Judicial Lawmaking and Precedent in Supreme Courts

What does it mean for a supreme court to ‘make law?’ When is it possible to say that its decisions are ‘precedents?’ To what extent should a supreme court’s pronouncements be taken into account by others – lower courts and political branches? And how should these other actors reason with such precedents? This article shows how a particular approach to judicial lawmaking and precedent shapes answers to these questions and examines them in relation to the US Supreme Court and the French Cour de cassation. The findings are then used for a critical analysis of the European Court of Justice’s case law. It is suggested that while the US and French systems have found some ways of reconciling judicial lawmaking with the basic premises of their constitutional and political systems (although they are not entirely satisfactory), the EU system is still waiting for an account of the Court’s lawmaking and precedent. The conclusion indicates directions of possible further research relevant for all courts examined.


INTRODUCTION
The status of the European Court of Justice's decisions in EU law is somewhat puzzling. Article 288 TFEU indicates that decisions are binding only on those to whom they are addressed. However, this underplays their significance. When applying the text of the Treaty itself, such as Article 30 TFEU prohibiting quantitative restrictions on imports between the Member States, lawyers always need to consult decisions of the Court in order to know whether the prohibition applies in their case. In the absence of the Court's decisions much of EU law would not exist: foundational doctrines of EU law, such as direct effect and primacy, can be found only there. 1 The puzzle is well illustrated by a major treatise on the EU judicial system which states that the 'case law -though in theory not formally binding -is often the most important source of law'.
On the other hand, it is 'no longer especially controversial to insist that common law judges make law', 2 and a doctrine of precedent is a paradigmatic feature of the US legal system, which belongs to the common law tradition. The Cour de cassation, however, cannot make law -properly so called -and its jurisprudence is a mere authority, although nobody denies its practical importance. And since the Cour does not make law, its decisions can be extremely brief, lacking any 'case law technique' familiar from common law.
In this article I look at judicial lawmaking and precedent in the context of the US Supreme Court's constitutional adjudication and the French Cour de cassation's application of the Civil Code and examine whether the foregoing characterisations of the US and the French legal system are entirely true. I then use these findings in order better to understand the European Court of Justice's case law.
The first section of this article provides a working definition of precedent and briefly discusses the persistent debate on the dichotomy (or convergence) between common and civil law. I stress that depending on the particular features of precedent on which one focuses, one may or may not be able to conclude that these legal traditions are converging as regards their treatment of previous judicial decisions. I then discuss the questions that I see to be important in connection to my inquiry into judicial lawmaking and precedent in EU law. The questions are, firstly, in what sense is it said that these courts make law; second, who is constrained by their pronouncements; and, thirdly, how does the precedential constraint actually work in each of the systems.
The two sections that follow examine these questions in the context of two legal systems: the US and the French. I examine the role of the Supreme Court's precedents in constitutional adjudication and the Cour de cassation's jurisprudence interpreting the Civil Code. While many legal thinkers acknowledge that judges make law in both systems, they also insist that this judicial lawmaking is based on creative interpretation of the foundational documents -the Constitution and the Civil Code. Both courts are often said to declare what these documents mean; they are not acknowledged as having autonomous power to create norms independent of the foundational document. In the US this understanding of the judicial lawmaking activity (in the context of constitutional adjudication) allows other branches of government to claim that they can provide competing interpretations of the Constitution. In France the fiction of 'lawmaking as creative interpretation' allows the legislator to control the Cour by adopting legislative provisions which correct interpretations adopted by the Cour.
I also argue that a particular conception of a case before the two courts is important for understanding the nature of precedential constraint, firstly since it limits the precedent court's lawmaking power, and secondly because it brings elements of 'real life' into lawmaking through adjudication (in contrast to lawmaking by legislatures). The two systems have distinct mechanisms to achieve this.
The concluding section applies these findings to the functioning of the European Court of Justice's case law. I suggest that while in some respects the functioning resembles the two systems, it does not contain (or does not employ in practice) the elements which would allow other actors -lower courts and political branches -to constrain the Court or moderate its lawmaking activity. The Court enjoys expositional supremacy and exclusivity, with regard to both political institutions in the EU and other courts.
Nor does the case play the central role in the reasoning with the Court's precedent. In its decisions the Court says too much when compared with the French Cour de cassation, which must respond strictly to the legal grounds of an appeal in cassation, and too little when compared with the US Supreme Court, which must persuade other courts to follow its precedents in the way it wants them to. However, the European Court of Justice cannot command other courts in the same way as the Cour de cassation, since it does not have an effective sanction to enforce their obedience.
The conclusion indicates directions for further research on judicial lawmaking and precedent, addressing problems identified in this article.

APPROACHING PRECEDENT
I propose a broad definition of precedent: a prior judicial decision which has normative implications beyond the context of the particular case in which it was delivered. I use the notion of precedent as a generic term which, for the purposes of this article, covers both the common law doctrine of stare decisis (or precedent) 3 and the French concept of jurisprudence. 4 That something has normative implications does not imply strict formal binding force which would require the court to choose between recognising a relevant prior judicial decision as binding and following it, or overruling it (or possibly to break the law if the court is not entitled to overrule prior judicial decisions it does not want to follow). 5 This is how 'legally binding' is sometimes understood, and I am interested in the wider effects precedent has. Another reason I use the term 'normative implications' instead of the more common terms used in relation to precedent (for example 'strictly binding,' contrasted with 'persuasive') 6 is the very peculiar nature of precedential constraint, which resists such binary classifications. It is one of this article's aims to explore the nature of precedential constraint in each of the systems. The difficulties of such an enterprise can be illustrated by the controversy surrounding the debate on the convergence between the civil and common law legal traditions as regards their treatment of precedent.
Interpreting Precedents, 7 the product of a group of leading legal theorists thoroughly examining a number of legal systems on the basis of a set of agreed questions, claims 'that precedent counts for a great deal in civilian countries' and even asserts that '[t]he tendency to convergence between systems of two types is a salient fact of the later twentieth century, although there remain real differences, some of great importance'. 8 Adams, on the other hand, notes in his critical review of Interpreting Precedent that 'despite academic arguments to the contrary, Belgian courts mostly cite precedents or case-law, if at all, in an opportunistic manner -[ie], when they confirm the position taken by a court'. 9 Therefore 'to conclude that 3 F. Schauer, Thinking Like a Lawyer. A New Introduction to Legal Reasoning (Cambridge, Mass. and London: Harvard University Press, 2009), 37 reserves the term 'stare decisis' only for horizontal precedent (concerning the same court) and distinguishes it from precedent. The term 'vertical stare decisis' is nevertheless also used, and I do not make this distinction in this article. I use stare decisis and precedent synonymously. 4 I italicise the term 'jurisprudence' to make clear its distinctiveness from the common law understanding of precedent and also, to distinguish it from the English use of the term. For different understandings of the word 'jurisprudence', which has its origin in the Latin word iurisprudentia, see C. Grzegorczyk, 'Jurisprudence: phénomène judiciaire, science ou méthode? ' (1985) 30 Archives de philosophie du droit 35. 5 See A. Peczenik, 'The Binding Force of Precedent' in N. MacCormick and R.S. Summers (eds), Interpreting Precedents. A Comparative Study (Dartmouth: Aldershot, 1997), 478, who notes that 'formal bindingness may be regarded as a non-graded concept, like "pregnant"' and then explains that it is too narrow a view. 6 See eg the classic, R. Cross and J.W. Harris, Precedent in English Law (Oxford: Clarendon Press, 4th ed, 1991), 4. 7 note 5 above. 8 N. MacCormick and R.S. Summers, 'Introduction' in MacCormick and Summers (eds), n 5 above, 2. See also ibid, 'Further General Reflections and Conclusions' in MacCormick and Summers (eds), n 5 above, 546-547. 9  the civil and common law are significantly converging is [according to Adams] surely an exaggeration'. 10 Another comparatist, Mauro Cappelletti, also takes a rather cautious approach: '[s]tare decisis is still an important difference, even though, admittedly, a diminishing one'. 11 Cappelletti highlights three 'still important differences': a) the organization of higher courts (into different hierarchies, which leads to 'more diffuse authority of both the organs themselves and their decisions'), b) their lack of discretion to select cases that they want to hear (which, apart from a huge workload, has an impact on judges' understanding of their task in the legal system -which is not to create precedents but rather to control the application of the law by lower courts in thousands of cases), and finally c) the kind of people who sit on the bench in the highest courts. 12 He illustrates his thesis with the example of judicial review in the civil law tradition, which is centralised in the hands of one court, and argues that the absence of precedent was one of the reasons for its centralisation. 13 The question of the difference between the two traditions' approaches to precedent is far from theoretical: as I mentioned, the European Union brings the two traditions together, and we can expect each to treat the Court's decisions somewhat differently. It can seem that Adams was justified in criticizing Interpreting Precedents as regards the 'convergence thesis'. 14 But I think much depends on what one sees to be important about precedent. In the following section I try to discuss at least some of the features of precedent which seem to be important in relation to the working of the EU judicial system. The US Supreme Court's doctrine of stare decisis in the context of constitutional adjudication and the concept of jurisprudence as employed by the French Cour de cassation or, more precisely, the French civil law scholarship, will assist me in this inquiry.
