1669-1690.

LLEWELLYIN, Karl N. (1930): The Bramble Bush, New York: Oceana.

LLEWELLYN, Karl N. (1962): Jurisprudence. Realism in Theory and Practice, Chicago: University of Chicago Press.

LYONS, David (1982): Moral aspects of legal theory, in: Midwest Studies in Philosophy, 7,223-254.

LYONS, David (1993): Moral Aspects of Legal Theory. Essays on Law, Justice, and Political Responsability, Cambridge: Cambridge University Press.

BmLIOGRAPHY 185

MACCALLUM, Gerald C. (1968): Legislative intent, in: R. S. Summers (ed.), Essays in Legal Philosophy, Oxford: Basil Blackwell, 237-273.

MACCORMICK, Neil (1978): Legal Reasoning and Legal Theory, Oxford: Oxford University Press.

MACCORMICK, Neil (1994): Foreword, in: id., Legal Reasoning and Legal Theory, 2nd ed., Oxford: Oxford University Press, ix-xvii.

McDOWELL, John (1976): Truth-conditions, bivalence and verificationism, in: EvansJ McDowell (1976), 42-66.

McDoWELL, John (1981): Following a rule and ethics, in: HoitzmanlLeich (1981), 141-162.

MCGINN, Colin (1984): Wittgenstein on Meaning, Oxford: Basil Blackwell.

MACKIE, John (1977a): Ethics. Inventing Right and Wrong, London: Penguin.

MACKIE, John (1977b): The Third Theory of Law, in: Philosophy and Public Affairs, 7, 3-16.

MAKINSON, David (1981): Quantificational Reefs in Deontic Waters, in: R. Hilpinen (ed.), New Studies in Deontic Logic, Dordrecht: Reidel, 87-91.

MARMOR, Andrei (1992): Interpretation and Legal Theory, Oxford: Oxford University Press.

MARMOR, Andrei (ed.) (1995): Law and Interpretation, Oxford: Oxford University Press.

MARMOR, Andrei (1995a): Authorities and Persons, in: Legal Theory, 1,337-360.

MARMOR, Andrei (1995b): Preface, in: Marmor (1995), v-vii.

MAZZARESE, Tecla (1991): 'Norm proposition': Epistemic and semantic queries, in: Rechtstheorie, 22, 39-70.

MAZZARESE, Tecla (1996): Forme di Razionalita delle Decisioni Giudiziali, Torino: Giappichelli.

MAZZARESE, Tecla (1997): Fuzzy logic and judicial decision-making: the peril of the rationalist fallacy, in: E. Garz6n Valdes et al. (eds.), Normative Systems in Legal and Moral Theory. Festschrift for C. E. Alchourr6n and E. Bulygin, Berlin: Duncker & Humblot, 483-508.

MEHLBERG, H. (1958): The Reach of Science, Toronto: University of Toronto Press.

MENDONCA, Daniel (1992): Introducci6n al Analisis Normativo, Madrid: Centro de Estudios Constitucionales.

MENDONCA, Daniel (1996): Como Alicia en el Pais de las Maravillas. En defensa de las proposiciones normativas (unpublished manuscript).

186 BmLIOGRAPHY

MENDONCA, Daniel, MORESO, Jose Juan and NAVARRO, Pablo (1995): Intorno alle norme di competenza, in: ComanduccilGuastini (1995), 219-236.

MERKL, Adolf J. (1968): Das Problem der Rechtskontinuitiit und die Forderung des einheitlichen rechtlichen Weltbildes, in: H. Klekatsky, R. Marcie, H. Scharnbeck (eds.), Die Wiener rechtstheoretische Schule, Vienna: Europa-Verlag, 1267-1300.

MICHELMAN, Frank (1988): Law's republic, in: Yale Law Journal, 97, 1493-1537.

MOORE, G. E. (1922): External and internal relations, in: id., Philosophical Studies, London: Routledge & Kegan Paul, 276-309.

MOORE, Michael (1981): The semantics of judging, in: Southern California Law Re­view, 54, 151-294.

MOORE, Michael (1982): Moral reality, in: Wisconsin Law Review, 1061-1156.

MOORE, Michael (1985): A natural law theory of interpretation, in: Southern California Law Review, 58, 277-398.

MOORE, Michael (1987): Metaphysics, epistemology, and legal theory, in: Southern California Law Review, 60, 453-506.

MOORE, Michael (1989): The interpretive turn in modem theory: A tum for the worse?, in: Stanford Law Review, 41,871-957.

MOORE, Michael (1992a): Law as a functional kind, in: R. P. George (ed.), Natural Law Theory. Contemporary Essays, Oxford: Oxford University Press, 188-244.

MOORE, Michael (1992b): Moral reality revisited, in: Michigan Law Review, 90,2423-2523.

MORESO, Jose Juan (1991): Disposiciones de reforma constitucional, in: Doxa, 10,201-222.

MORESO, Jose Juan (1992): La Teoria del Derecho de Bentham, Barcelona: Promoci6n Publicaciones Universitarias.

MORESO, Jose Juan (1994a): Introducci6n a La Teorta del Derecho en el Sigloxx., in: P. Casanovas and J. J. Moreso (eds.), El Ambito de 10 Jurtdico. Lecturas de Pensamiento Jurtdico Contemporaneo, Barcelona: Crftica-Grijalbo, 351-360.

MORESO, Jose Juan (1994b): Verfassungswidrige Normen, in: Rechtstheorie, 25, 417-449.

