I

How, if at all, do social conventions constitute reasons for action? Following George Letsas, my answer is that conventional reasons have three constitutive elements.¹ The first is that there exists a common practice: a regularity of behaviour widely observed by a group of agents. This should be uncontroversial. If there are conventions governing how we dress for work or greet people on the phone, for instance, no doubt they obtain in virtue of descriptive facts about what we say and do around here.²

Yet the existence of a common practice is insufficient for the obtaining of a convention. This is because there are all sorts of behavioural regularities that are not conventional in the requisite sense. Consider the following facts. The one is that everybody drives on the left in the UK. The other is that nearly all Englishmen refrain from torturing people.³ Plainly I have reason not to torture people. But that most Englishmen refrain from doing so is not part of my reason so to act. I would have that reason despite the existence of practices such as amputation and the rack. This is likely why we find offensive the notion that our not torturing people is properly explained in terms of some conventional standard.

In any case, the matter is otherwise when it comes to the rules of the road. Here the existence of a common practice is, it seems, part of my reason to drive on the left. Supposing counterfactually, for instance, that everyone drives on the right in the UK and thus expects other drivers to do the same, then I have reason to drive on the right, not the left. My reason to drive on a given side of the road constitutively depends on the fact that others drive on it too.

This contrast gestures towards a second constitutive element of conventional reasons. Conventional reasons, I shall say, are essentially compliance dependent.⁴ By this I mean that a convention obtains not merely in virtue of facts about the existence of a common practice. Equally relevant is the fact that the material convergence of behaviour is typically a motivating reason to conform accordingly.

There are reasons that explain what we do and reasons that justify what we do. Motivating reasons are of the former kind. Suppose I do not turn up for work today on the strength of a horoscope predicting that some calamity is due to befall me there. Do I have any reasons to act in this way? No, if what we have in mind are normative reasons: considerations that make my action good, right, or justified. Living in fear and letting my colleagues down on the basis of pseudoscience published by charlatans is not the thing to do. Still, my belief in the truth of the horoscope explains my action. For we can cite that belief as the reason I took myself as having; a consideration in light of which I did not turn up for work today; the fact that furnishes my motivation so to act.

Now imagine the following scenario. Anne is getting married. I accept her invitation. The day of the wedding has come. What should I wear? Suppose I reason thus: 'everybody will be wearing a suit so, rather than standing out like a sore thumb, I had better wear one too'. Suppose further that the other guests reason in the same way. Here we have an illustration of the necessary connection between reasons for action and convergent practice that distinguishes social conventions from mere regularities of behaviour. The fact that people converge in wearing suits to weddings explains my action. For that regularity of behaviour is itself a consideration in light of which I chose to wear a suit. The fact that there exists a common practice to this effect is a motivating reason for me to conform accordingly.

But motivating considerations are not the whole story. One of the clear fixed points about social conventions is that we often invoke them by way of governing the behaviour of ourselves and others. As Margaret Gilbert puts the point:⁵

Social conventions are undoubtedly both ubiquitous and influential. However local or temporary, they are always forces to be reckoned with. To give some examples: if there is a convention in my social circle that one send one's hosts a thank-you note after a dinner party, then I court censure if I never send such notes. If I arrive in a country where there is a convention that only close family members ever kiss in public, and I kiss an old friend on meeting her in a restaurant I risk being thought outrageous, alien, insane, or all three. Once I know that there is such a convention in the country where I am living, I have an argument for acting in accordance with the convention, though not necessarily one which will finally dictate how I act. The existence of a convention is, then, apt to encourage conformity to the convention.

Gilbert does well to highlight how conventional practices are often premises in support of conclusions about what we are justified, mandated, or permitted to do. Take her example of a convention requiring guests to send their hosts thank-you notes after a dinner party. What makes it the case that the convention obtains? To be sure, we have to begin by heeding the fact that people participate in a common practice, as well as facts about their motivating reasons for doing so. But we cannot stop there. We would not say that people had a convention concerning the sending of thank-you notes after dinner parties unless they regarded the material practice as binding: something that to some extent or degree ought to be done. The upshot is that it remains for us to give reckoning of how it is exactly that such facts constitute a standard, and thereby serve as a basis for the evaluation or criticism that Gilbert draws attention to in her example.