I want to emphasise at the outset that I am aware of the challenge of any such wide comparative exercise: there is no single understanding of the concept of stare decisis or jurisprudence within the legal systems which I am going to study. So in the course of my inquiry I make conscious choices, based on the purpose of this article: among the various plausible understandings and theories I focus on those which can say something relevant about the practice of the Court of Justice (while of course, I am trying to do justice -or rather to avoid doing injustice -to their alternatives). My attempt is to take what William Ewald called the 'jurisprudential approach to comparative law' 15to dig deeper into the minds of lawyers, and particularly legal thinkers, in the two selected legal systems to see how they understand their practice and its place within their legal systems. As will become apparent particularly in relation to my analysis of the concept of the jurisprudence of the Cour de cassation, much of its widespread understanding (especially in Anglophone writings) is based on a failure to approach it with the French conceptual framework in mind.
What are the features of precedent which I propose to examine?
The first question can be labelled the 'hermeneutics of precedent'. It seeks to understand what the normative basis for precedent's effects beyond a particular case is. It is related to the (once popular) question whether the court 'makes' or 'creates' law or merely 'finds' or 'declares' it. An intuitiveand too fast aresponse to this question would be that of course the US judges (like all common law judges) create law, and their French counterparts vigorously deny this. However, we will see that the response to this question is not so straightforward and is linked to the (ever popular and I am afraid never to be answered) question of what law is. Happily, and in line with the comparative jurisprudential approach, I do not examine this question in the abstract as a legal philosopher, which I am not. I want only to understand how the participants in the two legal systems being examined, the US and the French, tend to respond to it, in order to throw light on my inquiry.
The second question asks who is constrained (in the widest sense of the word) by precedent. I examine it on three different levels: vertical, which refers to the courts inferior to the two examined courts; horizontal, that is the precedent court itself; and, finally, I look at the relationship between the two courts' precedents and other branches of government.
Finally, I ask about the nature of the effects which precedent produces. How does precedent constrain? To make it more concrete: precedential reasoning in common law is distinguished from reasoning on the basis of legislation since the constrained court (the one which is to apply the precedent rule) has comparatively greater freedom in moderating the rule on the basis of the facts which underlie the precedent decision. I examine two more specific and interrelated problems: firstly, what role the factual context of the case -a 'real life' situation -plays in reasoning from precedent, and, secondly, how is the case -as opposed to the abstract context in which the legislator makes lawrelevant for the extraction of the norm implied in the precedent decision.
Necessarily, these questions do not exhaust all problems of judicial lawmaking and precedent in each of the systems. I do not explore here how judicial decisions are reported and what consequences publication practices have for the normative implications of precedent. I also leave unexamined the role of other actors, particularly academics, who in France -in the form of la doctrineplay a more important role in the functioning of jurisprudence than their counterparts in the US. However, the often emphasised differences between the common law and civil law traditions do not seem so sharp when we move from large scale studies of the two traditions to particular legal systems. As with the inquiry into precedent in the two traditions the extent of the difference depends on what questions one has in mind. 16

IT
There is a fundamental difference between common law adjudication and constitutional adjudication. The view that it is 'no longer especially controversial to insist that common law judges make law' 17 concerns just common law and not the interpretation of the Constitution. In constitutional adjudication, all creative power of the Court is justified by its task of interpreting and applying the Constitution. At the dawn of US constitutionalism Chief Justice Marshall expressed the view that '[i]t is emphatically the province and duty of the judicial department to say what the law is. Those who apply the rule to particular cases, must of necessity expound and interpret that rule'. 18 Many (surely not all) theories of constitutional interpretation are based on this premise: the Court does not make the Constitution, but merely gives meaning to it, no matter how creative this process 16 P. Jestaz and C. Jamin, 'The Entity of French Doctrine: Some Thoughts on the Community of French Legal Writers' (1998) 18 Legal Studies 415, 427, for example, contend that 'the countries using common law do not feel the need for rational constructions and only show mistrust of the creators of systems like [the French]. Thus, their best jurists stopped writing treatises at least thirty years ago'. The same authors present the US academic scholarship as an 'anti-model' to the French doctrine in La doctrine. (Paris: Dalloz, 2004), 265-306. But one may ask, for example, whether the Restatements are not 'rational constructions' of common law doctrines, or what one is to make of this statement by James Gordley: 'Before the 19th century, the common law was not organized by doctrines or even by areas of law such as contract and tort. It was organized according to forms of action, each with its own rules. In the 19th and 20th centuries, treatise writers rationalized and systematized these rules. '  is. 19 Rubenfeld's defence of the Court's 'judicial revolutions', which he qualifies as 'radical reinterpretations' of the Constitution, illustrates this. 20 The controversy concerns the question of what constitutes an interpretation mandated by the Constitution and how far this interpretation is, or ought to be, limited by the text of the Constitution or other norms and doctrines. 21 It is clear that the Court's interpretation determines the real obligations which flow from the Constitution and that in some sense the Supreme Court 'makes' constitutional law (sometimes the expression 'constitutional common law' is used). But it remains an interpretation, distinct from the Constitution itself. 22 WHO IS CONSTRAINED?
Within the judicial branch, we must distinguish between vertical precedent, concerning courts on lower levels of the judicial hierarchy, and horizontal precedent, concerning courts at the same level of the judicial hierarchy.
There are two separate judicial hierarchies below the Supreme Court: federal courts, 23 and state courts. Both the federal courts and the state courts are subject to the appellate jurisdiction of the Supreme Court. In practice, however, this control is very limited, if only because of the limited number of cases the Court hears. 24 At present the Supreme Court grants review in little more than one per cent of all petitions; of these granted petitions, even fewer come from state courts. 25 Thus the federal and state courts, particularly the latter, retain a high 19 It is possible to find a few authors who do not deny that the Court has full-blown constitution-making power outside (or alongside) the written constitution. See eg R.H. Fallon, Implementing the Constitution (Cambridge, Mass. and London: Harvard University Press 2001), 111-126. But Fallon also defends this by saying: 'I know of no case in which the unwritten constitution calls for results that cannot at least be reconciled with the language of the written Constitution, even if reconciliation sometimes depends on tenuous or even specialized interpretations'. ibid, 111-112. 20 21 It is difficult to provide a 'neutral' reference to some general treatment of constitutional interpretation in the US, since it is perhaps the most contested topic in constitutional scholarship. But see R. Post, 'Theories of Constitutional Interpretation' (1990) 30 Representations 13. 22 As I said before, this is by no means an uncontested position. But note how eg D.A. Strauss, 'Common Law Constitutional Interpretation' (1996) 63 University of Chicago Law Review 877, 878, introduces his defence of a common law approach to constitutional interpretation: 'the terms of debate in American constitutional law continue to be set by the view that principles of constitutional law must ultimately be traced to the text of the Constitution, and by the allied view that when the text is unclear the original understandings must control. An air of illegitimacy surrounds any alleged departure from the text or the original understandings'. 23 For the purposes of this article, I do not distinguish between the so-called 'Article III federal courts' (created by the Congress directly on the basis of Article III, § 1 of the US Constitution), and 'Article I (or also legislative) tribunals' (which have their basis in legislation adopted by the Congress in accordance with Article I and not the Constitution itself degree of autonomy, similar to that of the courts of EU Member States, which escape direct control by the Court of Justice. 26 Despite the Supreme Court's limited control, 'the doctrine of hierarchical precedent appears deeply ingrained in judicial discourseso much so that it constitutes a virtually undiscussed axiom of adjudication'. 27 The role direct appellate review plays in the doctrine of precedent is well illustrated by the attitude of state courts to decisions of lower federal courts. Since there is no direct control by federal courts, 28 and appeal is possible only from state supreme courts to the Supreme Court, many state courts maintain that they are not bound by prior decisions of lower federal courts. 29 Therefore only the Supreme Court is considered to have unquestionable authority by virtue of its being the ultimate authority as regards federal law and the Constitution. 30 Horizontal precedent at the Supreme Court 31 became particularly controversial when the Court refused to overrule Roe v. Wade, 32 relying on the special status of precedent in the US Constitution. 33 The Supreme Court expressly considered the criteria for overruling, which are 'customarily informed by a series of prudential and pragmatic considerations designed to test the consistency of overruling a prior decision with the ideal of the rule of law, and to gauge the respective costs of reaffirming and overruling a prior case'. 34 The criteria were: […] whether the rule has proven to be intolerable simply in defying practical workability; whether the rule is subject to a kind of reliance that would lend a special hardship to the consequences of overruling and add inequity to the cost of repudiation; whether related principles of law have so far developed as to have left the old rule no more than a remnant of abandoned doctrine; or whether facts have so changed, or come to be seen so differently, as to have robbed the old rule of significant application or justification. 35 26  Casey provoked a lively debate. Most reactions focused on the question of whether horizontal stare decisis was required by the US Constitution. 36 More radical critiques of Casey considered horizontal stare decisis to be against the Constitution. 37 Common to both critiques was a concept of the Court's role as interpreting the Constitution rather than creating it. Two separate authorities are embedded in precedent: the Constitution and the Court's interpretation of it. While the Supreme Court insists that 'the federal judiciary is supreme in the exposition of the law of the Constitution' 38popularly taken to mean 'the Constitution is what the judges say it is' 39much contemporary US scholarship (and occasionally the other branches of government too) denies this. 40 For example, when the President vetoes an act of Congress because he believes the act to be unconstitutional and declares such view, his understanding of the Constitution is unrestrained by the Court's precedent. 41 There is no remedy, and the presidential interpretation remains as a 'true' meaning of the Constitution. 42 Other branches of government may deny that they are bound by the Court's interpretations of the Constitution, since they are just that -interpretationsnot to be confused with the Constitution itself. This understanding of constitutional lawmaking by the Supreme Court as a mere interpretation of the Constitution allows other actors to emancipate themselves from the Court's determinations of what the Constitution is and to develop their own interpretations. As Devins and Fisher put it: […] the authority to say what the law is does not make the Court supreme, other than in that particular case. It is also the province and duty of Congress, in concert with the President, to say what the law is. The Court merely states 36  what the law is on the day the decision comes down. If Congress and the President disagree with that interpretation, the law may change after that. 43 In other words, according to this reading of the US Constitution other branches of government are not bound by the Supreme Court's precedents.