MORESO, Jose Juan (1996): On relevance and justification of legal decisions, in: Er­kenntnis, 44, 73-100.

MORESO, Jose Juan and NAVARRO, Pablo, (1992): Normas permisivas, sistemas jurfdi­cos y clausura normativa, in: Theoria, 16-17-18, 1079-1100.

BIBLIOGRAPHY 187

MORESO, Jose Juan and NAVARRO, Pablo (1993a): Orden luridico y Sistema Juridico. Una Investigacion sobre la Dindmica de los Sistemas Juridicos, Madrid: Centro de Estudios Constitucionales.

MORESO, Jose Juan and NAVARRO, Pablo (1993b): Some remarks on the notions of le­gal order and legal system, in: Ratio Iuris, 6, 48-63.

MORESO, Jose Juan and NAVARRO, Pablo (1996a): Applicabilita ed efficacia delle nor­me giuridiche, in: Comanducci/Guastini (1996), 15-36.

MORESO, Jose Juan and NAVARRO, Pablo (1996b): Reception of norms and open legal systems, in: Stanley Paulson (ed.), Normativity and Norms. Perspectives on Kelsenian Themes, Oxford: Oxford University Press, forthcoming.

MORESO, Jose Juan and NAVARRO, Pablo (1996c): Verdad y eficacia, in: Theoria, 26, 105-124.

MORESO, Jose Juan, NAVARRO, Pablo, REDONDO, Cristina (1992): Argumentaci6n ju­ridica, 16gica y decisi6n judicial, in: Doxa, 11,247-262.

MORESO, Jose Juan, NAVARRO, Pablo, REDONDO, Cristina (1995): Ubernahme von Normen und Fortbestand des Rechts, in: Archiv fUr Rechts- und Sozialphiloso­phie, 81, 321-334.

MUNZER, Stephen (1977): Retroactive law, in: Journal of Legal Studies, 6, 373-397.

MUNZER, Stephen (1985): Realistic limits on realist interpretation, in: Southern Cali­fornia Law Review, 58, 459-475.

MUNZER, Stephen and NICKEL, James W. (1977): Does the Constitution mean what it always meant?, in: Columbia Law Review, 77, 1029-1062.

NAVARRO, Pablo (1993): Sistemajurfdico, casos diffciles y conocimiento del Derecho, in: Doxa, 14,243-269.

NIINILUOTO, Ilkka (1981): On truth and argumentation in legal dogmatics, in: A. Aar­nio, I. Niiniluoto, J. Uusitalo (eds.), Methodologie und Erkenntnistheorie der juristischen Argumentation, Berlin: Duncker & Humblot, 3-76.

NINO, Carlos S. (1985): La Validez del Derecho, Buenos Aires: Astrea.

NINO, Carlos S. (1992): Fundamentos de Derecho Constitucional, Buenos Aires: As­trea.

NINO, Carlos S. (1994): Derecho, Moral y Politica. Una Revision de la Teoria General del Derecho, Barcelona: Ariel.

ORA YEN, Raul (1989): La 16gica relevante contra la deducibilidad chisica, in: id., L6gi­ca, Significado y Ontologia, Mexico-City: UNAM, 219-263.

PARODI, Gianpaolo (1996): Lacune e norme inespresse nella giurisprudenza costituzio­nale, in: Comanducci/Guastini (1996), 87-115.

188 BIBLIOGRAPHY

PATTERSON, Dennis M. (ed.) (1992a): Wittgenstein and Legal Theory, Boulder, San Francisco, Oxford: Westview.

PATTERSON, Dennis M. (1992b): Law's pragmatism: Law as practice and narrative, in: Patterson (1992a), 85-124.

PAITERSON, Dennis M. (1996): Law and Truth, Oxford: Oxford University Press.

PAULSON, Stanley (1988): An empowerment theory of legal norms, in: Ratio Iuris, 1, 58-72.

PEACOCKE, Christopher (1976): Truth definitions and actual languages, in: EvanslMc­Dowell (1976),162-188.

PEcEs-BARBA MARTiNEZ, Gregorio (1984): Los Valores Superiores, Madrid: Tecnos.

PEcEs-BARBA MARTiNEZ, Gregorio (1993): Los valores superiores, in: id., Derecho y Derechos Fundamentales, Madrid: Centro de Estudios Constitucionales.

PEREZ LLED6, Juan A. (1996): El Movimiento Critical Legal Studies, Madrid: Tecnos.

PLAITS, Mark (1979): Ways of Meaning. London: Routledge & Kegan Paul.

POSNER, Richard (1990): The Problems of Jurisprudence, Cambridge, Mass.: Harvard University Press.

POUND, Roscoe (1908): Mechanical jurisprudence, in: Columbia Law Review, 8, 605-638.

PRESBY, Shannon (1994): Interpretivism naturalized: Dworkin's minimalist metaphy­sics, in: Canadian Journal of Law and Jurisprudence, 7, 303-330.

PRIETO SANCHIS, Luis (1987): Ideologia e Interpretacion Juridica, Madrid: Tecnos.

PUTNAM, Hilary (1973): Meaning and reference, in: Journal of Philosophy, 70, 699-711.

PuTNAM, Hilary (1975): The meaning of 'meaning', in: id., Mind. Language. and Reality, Cambridge: Cambridge University Press.

PUTNAM, Hilary (1981): Reason. Truth and History, Cambridge: Cambridge University Press.