Part of coming to see why this is a difficult problem is appreciating how the concepts of motivating and normative reasons are simultaneously engaged in our understanding of what social conventions are and how they work. Motivating considerations play an essential role in distinguishing conventional practices from mere regularities of behaviour. Yet normative reasons are also present in our understanding of conventions, since we often invoke them by way of governing the behaviour of ourselves and others. So we have something of a puzzle. Why is it that the concepts of motivating and normative reasons are simultaneously engaged in our understanding of social conventions? How is it exactly that a combination of facts about the existence of common practices and motivating reasons for thus participating in them constitutes a standard to conform?

The basic idea underlying a moralized account of social conventions is that they are necessarily normative reasons that have a composite structure. The thought is that facts about the existence of common practices and motivating reasons for thus participating in them are characteristically elements or non-normative premises in support of conclusions about what we owe to each other. In this way, our moralized account insists on specifying a third constitutive element of conventional reasons: moral principles that constitutively explain what it is about the fact that there exists a common practice of Φ-ing and the fact that people are motivated to Φ precisely because of this convergence of behaviour that justifies conformity to that practice.

Let me proceed to sharpen this claim by discussing an example. Consider the case of driving on the left in the UK.⁶ Ignore the fact that such conduct is legally required. We can easily imagine a possible world where the rules of the road are not subject to legal regulation, and even in the actual world it is a truism that the fact that some conduct is legally required is insufficient to guarantee compliance. (Looking out of my window, it immediately becomes apparent how enacted statutes that prohibit littering, for example, are hardly obeyed and poorly enforced.) At any rate, two facts give me reason to believe that people will continue to drive on the left in the UK. The one is the existence of a common practice to this effect. The other is that people are motivated by the expectation that other drivers will do the same. On my view, the crucial fact of this expectation is necessarily part of a genuine normative reason to drive on the left. Suppose counterfactually that everyone drives on the right in the UK and thus expects other drivers to do the same. The upshot is that I have reason to drive on the right, not the left. My reason to drive on a given side of the road, as we have seen, constitutively depends on the fact that others drive on it too. But motivating considerations are not the whole story. The fact that drivers in the UK converge in driving on the left and are motivated so to act by the expectation that other drivers will do the same is only part of the explanation as to why I have a conventional reason to conform to that practice. Driving is a potentially dangerous activity. Every day it causes accidents that harm vital human interests. It matters that people converge in driving on a given side of the road. In this way, moral principles have a foundational role to play in a constitutive explanation of what makes it the case that the relevant convention obtains. To be clear about this point, the claim is not that facts about the existence of a common practice of driving on the left and the motivating reasons for thus participating in it constitute a conventional standard that we have independent moral reasons to follow. The claim is that moral principles, such as respecting the value of human life, make those facts relevant and thereby explain how it is that they constitute a conventional reason, a real reason, for me to drive on the same side as others.

To sum this up, the mark of social conventions is that they are necessarily normative reasons: a convention obtains in virtue of those moral principles that constitutively explain what it is about the fact that there exists a common practice of Φ-ing and the fact that people are motivated to Φ precisely because of this convergence of behaviour that justifies conformity to that practice. A properly worked out moralized account of social conventions would be a large project and worthy of attention in its own right. In this section, I have been concerned merely to establish the plausibility of this line of thought, since I want to lay greater stress on the claim that it proves fertile in philosophical discussions of the nature of law.

II

Consider our earlier example from trusts law doctrine. That the law requires Jack to manifest his trust in writing ('Φ' for short) constitutively depends, no doubt, on enacted statutes and decided cases. Yet those paradigmatic 'sources of law' are ultimately descriptive facts about what legislatures and courts have said and done. How is it possible, then, that such facts constitute the legal standards that they do? How do the facts, for instance, that 150 Members of the House voted in a certain way and that judges as a matter of settled practice are not disposed to enforce oral declarations of trusts of land make it the case that Jack is legally required to Φ?