THE NATURE OF THE SUPREME COURT PRECEDENT'S CONSTRAINT: CENTRALITY OF THE CASE
The importance of the case for precedent Context of the case distinguishes reasoning from precedent from that based on rules enacted in an abstract context of the legislative procedure. 44 As Neil Duxbury puts it: Since the recorded case [ie the precedent case] is not a strict verbal formulation of a principle, only exceptionally will judges conceive their task to be one of interpreting specific words or phrases within a case rather as they might focus on the precise wording of a statute. Instead, they will consider if the case is factually similar to or different from the case to be decided. Caselaw, we might say, unlike statute law, tends to be analogized rather than interpreted. 45 In the classical common law adjudication -which lies in the background of most theoretical studies of precedent -cases are analogised through their facts. It is on the basis of the facts of the precedent case that subsequent courts identify binding elements of the precedent decision -its ratio ('holdings' in the American parlance), or distinguish the precedent case from the one before them in order to explain that they are not bound by the precedent rule. 46 This characteristic of precedent led Raz to observe that 'judge-made law' (as he calls it) is revisable, constantly open to the possibility of being distinguished. 47 As we will see below, the constitutional interpretation of the Supreme Court differs from this traditional (common law) understanding of precedent, particularly because facts play a less significant role. But the case -defined otherwise than through its underlying facts remains important for precedential reasoning.
Two principal reasons are offered to explain why the context of the case matters in determining the extent to which a prior later one. One stresses the importance of a 'real-life situation' for creating an optimal rule implicated in the precedent decision; the other concerns limitations on the lawmaking power of judges.
The real-life context of judicial lawmaking Many people believe that the case gives a judge a special insight not available to the legislator. 'Treating the resolution of concrete disputes as the preferred context in which to make law […] is the hallmark of the common law approach,' says Schauer. 48 This view is prevalent in much of the Supreme Court's case law and is not limited to questions of precedent. The Court for example stated that 'a concrete factual context [is] conducive to a realistic appreciation of the consequences of judicial action'. 49 The emphasis on some 'real-life' disputes, in contrast to the abstract context in which legislators take their decisions, is perhaps best illustrated by Justice Holmes' famous statement that '[t]he life of the law has not been logic: it has been experience'. 50 Lawmaking in the context of deciding concrete disputes keeps the development of law in contact with reality. 51 However, Schauer has recently questioned this fundamental premise of common law adjudication by reference to another of Holmes' dicta: '[g]reat cases like hard cases make bad law', 52 and pointed to the many distortive effects of a concrete situation in front of a court which faces it. 53 The answer to Schauer's concerns can be the adaptability of the common law, which can be greater than he himself admits, 54 and also the fact that judicial lawmaking is not exclusive, but operates in mutual communication with the legislator. 55 The constraining function of the case Another reason why a case should be relevant for determining holdings is a desire to constrain judicial lawmaking activity. If creating law is seen as only a corollary to the courts' main task of deciding disputes, then law creation will occur only in this context. One can therefore insist that only what has been said in this context (deciding the concrete dispute) matters. However, this view relies on an assumption that it is the primary role of the Supreme Court to decide disputes. Is the assumption correct? 48  To a great extent, it is still Marbury v Madison, 56 which continues to shape the vision of what the Supreme Court should be doing, apart from the textual limitations of Article III of the US Constitution. 57 There are two conflicting views: the 'dispute resolution' (or 'private right') model on the one hand and the 'public rights model' on the other. 58 The first, dispute-resolution model reflects the traditional conception of adjudication as the resolution of disputes between two parties and the enforcement of their rights. It is characterised by Marshall in Marbury: 'the province of the Court is, solely, to decide on the rights of individuals'. 59 Rulemaking is a mere coincidence in the course of the resolution of a dispute. 60 This does not necessarily mean that the case in this model must be defined by factual circumstances. Especially in constitutional adjudication, but also when interpreting statutes, concrete facts do not matter. 61 The case can be defined by a legal issue presented to the Court for its decision. 62 Consider the Court's rule: '[o]nly the questions set forth in the petition [for certiorari], or fairly included therein, will be considered by the Court'. 63 It is true that the Court can be quite flexible in what it considers to be 'fairly included' in the questions set forth in the petition, but, as Hartnett notes, 'the leading treatise urges counsel not to phrase their questions presented in terms of error correction'. 64 So, for example, in Roper v Simmons one of the questions presented for review was: 'Is the imposition of the death penalty on a person who commits a murder at age seventeen "cruel and unusual", and thus barred by the Eighth and Fourteenth Amendments?' 65 Although the factual circumstances of the case are quite shocking -Simmons was a minor who, at the age of 17, together with one accomplice, broke into the home of their victim (an older woman), tied her up, and threw her off a bridge so that she drowned (all this having been planned beforehand), the question before the Court was fairly abstract and general, and the Court addressed it in that manner. 66 56 n 18 above. 57 The 'case or controversy requirement'. 58 See Fallon, et al, n 28 above, 67-73 with further references. 59 Marbury v Madison, n 18 above, 170. 60 See also a quotation from Marshall's opinion, n 18 above. 61 66 Justice Kennedy writing for the Court started his analysis in the following way: 'The prohibition against "cruel and unusual punishments", like other expansive language in the Constitution, must be interpreted according to its text, by considering history, tradition, and precedent, and with due regard for its purpose and function in the constitutional design.' 543 U.S. 551, 560. But it might still be possible to distinguish elements of the decision which are more relevant for the Court's conclusion than others, albeit only imprecisely. 67 In contrast, the second, public-rights model implies that what the court is doing has significance beyond the particular dispute before it. The Supreme Court is not 'a mere settler of disputes, but rather […] an institution with a distinctive capacity to declare and explicate public values'. 68 This role of the Court is reinforced by its power to select the cases it wants to hear. Clearly, the interests of the parties or the need to resolve their particular dispute are only secondary to the Court's main aim: to set a uniform rule which will be binding on other courts and possibly also on other actors. From that perspective, whatever the Court says matters. Especially in the hierarchical context, as Schauer observes, 'the advocate in a lower court urging a result plainly inconsistent with the language in the higher court opinion has a steep uphill climb, and arguments that the obstructing language is mere dicta, or not part of the ratio decidendi, are usually unavailing'. 69 But even if such model is adopted, it does not necessarily mean that the case becomes irrelevant for determining the scope of the holding. What differs is the definition of the casethe case is not the dispute between the parties, but rather an abstract issue presented to the Court.