PUTNAM, Hilary (1983): Vagueness and alternative logic, in: id., Realism and Reason, Cambridge: Cambridge University Press, 271-286.

PUTNAM, Hilary (1987): The Many Faces of Realism, La Salle, Ill.: Open Court.

PUTNAM, Hilary (1990): Realism with a Human Face, Cambridge, Mass.: Harvard University Press.

PUTNAM, Hilary (1995): Are legal and moral values made or discovered?, in: Legal Theory, 1, 1-19.

BIBLIOGRAPHY 189

QUINE, W. V. o. (1960): Word and Object, Cambridge, Mass.: MIT Press.

QUINE, W. V. o. (1969): Reply to 'On Saying That', in: D. Davidson and G. Harman (eds.), Words and Objections, Dordrecht: Reidel.

QUINE, W. V. O. (1970): Philosophy of Logic, Englewood Cliffs, NJ: Prentice-Hall.

QUINE, W. V. O. (1972): Methods of Logic, New York: Holt, Rinehart and Winston.

QUINE, W. V. O. (1981): What price bivalence?, in: id., Theories and Things, Cam-bridge, Mass.: Harvard University Press, 31-37.

RADIN, Margaret Jane (1992): Reconsidering the rule of law, in: Patterson (1992), 125-156.

RADIN, Max (1930): Statutory interpretation, in: Harvard Law Review, 43,863-885.

RAWLS, John (1971): A Theory of Justice, Cambridge, Mass.: Harvard University Press.

RAWLS, John (1987): The idea of an overlapping consensus, in: Oxford Journal of Le-gal Studies, 7, 1-25.

RAWLS, John (1993): Political Liberalism, New York: Columbia University Press.

RAZ, Joseph (1970): The Concept of a Legal System, Oxford: Oxford University Press.

RAZ, Joseph (1972a): Legal principles and the limits of law, in: Yale Law Journal, 81, 823-854.

RAZ, Joseph (1972b): Professor Alf Ross and some legal puzzles, in: Mind, 81, 415-421.

RAZ, Joseph (1979): The Authority of Law, Oxford: Oxford University Press.

RAZ, Joseph (1980): Postscript: Sources, normativity, and individuation, in: id., The Concept of a Legal System, 2nd ed., Oxford: Oxford University Press.

RAZ, Joseph (1986a): Dworkin: A new link in the chain, in: California Law Review, 74, 1103-1119.

RAZ, Joseph (1986b): The Morality of Freedom, Oxford: Oxford University Press.

RAz, Joseph (1992): The relevance of coherence, in: Boston University Law Review, 72, 273-321.

RAZ, Joseph (1993): On the autonomy of legal reasoning, in: Ratio [uris, 6, 1-15.

RAZ, Joseph (1994): Ethics and the Public Domain, Oxford: Oxford University Press.

RAZ, Joseph (1994a): Authority, law, and morality, in: id. (1994), 194-221.

RAZ, Joseph (1994b): The problem about the nature of law, in: id. (1994), 179-193.

RAZ, Joseph (1995): Interpretation without retrieval, in: Marmor (1995), 155-176.

190 BIBLIOGRAPHY

RAZ, Joseph (1996a): Intention in interpretation, in: R. P. George (ed.), The Autonomy of Law, Essays on Legal Positivism, Oxford: Oxford University Press, 249-286.

RAZ, Joseph (1996b): Why interpret?, in: Ratio Iuris, 9,349-363 ..

REICHENBACH, Hans (1951): The Rise of Scientific Philosophy, Berkeley: University of California Press.

RICHARDS, David (1983): Constitutional interpretation, history, and the death penalty: A book review, in: California Law Review, 71, 1373-1398.

Ross, Alf (1941): Imperatives and logic, in: Theoria, 7, 53-71.

Ross, Alf (1958): On Law and Justice, London: Stevens & Sons.

Ross, Alf (1968): Directives and Norms, London: Routledge & Kegan Paul.

Ross, Alf (1969): On self-reference and a puzzle in Constitutional Law, in: Mind, 78, 1-24.

RUIZ MANERO, Juan (1990): Jurisdicci6n y Normas, Madrid: Centro de Estudios Cons­titucionales.

RUMFITT, Ian (1995): Truth Conditions and Communication, in: Mind, 104,827-842.

RUSSELL, Bertrand (1919): Descriptions, in: A. W. Moore (ed.), Meaning and Refe­rence, Oxford: Oxford Universiy Press 1993,46-55.

RUSSELL, Bertrand (1923): Vagueness, in: Australasian Journal of Philosophy and Psychology, 1,84-92.

RUSSELL, Bertrand and WHITEHEAD, A. N. (1910): Principia Mathematica, Cambridge: Cambridge University Press.

SAINSBURY, R. M. (1991): Is there higher-order vagueness?, in: Philosophical Quar­terly, 41,167-182.

SAINSBURY, R. M. (1995): Paradoxes, Cambridge: Cambridge University Press.

SANFORD, David H. (1989): If P, then Q. Conditionals and the Foundations of Rea­soning, London and New York: Routledge.

SATTA, Salvatore (1968): Soliloqui e Colloqui di un Giurista, Padova: CEDAM.

SCALIA, Antonin (1989): Originalism: the lesser evil, in: University of Cincinnati Law Review, 57,849,866.

SCHAUER, Frederick (1988): Formalism, in: Yale Law Journal, 509-548.

SCHAUER, Frederick (1991): Playing by the Rules, Oxford: Oxford University Press.