Let us refer to this foundational problem of legal theory as the constitutive question. There is an understandable tendency of philosophers to miss the issues here. True, English lawyers have little difficulty identifying the Law of Property Act 1925 as being especially relevant to Jack's case. And yes, Section 53(1)(b) of the Act states quite plainly that: 'a declaration of trust respecting any land or any interest therein must be manifested and proved by some writing signed by some person who is able to declare such trust or by his will'. But the concerns are not so easily dismissed. The constitutive question demands that we provide an account of what it is for the statute to have that effect of legally requiring Jack to Φ. Even if we agree, more abstractly, that enacted statutes and decided cases are relevant in determining what the law requires on a particular issue, it still seems pertinent to ask: how and why is that so? What makes them relevant? What is it exactly about these practices that generate the legal rights and duties that obtain in a given jurisdiction at a given time?⁷

One way of answering the constitutive question is by appealing to social conventions followed by judges and other officials. Let us say that those concerned to defend such a thesis have embarked on the project of legal conventionalism. The root idea underlying that project is that a conventional practice of the courts and other officials, whose role it is to identify and enforce certain standards as law, holds the key to explaining how and why it is exactly that certain acts or events have legal significance in a given jurisdiction. Stated in this way, I appreciate that the root idea is decidedly vague. But there is a very good reason for that: to adopt the root idea as a plausible starting point is not yet to embrace any one specific proposal concerning how that thought may be developed.

Reflect a moment on the explanatory burdens involved when it comes to elaborating the root idea underlying the project of legal conventionalism. It seems clear that the would-be legal conventionalist has to proceed in three discrete steps: first, she owes us an account of what social conventions are and how they work; second, she has to explain how that account is supposed to extend to the legal domain; third, she has to do so in a way that makes convincing the thought that legal standards obtain in virtue of some, by hypothesis, conventional practice among judges and other officials in a given jurisdiction. The present point is that we can sensibly argue and legitimately disagree about how we should go through these three argumentative steps. It should come as no surprize, then, that the project of legal conventionalism, as we shall see, encompasses very different answers to the constitutive question.⁸

Before getting to that, however, let me develop certain aspects of my own view. The previous section sketched a moralized account of social conventions. The time has come to extend that account to the legal domain. Start with the thought that there exists a common practice of judges recognizing certain acts or events as having legal significance in a given jurisdiction.⁹ Understood in the right way, this should be uncontroversial. Philosophers sometimes use the term 'practice' in a thicker sense than I am at present. As we have seen, my general policy is to use that term to pick out descriptive or contingent facts about mere regularities of behaviour not the standards that obtain, when they obtain, in virtue of practices so understood. With that, I propose to sharpen my starting point in the legal case by considering the following counterfactual.

Suppose judges in England and Wales J are not disposed to converge in holding written declarations necessary to create valid trusts of land R. What matters for present purposes is that R is not, it seems, a normative standard that obtains in J. True, the Law of Property Act 1925 says otherwise. But unless and until it becomes a matter of settled practice for the courts to regard statutes emanating from the Westminster Parliament in the UK as constitutive of legal rights and duties-as standards that they have a duty to recognize and enforce in dealing with matters that come before them in their official capacity-it seems odd to say that they have such an effect. The upshot is that legal standards constitutively depend for their existence and content, if only in part, on descriptive facts about what judges consider enforceable in court. And yet, as Dworkin explains, this is just as much a source of perplexity as it is of insight:¹⁰

But it is a formidable problem for legal theory to explain why judges have such a duty. Suppose, for example, that a statute provides that in the event of intestacy a man's property descends to his next of kin. Lawyers will say that a judge has a duty to order property distributed in accordance with that statute. But what imposes that duty on the judge? We may want to say that judges are 'bound' by a general rule to the effect that they must do what the legislature says, but it is unclear where that rule comes from. We cannot say that the legislature is itself the source of the rule that judges must do what the legislature says, because that explanation presupposes the rule we are trying to justify. Perhaps we can discover a basic legal document, like a constitution, that says either explicitly or implicitly that the judges must follow the legislature. But what imposes a duty on judges to follow the constitution? We cannot say the constitution imposes that duty without begging the question in the same way.

From the perspective of our moralized account of social conventions, then, how do we set about explaining a judicial duty to recognize and enforce certain standards as law? What we do, essentially, is rehearse the basic movements in our discussion of how it is exactly that a convention obtains for people to converge in driving on a given side of the road. There are two such movements. The first is to make plausible the idea that the common practice of judges recognizing certain acts or events as having legal significance in a given jurisdiction is compliance dependent. The second is to show how the fact of such compliance dependence is capable of generating a genuine normative reason for judges to conform to the common practice, in particular by identifying the principles or values that make that so.