The relevance of the case in which the precedent court adopted its decision establishing the norm 'implicated' in it gives subsequent courts a flexibility which legislative rules do not allow. 70 I mentioned that the Supreme Court's supervision of lower courts is relatively limited. 71 The Court cannot correct every misinterpretation of law, including in its own precedents. Scholars have noted that in order to know 'what the law is' one must look at decisions made by federal Courts of Appeal, since they are authoritative pronouncements of federal law in the absence of a clear precedent of the Supreme Court. 72 Not only do lower courts effectively fill in the gaps left by the Supreme Court's precedents, but they can also substantively reformulate their original scope and function as authentic expositions of federal law. Sometimes they can effectively resist them, without rebelling openly. It can be illustrated by the title of a recently published article: 'What if the Supreme Court held a constitutional revolution and nobody [meaning the lower and state courts] came?' 73 67 Schauer, n 3 above, 181-182 (and also 54-56) is somewhat sceptical about the ability to identify reasons which were 'necessary' for the court to adopt a particular outcome, but himself asserts that 'this idea should not be taken too far', providing examples of statement made by judges which can be clearly characterised as dicta. 68 Fallon, et al, n 28 above, 68. 69 Schauer, n 2 above, 57-58. 70 It is not suggest that legislated rules are more determinate or that they allow less flexibility. But, contrary to precedent, their interpretation is not inextricably intertwined with a particular case. 71 See the text to n 24 above. 72  Emancipating precedent from the confines of the case However, the importance of the case seems to diminish in certain contexts. James Pfander has recently observed that the US federal courts increasingly tend to say what the law is regardless of the context of a dispute and calls this 'the principle of expositional supremacy'. 74 For example, when federal courts decide on the liability of state officials, they must identify the applicable constitutional standard of officials' conduct before they address the question of official immunity, which would prevent any suit from succeeding regardless of the applicable constitutional standard of official behaviour. Courts are therefore required to lay down a constitutional standard (if the standard was not clear before) and only then to decide whether it is relevant for deciding the case at hand. The Court justifies such an approach in the following way: 'Deciding the constitutional question before addressing the qualified immunity question also promotes clarity in the legal standards for official conduct, to the benefit of both the officers and the general public.' 75 A different phenomenon which has the same effectgiving the Court more control over the subsequent application of its precedentconcerns the diminishing importance of the holding/dicta distinction. Now I do not mean its flexibility (or malleability), which weakens the control exercised by the precedent court over subsequent decision makers. I mean the assumption that whatever the precedent court says matters. Some federal courts of appeal require lower courts to follow not only holdings, but also the dicta of their precedentsa trend which has recently provoked thorough criticism by a circuit court of appeal judge. 76 As I noted above, this approach is based on the public rights model.
Finally, the way in which the Court drafts its decisions also undermines the case-bound conception of precedent. As Frederick Schauer observed in 1995, 'It is a routine charge against contemporary judicial opinions that they read more like statutes than like opinions of a court.' 77 Defending this approach, Schauer highlights the function of a judicial opinion to 'guid[e] lower courts and legally advised actors' 78 and submits that 'it may be appropriate to think of opinion writing as (at least in part) a conscious process of rule-making'. 79  and no less likely to be constrained and informed by the kinds of considerations we would employ with respect to any other rule-making enterprise.' 80 But here we reach the difficult question of the democratic legitimacy of judicial lawmaking. Legal certainty and predictability provided by opinions drafted in this way are not the only considerations that matter. Michael Dorf argues: Schauer is surely correct that, other things being equal, whenever notice is an important value, statute-like judicial opinions will be preferable to vaguer, more principle-like judicial opinions. But other things are rarely equal. There are legitimacy costs that arise out of a court's attempt to emulate a legislature, because legislatures and courts derive their legitimacy from different sources. 81 Dorf's concern is precisely what I discussed above: the possibility of later courts modifying the rule implicated in the precedent decision. Dorf explains that 'more statute-like opinion entails more particular commitments' 82 and contends that Schauer does not 'explain why we would want our courts to commit themselves in the future to the concrete applications the present judges happen to favor in cases not currently before them'. 83 In sum, precedent unbound from the particular case in which the judicial decision forming its basis was made gives more control to the Supreme Courtthe court which pronounces rules which are relatively independent of the cases before it. The traditional common law concept of precedent is significantly redrawn.  168-169. and is duly reflected in all standard comparative accounts. 85 According to Lasser, the republican ideology insists that '[j]udges in some important sense cannot usurp [the] legislative law making power because law is defined categorically as legislative in origin'. 86 This does not mean that the French are 'blind to the fact that judges play a significant role in the elaboration, development, and modification of normative rules'. 87 This awareness is 'hardly recent', and Lasser recalls Portalis, the primary author of the French Civil Code, who highlighted the inability 'to predict and settle everything' and referred to 'the judge and the jurisconsults, penetrated by the general spirit of the [codified] laws, to direct their application'. 88 Still, according to Lasser the French conception of law is 'above all a legal rule (or a set of legal rules) that has been formally adopted by the legislature in the form of "loi" (legislation)'. 89 Lasser considers the American approach to comparative law parochial, arguing that it is based on the 'American realist definition of law, [which] when applied uncritically to a civilianand especially to the Frenchlegal system, leaves little or no possibility of encountering that system.' 90 In a series of works culminating in Judicial Deliberations, 91 Lasser stresses that the French legal system employs a concept of law and legality which is fundamentally different from the American realist account of law. While his account is insightful, I think the presentation of the French doctrine of sources of law is incorrect. 92 Malaurie and Morvan write that 'the majority of authors recognize jurisprudence as a source of law'. 93 Whilst this statement may be an exaggeration (since considerable debate about jurisprudence remains), 94  sources of law in the following way: 'For centuries, the designation "sources of law" serves to mark la loi, la jurisprudence, la coutume, etc.' 95 This speaks for itself.

THE COUR DE
Much depends on how we define 'the law'. 96 The real debate in France turns on this, together with the place of jurisprudence and the judiciary in general within the French legal system. The debate considers the legitimacy of judicially created norms and their normative effects beyond particular cases; in this respect the French debate is not so very different from the debates taking place in other legal systems. It is true that '[o]ne need hardly call judicial decision-making "law" in order to stress that judges make normative choices and thus exercise highly significant normative authority'. 97 But that is exactly how it is termed in Frenchla création du droit par le juge. 98 This can safely be translated as 'judicial lawmaking'.
Judicial creative activity is based on different premises from those put forward by Lasser. These premises can be found in Portalis's Discours préliminaire: There is a science for the legislator, as there is one for the judges; and the one does not resemble the other. The science of the legislator consist in finding, in each matter, the principles most favourable to the common good; the science of the judge is to put these principles in action, to develop them, to extend them, by a wise and reasoned application, to private relations; to study the spirit of the law when the letter kills, and not to expose himself to the risk of being alternatively slave and rebel, or to disobey because of a servile spirit. 99 Judicial lawmaking (création du droit) is put forward as a creative interpretation of the Code; an interpretation which may go well beyond the Code's wording, while retaining its spirit. This creativity exceeds mere clarification; occasionally filling the gaps; or reconciling possible antinomies. Judicial creation extends to adapting law to meet societal developments. 100 95 Jestaz, n 93 above, 1. 96 See Malaurie and Morvan, n 93 above, 264-266 (these authors note that the debate is often a 'dialogue of the deaf', since each participant has a different conception of law in mind); Ghestin, et al, n 93 above, 451; Jestaz, ibid, 1-8. 97 ibid. 98 Which is a title of (2007) 50 Archives de philosophie du droit. 99  It also true, however, that Portalis added: 'We leave to [jurisprudence] the rare and extraordinary cases that do not enter into the plan of a rational legislation, the very variable and very disputed details that should not occupy the legislator at all, and all the things that it would be futile to try and foresee or that a premature foresight could not provide for without danger.' in Terre (ed), n 88 above, xxix (translation von Mehren At the same time, judicial creativity is presented as circumscribed by the Code or, more precisely, its spirit. Judges cannot 'rebel' against the Code and become completely free. Thus Tunc contends that experience contradicts the statement of Justice Holmes in Lochner that 'general propositions do not decide concrete cases', 101 'a phrase which [according to Tunc] seems to be self-evident, but which is in no way justified by reality and which unfortunately inspires thousands of common law lawyers to distrust of regards codification which lacks any grounds'. 102 While Tunc's view seems to be a naïve endorsement of the Code's power to constrain the judges (but only at first sight), a similar foundational belief (or arguments against it) surrounds the debate about interpretation of the US Constitution and freedom it gives to [the] judges. 103 Holmes' position is a neat slogan, but by no means universally accepted in the US context. 104 The force with which the (spirit of the) Code constrains judges is relative. This is more so once general principles of law are accepted, since these principles enable judicial creativity to 'acquire[ a very particular autonomy'. 105 The Cour de Cassation does not pretend to ground the general principles in a specific provision of the Code but relies on them as the sole basis for its decisions. 106 However, the central idea remains: it is on the basis of the (spirit of the) Code that the judges formulate the principles.