SCHAUER, Frederick (1992a): Rules and the rule-following argument, in: Patterson (1992), 225-232.

BIBLIOGRAPHY 191

SCHAUER, Frederick (1992b): The occasions of constitutional interpretation, in: Boston University Law Review, 72, 729-746.

SCHAUER, Frederick (1993): Introduction, in: id. (ed.), Law and Language, Aldershot: Dartmouth.

SCHEFFLER, Israel (1979): Beyond the Letter, London: Routledge & Kegan Paul.

SCHILPP, P. A. (ed.) (1989), The Philosophy of Georg Henrik von Wright, La Salle, Ill: Open Court.

SCHLEGEL, John (1979): American legal realism and empirical social science: From the Yale experience, in: Buffalo Law Review, 28, 459-586.

SCHMITT, Carl (1928): Verfassungslehre, MunichlLeipzig: Duncker & Humblot.

SCHURZ, Gerhard (1991): Relevant deduction. From solving paradoxes towards a gene­ral theory, in: Erkenntnis, 35, 391-437.

SCHURZ, Gerhard and LAMBERT, Karl (1994): Outline of a theory of scientific under­standing, in: Synthese, 101,65-120.

SEARLE, John (1969): Speech Acts. An Essay in the Philosophy of Language, Cam­bridge: Cambridge University Press.

SEARLE, John (1979): A taxonomy of illocutionary acts, in: id., Expression and Mean­ing, Cambridge: Cambridge University Press, 1-29.

SHINER, Roger A. (1992): Norm and Nature. The Movements of Legal Theory, Oxford: Oxford University Press.

SINGER, Joseph William (1984): The player and the cards: Nihilism and legal theory, in: Yale Law Journal, 94, 1-70.

SMART, J. J. C. (1984): Ethics, Persuasion and Truth, London: Routledge & Kegan Paul.

SMITH, Gene Anne (1992): Wittgenstein and the sceptical fallacy, in: Patterson (1992), 157-188.

SMITH, Michael (1994): The Moral Problem, Oxford: Basil Blackwell.

SPAAK, Tarben (1994): The Concept of Legal Competence: An Essay in Conceptual Analysis, Aldershot: Dartmouth.

SPERBER, Dan and WILSON, Deirdre (1986a): Relevance: Communication and Cog­nition, Oxford: Basil Blackwell.

SPERBER, Dan and WILSON, Deirdre (1986b): On defining relevance, in: R E. Grandy and R. Warner (eds.), Philosophical Grounds of Rationality, Oxford: Oxford University Press, 243-258.

192 BIBLIOGRAPHY

STRAWSON, P. F. (1950): On referring, in: A. W. Moore (ed.), Meaning and Reference, Oxford: Oxford Universiy Press, 1993,56-79.

STRAWSON, P. F. (1967): Singular terms and predication, in: id. (ed.), Philosophical Logic, Oxford: Oxford University Press, 69-88.

STRAWSON, Peter (1971): Logico-Linguistic Papers, London: Methuen.

SUBER, Peter (1990): The Paradox of Self-Amendment, New York: Peter Lang.

SUMMERS, Robert S. (1982): Instrumentalism and American Legal Theory, Ithaca, NY: Cornell University Press.

SUPPES, Patrick (1957): Introduction to Logic, New York: Van Nostrand.

TARELLO, Giovanni (1962): II Realismo Giuridico Americano, Milan: Giuffre.

TARELLO, Giovanni (1980): L'Interpretazione della Legge, Milan: Giuffre.

TARSKI, Alfred (1944): The semantic conception of truth and the foundations of seman­tics, in: Philosophy and Phenomenological Research, 4, 151-181.

TARSKI, Alfred (1956): The conception of truth in formalized languages, in: id., Logic, Semantics, Metamathematics. Papers from 1923 to 1938, transl. J. H. Woodger, Oxford: Oxford University Press, 152-278.

TRIBE, Laurence H. (1985): Constitutional Choices, Cambridge, Mass.: Harvard Uni­versity Press.

TRILLAS, E., ALSINA, c., TERRICABRAS, J. M. (1995): Introduccion a la LOgica Borrosa, Barcelona: Ariel.

TuSHNET, Mark (1983): Following the rules laid down: A critique of interpretivism and neutral principles, in: Harvard Law Journal, 96, 781-827.

TuSHNET, Mark (1986): Critical Legal Studies: An introduction to its origins and un­derpinnings, in: Journal of Legal Studies, 36, 505-517.

TUSHNET, Mark (1988): Red, White, Blue. A Critical Analysis of Constitutional Law, Cambridge, Mass.: Harvard University Press.

TwINING, William (1973): Karl Llewellyn and the Realist Movement, London: Wei­denfeld & Nicholson.

TYE, Michael (1994): Why the vague need not be higher-order vague?, in: Mind, 103, 43-45.

UNGER, Peter (1979a): There are no ordinary things, in: Synthese, 41,117-154.

UNGER, Peter (1979b): I do not exist, in: G. MacDonald (ed.), Perception and Identity: Essays Presented to A. J. Ayer, London: Macmillan, 235-251.

UNGER, Roberto M. (1975): Knowledge and Politics, New York: The Free Press.

BIBLIOGRAPHY 193

UNGER, Roberto M. (1983): The Critical Legal Studies movement, in: Harvard Law Journal, 96, 563-675.