'[S]urely an English judge's reason for treating Parliament's legislation (or an American judge's reason for treating the Constitution) as a source of law having supremacy over other sources includes the fact that his judicial colleagues concur in this as their predecessors have done'.¹¹ Perhaps, but as Daniel Dennett writes, philosophy students are well advised to avoid using the 'surely' operator in their essays: 'it marks the very edge of what the author is actually sure about and hopes readers will also be sure about'.¹² Let us therefore try to be more systematic in our suggestion that the common practice of judges recognizing certain acts or events as having legal significance in a given jurisdiction is indeed compliance dependent.

As we have seen, a convention obtains only when the participants in a common practice are motivated to conform for certain reasons. Conventional reasons, to reiterate, are essentially compliance dependent: a convention obtains not merely in virtue of facts about the existence of a common practice. Equally relevant is the fact that the material convergence of behaviour is typically a motivating reason to conform accordingly. In the legal case, the suggestion is that judges are motivated to recognize and enforce certain standards as law precisely because of the existence of their shared practice to this effect. How do we make that suggestion plausible? In the main, we proceed by examining motivating reasons in greater detail than we have done hitherto.

What does it mean to say that a motivating reason relation R (p, x, c, φ) holds between a fact p, an agent x, a set of conditions c, and an action φ (e.g. the fact that it is raining is a motivating reason for Adam to take an umbrella to work today)?¹³ Thus far, I have relied on an intuitive notion of motivating reasons as considerations in light of which one acts. But, to tidy this up, it bears emphasizing that a gamut of mental states and propositional attitudes may be relevant in this connection. That x expects, hopes, or fears that p are all candidate considerations that explain her φ-ing. Similarly, that Adam is disposed to take an umbrella to work on rainy days or believes that this is something that he ought to do in these circumstances c are both candidate explanations of why he took an umbrella to work today.

In consequence, the suggestion that judges are motivated to recognize and enforce certain standards as law precisely because of the existence of their shared practice to this effect is not as restrictive as it sounds. Once we realize that motivating reasons comprise a number of different mental states and propositional attitudes, we can afford to take a fairly inclusive view of the types of motivation that judges may have for engaging in legal practice. For instance, it is not required on my view that judges regard their convergence of behaviour as an arbitrary solution to a coordination problem of some description, in particular one of securing uniformity of practice as regards the identification of law. Furthermore, I can perfectly well accommodate the possibility that judges typically have a belief that the legal regime constituted by their shared practice is morally justified, such that they ought to conform to it accordingly. Less savoury reasons for engaging in legal practice, such as a cynical interest in career advancement, are also within the realms of possibility on my view. This is because the only thing I mean to insist on here is that whatever other reasons judges may have for conforming to their shared practice of recognizing and enforcing certain standards as law, they would not do so but for the existence of the shared practice, which in that sense may plausibly be said to be compliance dependent.¹⁴

Even so, it is by no means clear how this generates a duty to conform. How do facts about the existence of a compliance dependent practice of recognizing certain acts or events as having legal significance in a given jurisdiction constitute good reasons for judges to conform accordingly? Our moralized account of social conventions gives us a preview of how the argument will go. In the general case, the basic idea is that the various mental states and propositional attitudes constituting motivating reasons for engaging in a common practice can be normative reasons for the action of others. In the legal case, the claim is that therein lies the key to explaining a judicial duty to recognize and enforce certain standards. Judges converge in determining legal rights and duties in established ways. They are motivated to do so by their expectation that other judges will do the same. Together those facts constitute a genuine normative reason for them to converge in recognizing certain acts or events as having legal significance in a given jurisdiction. They have a duty to enforce certain standards in matters that come before them in their official capacity. The duty obtains in virtue of a plurality of moral principles including, but by no means limited to, fairness and the protection of legitimate expectations. It is those principles that constitutively explain how and why the fact that judges participate in a common practice as well as facts about their motivating reasons for doing so justify conformity to that practice-they make it right that their judgements of what the law requires on a particular issue are in line with that of other judges.

III

Rather than doing so in the abstract, I shall develop this view with reference to a position that Dworkin calls 'conventionalism'.¹⁵ In many ways, this conception of law is superficially similar to mine. It will therefore be fruitful for me to proceed by distinguishing these approaches to the constitutive question, both of which in their own way and for very different reasons appeal to social conventions followed by judges and other officials.