The prohibition on denying justice on grounds of silence, obscurity, or insufficiency of the legislation, 107 at the time of the Code's inception the expression of its creators' belief in its completeness, 108 has become a source of the normative power of courts and is now read as a duty to say what the law is. 109 The notion of a judge who would be 'no more than the mouth that pronounces the words of the law' 110 was, according to Zénati, abandoned much earlier than is generally believed; référé legislatif, a symbol of distrust of the judiciary, was dysfunctional long before it was formally abolished in 1828. 111 As Zénati argues, it is a mistake to consider the Cour de cassation as a judicial body. Instead it is a body outside the judiciary, the mission of which is not, in Robespierre's words, 'to judge the citizens, but to protect enacted laws'. In other words, the Tribunal (later turned into the Cour) was created to protect the sovereign will of the people, embodied in the law, against encroachment upon it by the ordinary courts. According to Zénati, the Cour de cassation has two principal tasks: to protect the unity of the legislation and to serve as a guardian of the principle of the separation of powers. The latter task can seem contradictory to its nature -how can a court be a protector against other courts? However, this is contradictory only if we think of the Cour as a judicial body and not as a quasi-legislator. We will see further what consequences such an understanding of the Cour has for its jurisprudence.

WHO IS CONSTRAINED?
Does it follow that if the Cour de Cassation makes law then the product of this 'making' (ie judicial decision) is automatically binding on others? No. We must distinguish between making law in a particular case, which refers to a situation in which the judge's decision is not constrained by the existing sources (however defined), and making the judge's creation binding on others. Much of the discussion about the lawmaking power of civilian judges focuses on the first question and somewhat confuses this with the second question. For example Dawson calls a 'subsidiary issue' the question of 'the extent to which courts should be bound by high court decisions', 112 when considering the German courts', particularly the Reichsgericht's, practice of disregarding statutory provisions on the basis of general clauses contained in the German Civil Code, particularly that on good faith. 113 Dawson notes that the debate on whether Germany should have a Montesquieu and la bouche de la loi revisited' (2008) 4 European Constitutional Law Review 274 shows, however, this phrase was taken out of the context of The Spirit of the Laws, and Montesquieu did not mean it in the sense generally ascribed to it today. 111 Référé legislatif was a procedure introduced in 1790, which imposed a duty on the (then) Tribunal de cassation to refer a question of interpretation of law to the legislator if the Tribunal was disagreed with three consecutive times by the lower court in the same case. See Dawson, n 85 above, 378-379. A provision making the decision of the Cour de cassation on a second appeal in cassation binding on the lower court was introduced only in 1837. This, together with abolition of the référé legislatif (in 1828), was of paramount importance for establishing the Cour de cassation's authority. On the significance of these two changes see F. Zénati, 'La nature de la Cour de cassation' (Bulletin d'information de la Cour de cassation no. 575, 15 April 2003), at <http://www.courdecassation.fr> (The printed version was not available to me, so I could not provide more precise references); Zénati, n 100 above, 71; Ghestin, et al, n 93 above, 416-418). According to (current) Article L.431-6 of the Code on the organisation of the judiciary, the second cassation based on the same legal grounds (moyens) must be heard by the Full Court (l'assemblée plénière). 112 Dawson, n 85 above, 487. 113 See ibid, 461-479. system of following previous decisions 'ha [d] a weird, other worldly quality, [since according to him] Germany already had a precedent system working order'. 114 That remark is, I suspect, based on the failure to distinguish the two issues.
In France the distinction between judicial lawmaking and the normative force of law made in this way is much clearer due to the express prohibition on arrêts de règlement contained in Article 5 of the Civil Code. 115 Arrêts de règlement, decisions made by the pre-revolutionary courts (Parlements) in particular disputes, but announcing an abstract and general rule which was binding on all courts within the jurisdiction of the Parlement which issued them, were the symbols of the excessive judicial power of courts, and thus expressly prohibited. 116 Therefore to admit that judges make law in the first sense (ie for a particular case before them) does not explain the extent to which other courts or participants in the legal system are bound by such rules. And Article 5 prevents any open recognition of the general binding force of prior decisions. This is further reinforced by Article 1351 of the Code, which establishes only the relative force of res judicata, limiting it to the particular case decided by the court. 117 These two provisions (together with Article 4) 118 form the framework within which the French debate on the normative effects of jurisprudence takes place and which shapes many arguments which can seem formalistic and blind to crude reality. In fact, they are elaborated responses to the express limitations embodied in the Code by the Enlightenment legal thinkers.
Cassation, whereby the Cour de cassation quashes a lower court's decision and sends it back to another court for the final decision (while the Cour's review is limited to points of lawles moyens), 119 and the Cour's wide supervisory power over the lower courts are the crucial elements giving the Cour's jurisprudence its force.
Cassation is conceived as a sanction imposed on lower judges (juges du fond) for disobeying the legislator. 120 The Cour de cassation is a 'secular arm of the legislated law' (le bras séculier de la loi). It therefore does not need to justify its interpretation of the law; to do so would only weaken it -'imperatoria brevitas of the 114 ibid, 484. 115 'Judges are forbidden to decide cases submitted to them by way of general and regulatory provisions.' On arrêts de règlement in general see Dawson, n 85 above, 305-314. 116 In the revolutionary period it was truly believed that judges should exercise no normative power, if only by interpreting laws enacted by the legislator; thus Robespierre's famous desire to erase the word jurisprudence from the French language. See Zénati, n 100 above, 45-55. 117 'The force of res judicata takes place only with respect to what was the subject matter of a judgment. It is necessary that the thing claimed be the same; that the claim be based on the same grounds; that the claim be between the same parties and brought by them and against them in the same capacity.' 118 Law 201,[204][205][206][207][208], who also discusses changes in the scope of the Cour's review and its ability finally to dispose of the case itself. 120 As Zénati, n 100 above, 43 says, cassation was introduced to the French procedure by the King in the 18th century for the same purpose: to control judicial obedience to the royal authority. supreme judgments borrows the concise and closed style of enacted laws'. 121 Such a conceptualisation of the Cour in fact corresponds to Austin's conception of a judge as the sovereign's delegate: The portion of the sovereign power which lies at [the judge's] disposition is merely delegated. The rules which he makes derive their legal force from authority given by the state: an authority which the state may confer expressly, but which it commonly imparts in the way of acquiescence. 122 For Zénati, the delegation of power to the Cour is crucial for determining the legal effects of its jurisprudence. Jurisprudence borrows the legal effects from the interpreted rules, but at the same time is seen as a separate source of law, supported by a sanction -cassation. In this respect Zénati distinguishes his conceptualisation from the 'incorporation theory', which suggested that the Cour's interpretations of a legislative norm are incorporated into it and must have the same effects. On this basis Zénati distinguishes the jurisprudence of the Cour from that of lower courts. Because of the sanction in the form of cassation only the former is a true source of law. 123 The authority of the Cour de cassation's jurisprudence is imposed through its wide supervisory power. This power was supported by the ideology of 'intermediate law', which 'recognised all citizens' power to impose on a judge [understood as a lower judge, "juge du fond"] respect for the law.' 124 An attentive reader will perhaps note how similar this sounds to the Court of Justice's doctrine of direct effect and its use for the purposes of private enforcement of EU law by ordinary citizens. 125 The Cour has always dealt with a much greater number of cases than courts in countries where the supreme courts' authority has rested on different grounds. 126 At the same time, as regards the lower courts its power to impose its interpretation is absolute, since after the second cassation the lower court is bound by the Cour de cassation's interpretation of the law. 127 Zénati contends that the only freedom which the lower courts have is to do mischief and 121 Zénati, n 111 above. . 126 This applies not only to supreme courts in common law jurisdictions but also to supreme courts based on the German model. For a comparison of different review models see Geeroms, n 119 above (providing a rich historical account of different models of review, and noting at 215 that the German revision model 'as a reaction against the cassation ideal, [...] explicitly intended not to supervise the lower court. Instead, its primary purpose was, and still is, the assurance of uniformity in case law and the harmonious development of existing law without disregarding the interests of the parties'). 127 See n 111 above. be exposed to sanction in the form of cassation. 128 We will see below, when I discuss the nature of jurisprudence's force, that this is not completely true and that it is important that lower courts do sometimes disobey.
The concept of lawmaking as creative interpretation finds its limits when the Cour wants to change the existing jurisprudence. Overturning (revirement de la jurisprudence) conflicts with the principle of legal certainty, the French doctrine therefore debated the possibility of limiting the effects of overturning only for the future. After a group of academics and practitioners submitted a report to the First President of the Cour, 129 the Cour actually declared that it had such competence. 130 Morvan considers this to mark the crossing of the Rubicon, 131 since it is taken as proof of the lawmaking power of the Cour 132 -now openly admitted by the Cour itself. One is then faced with the prohibition of arrêts de règlement imposed on the French civil courts by Article 5 of the Civil Code. But here remains another important element of the normative effects of jurisprudence: its relationship with the legislator.