V ALLEIO, Jesus (1993-1994): Ortograffa y heterograffa constitucionales, in: Anuario de Historia del Derecho Espanol, 63-64, 615-669.

VAN FRAASSEN, B. C. (1966): Singular terms, truth-value gaps and free logic, in: Jour­nal of Philosophy, 63, 136-152.

VERNENGO, Roberto J. (1993): On the logical interpretation of legal sentences, in: Ar­chiv for Rechts- und Sozialphilosophie, 93, 2-15.

WAISMANN, Frederick (1951): Verifiability, in: A. N. G. Flew (ed.), Logic and Lan­guage, Oxford: Basil Blackwell, 117-144.

WALDRON, Jeremy (1992): The irrelevance of moral objectivity, in: R. P. George (ed.), Natural Law Theory. Contemporary Essays, Oxford: Oxford University Press, 158-187.

WALDRON, Jeremy (1994): Vagueness in law and language: Some philosophical issues, in: California Law Review, 82,510-540.

WALDRON, Jeremy (1995): Legislator's intentions and unintentional legislation, in: Marmor (1995), 329-356.

WALUCHOW, W. J. (1994): Inclusive Legal Positivism, Oxford: Oxford University Press.

WASSERSTROM, Richard A. (1961): The Judicial Decision: Toward a Theory of Legal Justification, Stanford: Stanford University Press.

WECHSLER, Herbert (1959): Toward neutral principles of Constitutional Law, in: Har­vard Law Review, 73, 1-35.

WEINBERGER, Ota (1984): 'Is' and 'ought' reconsidered, in: Archiv for Rechts- und Sozialphilosophie, 70, 454-474.

WEINBERGER, Ota (1991): The logic of norms founded on descriptive language, in: Ra­tio Iuris, 4,284-307.

WIGGINS, David (1987): Needs, Values, and Truth, Oxford: Basil Blackwell.

WILLIAMS, Bernard (1973): Consistency and realism, in: id., Problems of the Self, Cambridge: Cambridge University Press, 187-206.

WILLLIAMS, Glanville (1946): Language and the law, in: Law Quarterly Review, 61, 185-195 and 293-302.

WILLIAMSON, Timothy (1994): Vagueness, London: Routledge.

WINCH, Peter (1958): The Idea of a Social Science, London: Routledge & Kegan Paul.

194 BIBLIOGRAPHY

WITTGENSTEIN, Ludwig (1921): Tractatus Logico-Philosophicus, transl. D. F. Pears and B. F. McGuinness, London and New York: Routledge & Kegan Paul 1974.

WITTGENSTEIN, Ludwig (1953): Philosophical Investigations, transl. G. E. M. Ans­combe, Oxford: Basil Blackwell.

WOOZLEY, A. D. (1984): No right answer, in: M. Cohen (ed.), Ronald Dworkin and Contemporary Jurisprudence, Totowa, NJ: Rowman & Allanheld, 173-181.

WRIGHT, Crispin (1975): On the coherence of vague predicates, in: Synthese, 30, 325-366.

WRIGHT, Crispin (1976): Language-mastery and the sorites paradox, in: EvanslMcDo­well (1976), 223-247.

WRIGHT, Crispin (1989): Critical notice, in: Mind, 98, 289-305.

WRIGHT, Crispin (1992a): Is higher-order vagueness coherent?, in: Analysis, 52, 129-139.

WRIGHT, Crispin (1992b): Truth and Objectivity, Cambridge, Mass.: Harvard Univer­sity Press.

WRIGHT, Crispin (1993): Realism, Meaning, and Truth, 2nd ed., Oxford: Basil Black­well.

WRIGHT, Georg Henrik von (1951): Deontic logic, in: Mind, 60,1-15.

WRIGHT, Georg Henrik von (1957): The concept of entailment, in: id., Logical Studies, London: Routledge & Kegan Paul, 165-191.

WRIGHT, Georg Henrik von (1963a): Norm and Action. A Logical Enquiry, London: Routledge & Kegan Paul.

WRIGHT, Georg Henrik von (1963b): The Varieties o/Goodness, London: Routledge & Kegan Paul.

WRIGHT, Georg Henrik von (1980): Problems and prospects of deontic logic. A survey, in: E. Agazzi (ed.): Modern Logic. A Survey, Dordrecht: Reidel 1980,399-423.

WRIGHT, Georg Henrik von (1983): Prqctical Reason. Philosophical Papers. Vol I., Oxford: Basil Blackwell.

WRIGHT, Georg Henrik von (1983a): Norms, truth and logic, in: id. (1983), 130-209.

WRIGHT, Georg Henrik von (1983b): The foundations of norms and normative state­ments, in: id. (1983), 67-82.

WRIGHT, Georg Henrik von (1984): Truth, Knowledge, and Modality. Philosophical Papers. Volume III, Oxford: Basil Blackwell.

WRIGHT, Georg Henrik von (1984a): Demystifying propositions, in: id. (1984), 14-25.

WRIGHT, Georg Henrik von (1984b): The logic of predication, in: id. (1984), 42-51.

BIBLIOGRAPHY 195

WRIGHT, Georg Henrik von (1984c): Truth and logic, in: id. (1984), 26-41.

WRIGHT, Georg Henrik von (1988): Truth-Logic. Paper presented to the Sociedad Ar­gentina de Amilisis Filos6fico, Buenos Aires (unpublished manuscript).

WRIGHT, Georg Henrik von (1989): Logiche della verita, transl.. G. Pezzini, in: Infor­matica e Diritto, 1, 15-39.