Proceeding in this fashion is also germane to my central thesis. On the one hand, I want to defend the view that there are social conventions at the foundations of law. On the other hand, I do not wish to disagree with Dworkin that certain moral principles feature in a constitutive explanation of legal standards, or even that citizens are entitled to principled consistency in the determination and coercive enforcement of their legal rights and duties.¹⁶ On the contrary, the plan is to show why we do better to reconcile the thought that there are social conventions at the foundations of law with the thesis that certain moral principles make propositions of law true and thus give law its content.

Conventionalism interests me, then, for two main reasons. The first is the centrality of this position, both in Dworkin's influential case against the project of legal conventionalism and, moreover, the arguments developed in support of his positive conception of law as integrity. A congenial way of putting this last point is that conventionalism functions as a picture in Dworkin's thought, by which I mean that it bespeaks a discrete set of assumptions or presuppositions about what our theoretical options are when it comes to pursuing the project of legal conventionalism. In what follows, I shall argue that these assumptions are unduly restrictive-in particular, that they foreclose on the possibility of there being a highly plausible and altogether more sophisticated version of legal conventionalism, one which not only remains untouched by Dworkin's influential arguments, but may also suggest a requisite turn and development in his own conception of law as integrity. Thus, my second reason for wishing to engage with conventionalism is that it acts as a helpful springboard for the development of my own views.

In the interests of space, I shall have to assume broad familiarity on the part of the reader with respect to the relevant backstory in Dworkin's discussion of conventionalism and, therefore, confine myself to the following by way of rudimentary overview. After taking himself to have exposed the shortcomings of the 'plain fact' view of the grounds of law,¹⁷ Dworkin's next thought is that it can nevertheless be recast as an interpretive thesis: more precisely, a constructive interpretation of legal practice that deserves a hearing. Constructive interpretations of this kind have a particular structure. They begin by accepting the proposition that our concept of law picks out the discrete political value of legality, such that the fundamental point of legal practice is to guide and constrain the collective use of force in a given community. In this way, legality demands that the coercive power of government should be deployed only when it is permitted or required by the rights and duties that obtain on account of past political decisions and practice concerning when that use of force is justified. What constructive interpretations of legal practice have to do, as a result, is elaborate on these abstract and provisional starting points. Having begun our enquiry with the thought that legal rights are those that people are properly entitled to enforce on demand though adjudicative and other coercive institutions, the pressing question that we now have to confront is: how do people acquire such rights? In order to answer that question, what we have to do is engage in substantive argument concerning the discrete political value that our concept of law picks out. To be sure the value of legality furnishes a significant constraint on the collective use of force in a given community. But what is the point of that constraint? Why is it valuable? What, if anything, is good about it?

What Dworkin has in mind by conventionalism is best summed up in the following passage:¹⁸

Conventionalism gives an affirmative answer to the first question posed by our "conceptual" description of law. It accepts the idea of law and legal rights. It argues, in answer to the second question, that the point of law's constraint, our reason for requiring that force be used only in ways consistent with past political decisions, is exhausted by the predictability and procedural fairness this constraint supplies; though as we shall see conventionalists divide about the exact connection between law and these virtues. It proposes, in answer to the third question, a sharply restricted account of the form of consistency we should require with past decisions: a right or responsibility flows from past decisions only if it is explicit within them or can be made explicit though methods or techniques conventionally accepted by the legal profession as a whole. Political morality, according to conventionalism, requires no further respect for the past, so when the force of convention is spent judges must find some wholly forward-looking ground of decision.

Although it oversimplifies, we can represent Dworkin as having two main points to raise in response to conventionalism, so understood. First, we can doubt whether the relevant conventions exhaust the grounds of law in a given jurisdiction. For when we consider how judges do not simply put aside their books and make all-things-considered judgements, but rather continue to sensibly argue and legitimately disagree, about what the law requires in the instant case, even when the explicit extension of the relevant conventions has run out, we may be sceptical about the suggestion that they do indeed constitutively determine the existence and content of legal standards. Second, it is by no means clear what if any good reasons there are for judges to slavishly follow the relevant conventions in the legal domain. Assuming we are right to dismiss the plain fact view of law, and thereby take seriously the challenge of explaining what it is exactly that justifies our practice, just what theoretically interesting role remains for the convergent practice of judges and other officials in a constitutive explanation of legal standards? To be sure their beliefs and attitudes may track what it is about their convergent practice that justifies the imposition of some legal standard in matters that come before them. But, then again, they may not. From this it follows that: 'it is implausible to think that any judge's conviction that he ought to decide cases in a "proper" way depends on the convergent behaviour of other judges. A judge would think he should decide in a proper way whatever other judges do or think'.¹⁹