Portalis predicted the creative role of judges and their jurisprudence. 133 He writes: 'It is necessary that the legislator keep an eye on [jurisprudence]. He can learn from it and he can, for his part, correct it.' 134 In the words of the today's commentator, this is 'essential', since 'while it was not possible to keep the system which reserved interpretation of the law to the legislator by way of référé legislatif, the democratic principle commands recognizing the legislator's power, and even a duty, to erase jurisprudence which it considers erroneous, shocking or inappropriate'. 135 This is also why the Cour produces its annual reports, although their role has changed since, while '[i]nitially conceived as an instrument of the subordination of the [Cour de cassation] to the legislative power, they became an instrument of diffusion of the jurisprudential innovations and the normative policy of the supreme court'. 136 The idea of 'lawmaking as creative interpretation' permits the borrowing of the effects of the interpreted rules, but at the same time contains a constraining element: the legislator can always undo jurisprudence it dislikes. 137 That is why Ghestin, who classifies jurisprudence as a source of law, at the same time opposes abolishing the prohibition of the arrêts de règlement. According to Ghestin, jurisprudence would acquire the same status as enacted law, which could lead to irreconcilable conflicts between the two. 138 However, according to Zénati such conflict cannot arise, precisely because of the unity between enacted laws and the jurisprudence which interprets them. If the text of an enacted law contradicts previous jurisprudence, it must be seen as amending it, and vice versa. If there is a contradiction between jurisprudence interpreting a hierarchically subordinate norm on the one hand, and the text of an enacted norm superior to it on the other, it must be conceived as if the Cour was also implicitly interpreting the superior norm. 139 This of course holds only theoretically, and it can find its limits in the very clear wording of the text of the enacted norm; in practice some scholars talk about contra legem jurisprudence. 140 Be that as it may, an increasing number of academics question whether the prohibition of the arrêts de règlement has any real meaning. 141

THE NATURE OF JURISPRUDENCE'S CONSTRAINT: LEGISLATIVE PRECEDENT
If we think of the Cour de cassation as the secular arm of legislated law, we cannot expect it to provide extensive reasoning: imperatoria brevitas says Zénati. The brevity of the decision is something which constantly perplexes observers from other legal systems, who also stress the scanty treatment of facts in the Cour's decisions. 142 Dawson comments on the brevity of the Cour de cassation's judgments in the following way: To readers trained in our own tradition [the US], the extreme parsimony of its statements of facts is even more striking than the brevity of its propositions of law. It is not only striking but in a way more important, for it raises issues that are central not only to workable case-law technique but to conceptions of the kind of law that judges are qualified to make. 143 137 This construction however encounters two fundamental problems: one concerning general principles of law, which, once formulated by the Cour, can be 'corrected' by the legislator only to a limited extent (see particularly P. Morvan, Le principe de droit privé (Paris: Panthéon-Assas, 1999), 735-749), and another related to the control of the compatibility of the legislation with international treaties (including EU law) binding on France (see eg Ghestin, et al, n 93 above, 248-258). 138 Ghestin, et al, ibid, 446-448. 139 Zénati, n 100 above, 224. By 'workable case-law technique [and] conceptions of the kind of law that judges are qualified to make' Dawson means what I described above as the centrality of the case and its facts for determining what is binding in the previous court's decision, primarily by separating findings from dicta or possibly distinguishing the case before the judges from that which led to a precedent decision. Extensive reasoning of the precedent judgment then serves as a means of limiting the rulemaking power of the precedent court: it is the subsequent court which defines the scope of holdings and can possibly distinguish the subsequent case on the basis of the facts before it, or at least on the basis of what the court said.
The whole concept of cassation speaks against the limiting of the Cour de cassation in this way. Unlike the US Supreme Court, the Cour is not part of the judicial system; it is a secular arm of the legislated law, which sanctions ordinary courts if they disrespect it. As Zénati notes, it 'represses jurisprudence in the name of the legislated law' 144 by not allowing a lower court to participate in the formulation of the 'precedent' rule and also by the way it reasons. Zénati therefore calls the Cour de cassation's production 'legislative jurisprudence' and distinguishes it squarely from 'genuine' jurisprudence produced by other supreme courts which have full appellate jurisdiction. While the latter serve as model for lower courts, the Cour de cassation issues legislation-like rules and controls obedience to them. 145 Does this mean that the Cour de cassation is virtually unconstrained as a result of being liberated from the duty to reason in the way common law judges do?
Here we can come to a paradoxical response to Dawson's concerns: it is brevity itself which constrains the lawmaking activity of the Cour. In its decision the Cour is expected to deal only with specific legal 146 grounds submitted in the appeal in cassation. 147 At the same time, everything the Cour says in its judgment has immediate legal consequences for the process before the lower court; 148 if there is a second appeal in cassation pursued on the basis of the same moyens, it will be the General Assembly of the Cour that will hear the case. 149 And its decision is then binding (on the points of law) on the lower court -no other appeal is possible. The ability to raise an issue on its own motion is very limited, and the Cour uses it only rarely. 150 The Cour de cassation is therefore far more constrained in what it can say in its decision than, for example, the US Supreme Court; it can thus produce fewer statements which can be taken as its authoritative pronouncements 144 Zénati, n 111 above. 145 Zénati, n 100 above, 40-44, 177-180 and more explicitly Zénati, n 111 above. 146 See Zenati-Castaing, n 161 above, 1557. 147 'Appeal in cassation' corresponds to 'pourvoi en cassation', while 'legal grounds' correspond to 'moyens'; these are translations used by the European Court of Justice.
148 And yet another translation -'lower court' refers to 'juge du fond', ie the court the decision of which was appealed in cassation to the Cour de cassation. 149 See n 127 above. 150 Ghestin, et al, n 93 above, 474-475. of law. Some authors even mention that dicta, which the Cour nevertheless occasionally utters, breach the prohibition on arrêts de règlement. 151 This understanding of the reasoning also explains why the Cour, after making public materials related to its decision, including the opinion of its avocat général and the report of the conseiller rapporteur, does not make them part of the official decision but puts them to one side, 152 as travaux préparatoires. Their title of travaux préparatoires again suggests that Cour decisions are legislative works rather than a process of judicial deliberation, reminding us that the Cour is an adjunct to the legislator. 153 Most comparatists note that the Cour de cassation's decisions pay only scant attention to the facts. 154 We can understand why this is so, once we view cassation in the sense discussed by Zénati -as a sanction enforcing lower judges' obedience to the legislated law and not as a mechanism for correcting individual injustices. For some scholars this is but an illustration of a much wider phenomenon dividing the common law and civil law traditions. While the former focuses on the particular and factual, the latter emphasises the general and abstract. 155 But is the notion of jurisprudence truly limited to a set of abstract norms expressed in legislation-like language contained in the Cour's decisions? According to Zénati, the Cour 'represses jurisprudence in the name of the legislated law'. 156 What does Zénati mean by this? Do the Cour's decisions not form jurisprudence proper?
For Zénati jurisprudence has a very special meaning of 'prudence of the law' (la prudence du droit). In that respect he refers to the Aristotelian virtue, prudence, 157 which is 'turned towards the action [and] aims at determining the good and the bad for acting well'. 158 Prudence is connected to a concrete factual situation, not a set of abstract norms. Zénati believes that '[w]hile the supreme courts of unlimited jurisdiction found the law in the prudence, the Cour de cassation searched for it in interpretation of the legal rules'. 159 Due to the nature of cassation (concerned with proper application of law, not individual justice in the case being reviewed), there is no jurisprudence of the Cour, properly so called, since the Cour does not decide in the prudential mode. 160 Lower courts play a crucial role here. Unlike the Cour, they are facing 'reallife situations' and see how the abstract rules are being applied in them. 161 They therefore decide in the prudential mode, like courts with unlimited jurisdiction. So when an abstract rule contained in the Cour de cassation's jurisprudence produces results which do not fit the conception of justice which would correspond to the situation before the lower courts, they can always try to provoke its change. The fact that the Cour de cassation's jurisprudence is not officially binding on the lower courts is of crucial importance here. While they do not have the means of moderating the precedent rule like courts below the US Supreme Court, French lower courts are always free to depart from the Cour de cassation's jurisprudence and invite the Cour to change it so that it is updated to reflect the needs of society. 162 In this way they incorporate experience into the life of the law -like doctrinal writers, 163 who are in a constant dialogue with the Cour and can force it to a change. 164

THE EUROPEAN COURT OF JUSTICE: PROBLEMS OF AN INCOMPLETE TRANSLATION
In the foregoing two parts I wanted to show how the conceptualisation of judicial lawmaking and the nature of precedential constraint in the US and France make the involvement of other actors -lower courts and legislators (or, more widely, political branches) -possible. In this section I want to argue that while the Court of Justice's precedent in some respects resembles precedents in both systems, it does not contain (or does not employ in practice) the elements which achieve this. That is why I call this section 'problems of an incomplete translation'. 158 Zénati, n 100 above, 86. 159

JUDICIAL LAWMAKING AS CREATIVE INTERPRETATION AND THE COURT OF JUSTICE'S SUPREMACY
While both systems -the US and the Frenchacknowledge that judges 'make law', they also insist that this judicial lawmaking is based on creative interpretation of the foundational documents -the Constitution and the Civil Code. Both courts are often said to declare what these documents mean; they are not recognised as having an autonomous power to create norms which would be independent of the foundational document. In the US this understanding of judicial lawmaking activity (in the context of constitutional adjudication) allows other branches of government to claim that they can come up with competing interpretations of the Constitution. In France the fiction of 'lawmaking as creative interpretation' allows the legislator to control the Cour by adopting legislative provisions which correct interpretations adopted by the Cour. Of course, this is a rather idealised picture, since the idea of other branches' involvement in constitutional interpretation is deeply contested in the US, and because in France it is possible to take the legislator out of the game -particularly by judicial control of legislation's compatibility with the European Convention, European Union law, and the general principles of law. Moreover, the recent introduction of ex post constitutional review (on references from ordinary courts to the Conseil constitutionnel) has made the legislature's control of the judiciary even more illusory. 165 But these idealisations help to accommodate the normative effects of decisions of both courts beyond the context of the cases in which they are delivered and allow other actors -particularly the political process -to react.