WROBLEWSKI, Jerzy (1971): Legal decision and its justification, in: H. Hubien (ed.), Le raisonnement juridique. Akten des Weltkongresses fUr Rechts- und Sozial­philosophie, Brussels, 409-419.

WROBLEWSKI, Jerzy (1985): Constitucion y Teor{a General de la Interpretacion Ju­rfdica, Madrid: Civitas.

WROBLEWSKI, Jerzy (1992): The Judicial Application of Law, Dordrecht: Kluwer.

YABLON, Charles (1987): Law and metaphysics, in: Yale lAw Journal, 96, 613-642.

ZADEH, L. A. (1965): Fuzzy sets, in: Information and Control, 8, 338-353.

ZIEMBA, Zdzislaw (1976): Appendix on deontic logic, in: Z. Ziembinski (ed.), Practical Logic, Dordrecht: Reidel, 360-430.

INDEX OF NAMES

Aarnio, A. 16, 155 Ackerman, B. 118, 167 Alarc6n Cabrera, C. 119 A1chourr6n, Carlos E. 3, 5, 10, 26, 45, 48,

57 f., 60 ft., 86 ff., 92 ff., 97, 101, 103 f., 109, 112 ff., 120, 122, 124 f.,

Alexander, L. 161 Alexy, R. 16 Allen, W. 164 Alonso, J. P. 59 Alonso Garcia, E. 168 Alsina, C. 77 Altman, A. 149 Amar, A. R. 131 Anderson, A. R. 19,33 f., 38 ff., 42 Anscombe, G. E. M. 13 Aristotle 26 Ascarelli, T. 151 Atienza, M. 45, 101, 138, 153 Austin, J. L. 12, 105

Backer, G. P. 154 Barnes, 1. 146 Bayon Mohino, J. C. 159, 169 Belnap, N. D. 19,33 f., 38 ft., 42 Bentham, J. 26, 101 Betti, E. 166 Bickel, A. M. 167 Bix, B. 73, 144, 148, 154, 161 Bjarup, J. 154 Blackburn,S. 50,65,142,154 Bobbio, N. 113 f. Bobbit, Ph. 159 Borges, J. L. 146 Bork, R. 160, 167 Braybrooke, D. 131 Brest, P. 160 Brink, D. 134, 136 f., 143 Brouwer, L. G. J. 54 Bulygin, E. 3, 5, 26, 57 f., 60 ff., 86 ff., 97,

101 f., 107ff., 111, 113, 119 f., 122, 124 ff., 147, 158

Bums, L. 76, 86

Calsamiglia, A. 149 Campos, P. 159 Caracciolo, R. 5, 101, 104, 118, 149 f. Carnap, R. 7 f., 73 Carrino, A. 149 Carrio, G. R. 73 Casebeer, K. M. 149 Chomski, N. 14 Choper,1. 167 Coleman, J. L. 159

Dante 146 Davidson, D. 7, 10 ff., 14, 166 Day, T. 84 De Morgan, A. 36 f. Dickens, Ch. 139 Dummett, M. 6 f., 26,49 f., 54, 74 ff., 83,

86 Dworkin, R. 50 f., 57, 59, 63 ff., 72, 82, 86,

99,109,134 f., 139 ff., 145 ff., 159, 161,165 ff.

Eco, U. 164 Edgington, D. 82 Eekelaar, J. M. 112 Elster, 1. 117 f. Ely, J. H. 167 Enderton, H. B. 29 Endicott, T. 65, 74, 154, 156, 161, 164 Engel,P.49,78,82 Eubulides 73 Evans, G. 136, 144 Ezquiaga, F. 1. 163

Feldman, L. E. 161 Ferreres Comella, V. 168 Fine, K. 83 Finnis, J. 111 f., 147 Fiss, 0.159 Flaubert, G. 50, 146 Fpllesdal, D. 44 f. Fraassen, B. C. van 83 Frank, J. 147 ff., 153

198 INDEX OF NAMES

Freeman, S. 161 Frege, G. 6,14,36,76,81 Freud, S. 148 Fuller, L. L. 72, 111, 144

Garda de Entema, E. 161 Gardenfors, P. 114 Garrido, M. 29 Garz6n Valdes, E. 118 Gentzen, G. 32, 38 Golding, M. P. 153 Gordon, R. W. 149 Grant, C. K. 23 Gray, J. Ch. 151 Grayling, A. C. 24 Greenawalt, K. 146, 154 Grey, Th. C. 166 f. Grice, P. 10, 14 ff., 32, 95 Guastini, R. 106, 118 f., 125, 151 Guibourg, R. 119

Haack, S. 18,37,51,54 Habermas, J. 169 Hacker, P. M. S. 154 Harman, G. 137, 143 Harris, J. W. 112 Hart, H. L. A. 24, 72 f., 101, 104, 109,

III f., 119, 132 f., 139, 141, 146, 152 ff., 156, 158, 165, 169 f.

Heck, R. G. 82 Hernandez Mann, R. 7, 23, 40, 49, 60, 101,

131,161 Heyting, A. 54 Hierro S. Pescador, 1. 7 Hilpinen, R. 44 f. Hintikka, J. 5 Hoadly, Bishop 151 Hoerster, N. 119 Hofstadter, A. 7,9 Holmes, O. W. 148 Holmes, St. 117, 129 Honore, A. 112 Horwich, A. 80 f. Hugly, Ph. 11 f. Hume, D. 9, 118, 150

Hunter, G. 36,40 Hurd, H. 134, 137 f., 143, 161 Hurley, S. 147 Hyde, D. 82

Jhering, R. von 134 JOrgensen, J. 9 f.