The Court of Justice's judicial activity is also presented as a mere interpretation of law. 166 So, for example, the principle of Member State liability, which goes 'well beyond the terms of the relevant treaties and legislation', 167 is justified as 'inherent in the system of the Treaty'. 168 The need to preserve the image of the Court as a mere interpreter of the law is nicely illustrated by the following statement by Alberto Trabucchi, at the time Advocate General at the Court of Justice: The judicial nature of the activity prescribed by Article 177 [now 267 TFEU] implies in particular its clear distinction from the legislative function. The Court should not, therefore, be called to determine the best way how to regulate a particular case, but to define what the best interpretation that can apply to the act, that is an expression of heteronomy, is. If this fundamental distinction disappeared, the principal pillar of the Community constitution would be wrecked, because the constitution has been conceived as an organisation ruled by the law in the attribution of different powers and with the guarantee of the judicial control exercised precisely by the Court of Justice. 169 Does such understanding of judicial lawmaking have the same effects as in the US and France, allowing other actors to control or at least constrain the Court of Justice's creative activity? I think that this question must be answered in the negative.
First, unlike the US Constitution and the French Civil Code, the Treaties are an unfinished project, the purpose of which is contested. 170 The Treaties are not a fixed point of reference, either in the form of a commitment of 'We the People' to 'form a more perfect Union', as in the US Constitution, nor an enlightened codification of a 'secular natural law ideal of one law applicable to all Frenchmen', 171 as in the French Civil Code. In short, the Treaties are not 'the holy books of the law'. 172 The uncertain nature of the European Union allows the Court to play the role of an institution responsible for giving the European project momentum, which is tacitly approved by the actors with decision-making power. The Court's pronouncements are sometimes even taken as conclusive evidence for the constitutional character of the Union; 173 but it is a generally accepted narrative of European integration that it was the Court that 'constitutionalised' it. 174 As Damian Chalmers notes, 'judicial supremacy has been a central seam in the EU legal order', although he explains that it has applied in an extremely limited domain. 175 Secondly, there are no EU institutions to compete with the Court. In other words, in Europe there is no constitutional voice other than the Court's. Whenever the Court issues a controversial judgment, it is a national government or other representative that complains; such complaints are always treated with suspicion and little serious attention. 176 To my knowledge there has been no declaration from the Council, let alone the Commission, that it felt discontent with the Court of Justice's ruling, comparable to those occasionally made by some governments. 177 Similarly, when a Member State's constitutional court challenges the Court of Justice's interpretation, such challenge is based on the Member State court's own frame of reference, not EU law. When the German Constitutional Court expresses its concern over the Union's exercise of its competences or the Union's respect for fundamental rights, it interprets the German Basic Law, not the Treaties. It is a competition between two distinct normative frames of reference, limited to one jurisdiction -that of the Member State in question.
Thirdly, the Union's legislature is not superior to the Court. As such it is unable to monitor and modify the Court's activity, as is traditional in France. In fact, the contrary is true; the Council (a part of the Union's legislative body) is sometimes presented as an institution which impedes the process of integration and must therefore be controlled by the Court. Mancini and Keeling describe Cassis de Dijon 178 as 'the judgment which best epitomizes "many interesting things you could do with the law"'. 179 There 'the Court imposed on the States a mutual recognition of their respective standards, which practically amounted to rendering the enactment of harmonizing directives unnecessary'. 180 In this way, the Court overcame the requirement of unanimity in Council decision-making, which had hitherto prevented the adoption of much harmonising legislation. 181 Moreover, since the Court derived the requirement of mutual recognition directly from the EC Treaty, it was impossible for the EU legislator to undo its workthe only way would be to adopt an express Treaty amendmentsomething as rare as amendments to the US Constitution in response to the Supreme Court's rulings. 182 To summarise, the fiction of judicial lawmaking as interpretation does not mobilise other actors to limit the power of the Court as it does in the US and French systems, which employ the same fiction.

WHO IS CONSTRAINED?
Since I exhausted the question of the relevance of precedent beyond judicial process in the last section, I will focus here on the relevance of precedent in the Union judiciary, starting with courts other than the Court of Justice.
As Monica Claes rightly observes, 'it has become truism' to think of Member States' courts as true European courts. 183 This is perhaps the single most important difference between the judicial systems of the EU and the US, where the central authority can rely on a complete system of federal judiciary, functionally distinct from the judicial systems of particular states. So although a significant part of US federal law is applied by states' courts, 184 the central (federal) authority can rely on its own system of courts and jurisdictional rules to secure sufficient control over the proper application of federal law throughout the United States. 185 As a consequence of this, much of the federal law is interpreted by lower federal courts, and the opinions of the courts of appeal play an important role in the development of federal law. 186 In contrast, it is in the main the Court of Justice itself which creates the whole body of Union case law in all areas of Union activity. The General Court and the Civil Service Tribunal are the only equivalents to the US lower federal courts, but they have limited jurisdiction and have comparatively less impact on Member States' legal systems. These courts seem to be constrained by the Court's precedent in a way not unlike lower federal courts, a fact which follows from the existence of an appeal. 187 There is potential for the General Court's moderating role, however; Jégo-Quéré, 188 where the General Court invited the Court to change its forty-year-old case law concerning the standing of private parties to challenge EU legal acts, 189 was rather exceptional. 190 The prominent role of the Court of Justice is highlighted by its reluctance to share jurisdiction to hear preliminary references with the General Court, as is possible now that the Treaty of Nice (2000) has introduced Article 225(3) TEC (now 256(3) TFEU). The 'centralisation of the interpretative function, which promotes uniformity' was described by Advocate General Colomer as the main justification for this reluctance, 191 and seems to be shared by other members of the Court. 192 On the other hand, the Court has very limited means of enforcing the authority of its precedents in Member States' courts although it clearly considers that they are binding on them. There is no appeal from Member States' courtsonly the preliminary ruling procedure, which lies entirely in their hands. 193 It is true that, according to Köbler, if a Member State's court of last instance gives a decision 'in manifest breach of the case-law of the Court in the matter', 194 private parties can claim damages from that state. However, such an action can potentially be decided by the same court, 195 and the criterion of manifest breach rather discourages the bringing to court of a liability claim. Infringement actions initiated by the Commission do not seem to be a promising avenue either. 196 So it can be that the authority of the Court of Justice's precedent and the exclusivity of its lawmaking role are 'moderated' by Member States' courts' simple disregard of them.
Moreover, if the Member States' courts were duly to obey the Court's prescriptions, the Court would collapse. Some authors believe that any court 'minded to diverge' from the previous Court's case law, 'must first attempt to obtain a change […] by making a request for a preliminary ruling'. 197 Although we will see that from one possible perspective the Court is given too few cases, compared with both the US Supreme Court and the Cour de cassation, 198 it does not have the means rationally to manage even the (comparatively small!) number of cases it receives. There is no case selection, 199 nor is it possible to deal with references for preliminary rulings in a mechanical way, as is possible in most cases before the Cour, 200 which in addition has far more judges. 201 Perhaps it is as good as it gets?
'The general position is and always has been that the Court of Justice is not bound by its previous decision but that in practice it does not often depart from them,' Arnull says. 202 This would be true if we saw precedential constraint in the binary way of binding/non-binding, which nevertheless does not fit reasoning from precedent. The US Supreme Court can overrule itself, yet it can do so only under certain circumstances, and it is common to talk about 'horizontal stare decisis'. The Court of Justice seems to do the samesometimes.