Kalinowski, G. 26 Kamp, H. 83 Kayris, D. 149 Kelman, M. 149 Kelsen, H. 1,24,57,59 ff., 65, 72, 87, 101,

106,111,119,158,164,168 Kennedy, D. 149 Kim, J. 6 Kneale, M. 73 Kneale, W. 73 Kress, K. 155 Kripke, S. 120, 136, 144, 149 f., 154 Kutz, Ch. L. 159

Lambert, K. 114 Langille, B. 149 Laporta, F. J. 170 Leff, A. A. 169 Leiter, B. 159 Levinson,S. 164, 166 Lewis, C. I. 36 f. Lewis, D. 12,81,83, 120, 158 f. Livingston, D. 149 Llewellyn, K. N. 148 Lyons, D. 17,72,161

MacCallum, G. C. 161 f. MacCormick, N. 16,64,105 McDowell, J. 12, 154 McGinn, C. 154 McKinsey, J. C. 7, 9 Mackie,1. 137, 143, 147 Makinson, D. 23,112, 114 Marco Polo 144 Marmor, A. 1,11,73, 144, 154 ff., 161,

163 Martino, A. A. 10

INDEX OF NAMES 199

Mazzarese, T. 78, 152 f. Mehlberg, H. 83 Mendonca, D. 3, 101, 152 Merkl, A. 104 Michelman, F. 167 Moore, G. E. 18 Moore, M. 131, 134 ff., 143 ff., 161 Moreso, J. J.3, 5,17,26,51,60, 101 f.,

104, 110 f., 113, 119, 154 Munzer, St. 106, 146, 165, 167

Navarro, P. E. 3, 5,17,60,101 f., 104, 111, 168

Nickel, J. W. 165,167 Niiniluoto, I. 155 Nino, C. S. 118 f., 153, 161

Orayen, R. 42

Parodi, G. 125 Patterson, D. M. 50 Paulson, St. 101 Peacocke, Ch. 11 Peces-Barba Martinez, G. 126 Perez Lled6, J. A. 149 Plato 54, 99, 147, 154 Platts, M. 6 f., 11 f., 14 f., 30, 76 Posner, R. 161 Pound, R. 134 Presby, Sh. 146 Prieto Sanchis, L. 169 Putnam, H. 83, 136 f., 143 ff.

Quine, W. V. O. 10,28,30,42,58, 78, 90, 166

Radin, M. J. 154 Radin, M. 155, 161 Rawls, J. 90, 118, 165, 169 Raz, J. 44,49,59,66 ff., 82, 101 f., 104,

107, 109 f., 112, 119, 139, 147, 160 f., 163f.,170

Redondo, C. 17, 104, 111 Reichenbach, H. 153 Richards, D. 145

Ross, A. 7, 24, 39, 41, 73,101,119,158 Ruiz Manero, J. 62 f., 101 Rumfitt, I. 15 Russell, B. 36,73,91 f., 94 ff., 99, 140

Sainsbury, R. M. 74,78, 82 f., 85 Sanford, D. H. 40 Satta,S. 151 Sayward, Ch. 11 Scalia, A. 160 Schauer, F. 73, 134, 144, 152 ff., 161, 169 Scheffler, I. 73 Schlegel, J. 149 Schmitt, C. 19 Schurz, G. 41 ff., 114 Searle, J. 7,12 f., 50, 95 Shakespeare, W. 139, 145, 164 Shiner, R. A. 161 Singer, J. W. 77 Smart, J. J. C. 7, 23 Smith, G. A. 154 Smith, M. 137 Spaak, T. 101 Sperber, D. 32 ff. Strawson, P. F. 10 f., 14,91 ff., 98 f., 140 Suber, P. 119 Summers, R. S. 148 Suppes, P. 18

Tare11o, G. 148, 161 Tarski, A. 7, 24 Terricabras, J. M. 77 Tribe, L. H. 167 Trillas, E. 77 Tushnet, M. 149 f., 168 Twining, W. 148 Tye, M. 82, 85

Ugolino de Pisa 146 Unger, P. 75 ff. Unger, R. M. 149

Vallejo, J. 131 Vemengo, R. J. 152

200

Waismann, F. 73 Waldron, J. 73, 145, 161 Waluchow, W. J. 110 Wang, H. 74 Wasserstrom, R. A. 153 Wechsler, H. 167 Weinberger, O. 9, 23 Whitehead, A. N. 36 Wiggins, D. 145 Williams, B. 24 Williams, G. 73 Williamson, T. 58, 79 f., 83, 85, 156 Wilson, D. 32 ff. Winch, P. 150

INDEX OF NAMES

Wittgenstein, L. 7,13 f., 73,144, 147, 149 f., 154 f., 157, 165

Woozley, A. D. 148 Wright, C. 50,75 f., 82, 87,134 f., 147,

154 Wright, G. H. von 1, 3, 5, 8, 17 f., 25 f., 31,

37,41 f., 48, 51 ff., 60, 70, 81, 89, 97, 101, 103

Wroblewski, J. 16, 156

Yablon, Ch. 149

Zadeh, L. A. 77 Ziemba, Z. 23

Law and Philosophy Library

1. E. Bulygin, J.-L. Gardies and I. Niiniluoto (eds.): Man, Law and Modern Forms of Life. With an Introduction by M.D. Bayles. 1985 ISBN 90-277-1869-5

2. W. Sadurski: Giving Desert Its Due. Social Justice and Legal Theory. 1985 ISBN 90-277-1941-1

3. N. MacConnick and O. Weinberger: An Institutional Theory of Law. New Approaches to Legal Positivism. 1986 ISBN 90-277-2079-7

4. A Aarnio: The Rational as Reasonable. A Treatise on Legal Justification. 1987 ISBN 90-277-2276-5

5. M.D. Bayles: Principles of Law. A Nonnative Analysis. 1987 ISBN 90-277-2412-1; Pb: 9O-277-2413-X

6. A Soeteman: Logic in Law. Remarks on Logic and Rationality in Nonnative Reasoning, Especially in Law. 1989 ISBN 0-7923-0042-4

7. C.T. Sistare: Responsibility and Criminal Liability. 1989

8. A Peczenik: On Law and Reason. 1989

9. W. Sadurski: Moral Pluralism and Legal Neutrality. 1990

10. M.D. Bayles: Procedural Justice. Allocating to Individuals. 1990

ISBN 0-7923-0396-2

ISBN 0-7923-0444-6

ISBN 0-7923-0565-5

ISBN 0-7923-0567-1

11. P. Nerhot (ed.): Law, Interpretation and Reality. Essays in Epistemology, Henneneutics and Jurisprudence. 1990 ISBN 0-7923-0593-0

12. AW. Norrie: Law, Ideology and Punishment. Retrieval and Critique of the Liberal Ideal of Criminal Justice. 1991 ISBN 0-7923-1013-6

13. P. Nerhot (ed.): Legal Knowledge and Analogy. Fragments of Legal Epistemology, Her-meneutics and Linguistics. 1991 ISBN 0-7923-1065-9

14. O. Weinberger: Law, Institution and Legal Politics. Fundamental Problems of Legal Theory and Social Philosophy. 1991 ISBN 0-7923-1143-4

15. J. Wr6blewski: The Judicial Application of Law. Edited by Z. BafIkowski and N. MacCor-mick. 1992 ISBN 0-7923-1569-3

16. T. Wilhelms son: Critical Studies in Private Law. A Treatise on Need-Rational Principles in Modern Law. 1992 ISBN 0-7923-1659-2

17. M.D. Bayles: Hart's Legal Philosophy. An Examination. 1992 ISBN 0-7923-1981-8

18. D.W.P. Ruiter: Institutional Legal Facts. Legal Powers and their Effects. 1993 ISBN 0-7923-2441-2

19. J. Schonsheck: On Criminalization. An Essay in the Philosophy of the Criminal Law. 1994 ISBN 0-7923-2663-6

20. R.P. Malloy and J. Evensky (eds.): Adam Smith and the Philosophy of Law and Economics. 1994 ISBN 0-7923-2796-9

21. Z. Bankowski, I. White and U. Hahn (eds.): Informatics and the Foundations of Legal Reasoning. 1995 ISBN 0-7923-3455-8

22. E. Lagerspetz: The Opposite Mirrors. An Essay on the Conventionalist Theory of Institutions. 1995 ISBN 0-7923-3325-X

23. M. van Hees: Rights and Decisions. Fonnal Models of Law and Liberalism. 1995 ISBN 0-7923-3754-9

24. B. Anderson: "Discovery" in Legal Decision-Making. 1996 ISBN 0-7923-3981-9

Law and Philosophy Library

25. S. Urbina: Reason, Democracy, Society. A Study on the Basis of Legal Thinking. 1996 ISBN 0-7923-4262-3

26. E. Attwooll: The Tapestry of the Law. Scotland, Legal Culture and Legal Theory. 1997 ISBN 0-7923-4310-7

27. J.C. Hage: Reasoning with Rules. An Essay on Legal Reasoning and Its Underlying Logic. 1997 ISBN 0-7923-4325-5

28. R.A. Hillman: The Richness of Contract Law. An Analysis and Critique of Contemporary Theories of Contract Law. 1997 ISBN 0-7923-4336-0; 0-7923-5063-4 (Pb)

29. C. Wellman: An Approach to Rights. Studies in the Philosophy of Law and Morals. 1997 ISBN 0-7923-4467-7

30. B. van Roermund: Law, Na"ative and Reality. An Essay in Intercepting Politics. 1997 ISBN 0-7923-4621-1

31. I. Ward: Kantianism, Postmodernism and Critical Legal Thought. 1997 ISBN 0-7923-4745-5

32. H. Prakken: Logical Tools for Modelling Legal Argument. A Study of Defeasible Reasoning in Law. 1997 ISBN 0-7923-4776-5

33. T. May: Autonomy, Authority and Moral Responsibility. 1998

34. M. Atienza and J .R. Manero: A Theory of Legal Sentences. 1998

35. E.A. Christodoulidis: Law and Reflexive Politics. 1998

ISBN 0-7923-4851-6

ISBN 0-7923-4856-7

ISBN 0-7923-4954-7

36. L.M.M. Royakkers: Extending Deontic Logicfor the Formalisation of Legal Rules. 1998 ISBN 0-7923-4982-2

37. J.J. Moreso: Legal Indeterminacy and Constitutional Interpretation. 1998 ISBN 0-7923-5156-8

KLUWER ACADEMIC PUBLISHERS - DORDRECHT / BOSTON / LONDON