There are few cases in which the Court of Justice has declared explicitly that it was changing its previous case law. The very first such case was HAG II, where the Court stated that it 'believe[d] it necessary to reconsider the interpretation given in [a previous] judgment in the light of the case-law which has developed with regard to the relationship between industrial and commercial property and the general rules of the Treaty, particularly in the sphere of the free movement of goods'. 203 In Keck the Court explained the need to depart from settled case law concerning the definition of an obstacle to trade caused by 'the increasing tendency of traders to invoke Article [28] of the Treaty as a means of challenging any rules whose effect is to limit their commercial freedom even where such rules are not aimed at products from other Member States'. 204 On the other hand in Metock the Court merely stated that a conclusion arrived at in Akrich, 205 (a previous decision on the point) 'must be reconsidered'. 206 In some cases the Court does not even acknowledge that it is departing from a previous decision. 207 It is therefore not very clear under what circumstances the Court of Justice decides to amend its previous interpretation. 208 Moreover, as some commentators note,  211 Tesauro also did not regard it as 'a revolutionary transition'. 212  We have seen that a conception of a case before the US Supreme Court and the Cour de cassation is important for understanding the nature of precedential constraint. Firstly, the case limits the precedent court's lawmaking activity. The scope of the Supreme Court's precedent can be limited primarily by subsequent courts 213 with reference to some elements of the case which gave rise to the precedent decision subsequent courts can either narrow the holding of precedent or distinguish the precedent case from that before them. The scope of the Cour's legislative precedents is limited by the Cour's duty to reply to the legal grounds of appeal in cassation and nothing more.
Secondly, the case brings elements of 'real life' into lawmaking through adjudication (in contrast to lawmaking by legislatures). While the real-life element is most clearly present in common law adjudication, in most of the constitutional cases the Supreme Court decides in a rather abstract fashion, disinterested in the individual circumstances of the applicant. However, 'real life' enters the adjudicative process before it reaches the peak of the judicial pyramid. It is noted more clearly in the context of the Cour de cassation's adjudication as the prudence of law, implanted into the Cour's jurisprudence through lower courts and also academic doctrine.
How about the case before the Court of Justice and its relevance for precedent?
The answer is complicated by the very different nature of procedures before the Court. These range from preliminary rulings which may concern the identity of two factual situations for the purposes of the application of the ne bis in idem principle, 214 to opinions on the compatibility of an intended international treaty with EU law prior to its conclusion, where the Court decides in a purely abstract context. 215 However the procedure tends to bring the Court closer to the Cour de cassation modela court deciding purely legal questions. The purity is not absolute since the Court has far more flexibility when it comes to its willingness to engage with facts which underlie the legal questions before it. While this is generally acknowledged in the context of the preliminary ruling procedure, 216 it is also true for other procedures, particularly in the light of an increasing tendency to deal with infringements consisting of the misapplication of EU law in concrete situations. 217 The Court can very often choose in a rather arbitrary fashion the level of generality on which it wants to engage with the case, including the level of attention it wants to pay to the facts. 218 However, we have seen that in many of the US Supreme Court's decisions facts are relatively insignificant, and yet a case, defined differently, can play a role in precedential reasoning. Curran explains it in the following way: The common-law recognition of precedents as a binding source of law is blending with the civil-law custom of norm-formation for general prospective deductive application. The manner of applying the norms derived from European Court of Justice precedents is emerging in the civil-law style of privileging the deductive process from norm to application, and departing from the common-law insistence on limiting the applicability of norms derived from precedents to factually analogous future cases. 219 At the same time, the Court is not constrained in what it can say in its judgments, as the Cour de cassation is. Ironically, the Court says both too little and too much in its judgments and possesses a freedom incomparable to that of the US Supreme Court and the Cour de cassation. From this Arnull infers that 'in principle everything that is said in a judgment of the Court of Justice expresses the Court's opinion and is therefore capable of having the same persuasive force'. 220 This appears incorrect, given the occasional practice of the Court's Advocates General and the Court of First Instance of using the distinction between holding and dicta. 221 But to my knowledge there have been no attempts to formulate a coherent approach to this distinction which aims at providing reasons why statements made by the Court can be ignored as mere 'dicta'. And a concept of a casehere different from that before both the US Supreme Court and the French Cour de cassationwould be crucial in such an effort.
The way in which courts communicate depends largely on the conception of authority they adopt. They can either seek to persuade through their precedents, or command by them. This distinction corresponds to two conceptions of the organisation of state authority and officialdom suggested by Mirjan Damaška -  ECR 213,242. 221 Note that Arnull acknowledges that '[o]ccasionally, the Court seeks to distinguish a case on which a party has sought to rely'. ibid, 631. However, distinguishing a case and identifying its ratio are analytically two different things, although they aim at the same result -the avoidance of precedent. By distinguishing the subsequent court seeks to show that precedent is not relevant in the case before it, while by identifying some statement in the precedent judgment as 'dicta' it assumes that it is not binding (even if relevant). coordinate and hierarchical 222 -and is transposed into the organisation of the judicial process. The distinction between persuading and commanding is also reflected in two types of judicial discourseauthoritative or authoritarianexamined (in a rather esoteric way) by Joseph Vining. 223 The US Supreme Court's precedent reflects coordinate authority and authoritative judicial discourse, while the Cour de cassation's jurisprudence/legislative precedent corresponds to hierarchical authority and authoritarian discourse. The Court of Justice seem to lie somewhere between the two.
If we were to examine its nature in Zénati's footsteps, we could be inclined to say that the European Court of Justice is not a court -especially in the context of the preliminary ruling procedure (if being a court primarily means deciding on the rights and duties of parties before it). Again, the diversity of procedures before the Court undermines such characterisation, since it sometimes acts as a court, especially in the appeals it hears from the Court of First Instance. Although it can decide only legal questions, the fundamental difference from the Cour de cassation is that it can directly and conclusively determine the rights of the parties before it.
However, even if the Court's main task were to say what the law is (in the abstract), it cannot act with the same kind of authority as the Cour de cassation. Remember, the Cour is viewed as a 'secular arm of the legislator', which has been delegated the task of supervising lower courts and possibly developing the law if needed. The legislator is believed (whatever the reality) to 'keep an eye' on it and possibly correct its jurisprudence. No such link exists in the case of the Court of Justice. With some exaggeration it can be said to be a secular arm of European integration -but such kind of legitimacy appears deeply problematic, sincecontrary to the ideal of representative government -European integration is a very much contested project.
More importantly, the Court of Justice lacks any sanction by means of which it could impose its authority on Member States' courts, 224 and exercises very limited control in terms of the number of decisions of other courts it can review or guide through the preliminary ruling procedure. In 2009 the Court got 561 new cases and completed 588 cases, and 741 cases have remained pending. 225 Compared to this, in the October 2008 Term, 226 the United States Supreme Court got 8.966 cases and disposed of 7.822, while 1.144 cases have remained on its docket. 227 Finally, the Cour de cassation got in 2009 28.025 new cases and decided 28.594 cases, and 22.165 have remained to be disposed. 228 The Court is therefore dependent on persuasion and cannot limit itself to giving brief unreasoned answers. On the other hand, as we saw in the previous subsection, there are few mechanisms which would limit its law-pronouncing activity. Again, the Court says both too little and too much in its judgments.

CONCLUSION: AN AGENDA FOR FUTURE RESEARCH
In conclusion I would like to suggest possible lines of further inquiry into judicial lawmaking and precedent in supreme courts, informed by the comparative findings offered by this article.
Involvement and communication among constitutional actors beyond the supreme courts, when they give meaning to the 'foundational documents', is an underlying theme of this article. Involvement and communication allow each constitutional actor's competence and legitimacy capital to be combined so that better and more legitimate decisions (in comparison to decisions taken by a single actor) can be taken. I hope to have shown that while the US and French systems seek to achieve this through their particular understanding of judicial lawmaking and precedent, in the EU this is far more problematic. More work is needed to examine how involvement and communication through judicial precedent work in the real context of constitutional adjudication, where the traditional conceptions do not hold. For example, it is difficult to find a satisfactory definition of a case in the context of constitutional adjudication. Is it an abstract issue only? Or do 'reallife' facts play any role in the definition of the case?
A case -either in the form of a real-life situation or as a legal issue presented for the court's decision -constitutes a framework through which the precedent decision can be connected to subsequent cases. There must be some criteria determining why the precedent case is 'relevantly similar' to a subsequent one in which a norm implicated in the precedent is intended to be applied. Schauer calls these criteria 'rules of relevance'. 229 What are the rules of relevance in constitutional adjudication? I hope this article has provided some material for further thought about these questions and, more importantly, shown their practical importance for our understanding of supreme courts' role as adjudicators and at the same time lawmakers. But the real work is only beginning.