It is precisely for these and other reasons why Dworkin takes his conception of law as integrity to have the better of the argument. The grounds of law are not exhausted by social conventions followed by judges and other officials: such grounds include those moral principles implicit in the decisions and practices drawn attention to by conventionalism and, moreover, presupposed by them by way of justification. There is more to legality than the ideal of protected expectations: it has the expressive value of enabling a political community to speak with one voice-in other words, to act on a coherent set of principles in circumstances of pervasive disagreement about what justice and fairness require, something which citizens are owed in view of their right to equal concern and respect. Only by accepting these claims, so the argument goes, do we present law in its best light. Conventionalism fails as a constructive interpretation of legal practice.

The argument as it stands does not work, however, because it rests on a false dichotomy. According to Dworkin, there are two main options available when it comes to elaborating the root idea underlying the project of legal conventionalism. In this way, the would-be legal conventionalist has something of a dilemma. On the one hand, she could subscribe to the plain fact view of the grounds of law. But this gives rise to a familiar difficulty. That judges and other officials manifest a complex attitude of acceptance of a rule of recognition along the lines suggested by H.L.A. Hart, say, does not make the rule binding or constitute a reason for them to act in accordance with the rule.²⁰ It may be that such dispositions are a necessary condition for the existence of a conventional standard to recognize certain acts or events as having legal significance in a given jurisdiction. But on Dworkin's view, with which I am inclined to agree, such dispositions are insufficient of themselves: the would-be legal conventionalist must proceed by specifying genuine normative considerations that constitutively explain what it is about those dispositions that make them good grounds for the assertion of a judicial duty to recognize and enforce those standards.²¹ On the other hand, we might endorse conventionalism. But, as we have just seen, the indications are that it neither fits nor justifies our practice. So, where do we go from here?

At first blush, the plain fact view and conventionalism do not appear to be false alternatives. On the contrary, initially they seem to be radically different. Certainly both perspectives on the nature of law give social conventions pride of place. But, ostensibly speaking, these accounts diverge when it comes to the role played by substantive considerations of political morality in their respective answers to the constitutive question. Whereas the former appeals to conventions in order to exclude the possibility of those substantive considerations having a pivotal role to play in constitutive explanations of what makes it the case that the law requires what it does, the latter comes at the question of what makes propositions of law true in a very different fashion. Conventionalism, after all, is supposed to be a constructive interpretation of legal practice. In this way, its appeal to social conventions in the legal domain is predicated on a substantive moral claim about what is really good about legality, and a corresponding view of what judges are up to in dealing with matters that come before them in their official capacity.

Appearances can be deceptive, however, and I should now like to suggest that the foregoing dialectic forces us to make a choice between a limited range of options. In actuality, the plain fact view and conventionalism are fundamentally similar. Both positions, I shall say, proceed on the basis of a non-moralized account of social conventions. And both positions, in my view, share a certain picture of what law is and how it works which is altogether antithetical to those of us who are receptive to the notion that law is a department of political morality.

Look again at the earlier passage setting out conventionalism. This conception of law does not proceed on the footing that moral principles such as fairness and the protection of legitimate expectations have a constitutive role to play in accounts of how and why it is exactly that certain conventions obtain in the legal domain. On the contrary, it proceeds on the basis of a familiar two-step procedure which is excellently discussed in the writings of Nicos Stavropoulos on the 'Orthodox View' and Mark Greenberg on the 'Standard Picture'.²² To flesh this out, according to conventionalism, whether certain conventions obtain in the legal domain exclusively depends on facts about the existence of a convergent practice of judges and other officials, as well as facts about their motivating reasons for thus conforming to that practice. This is the basic idea underlying a non-moralized account of social conventions and it constitutes step 1 of the aforementioned two-step procedure underlying conventionalism. To be sure, conventionalism identifies an ideal of protected expectations and certain other independent substantive reasons for conforming to those conventions so constituted. But notice how this substantive argument only emerges at an analytically later stage. Conventionalism does not appeal to moral principles in order to establish that a convention exists and has a particular content. It appeals to moral principles in order to explain why we have reason to follow conventions constituted independently of them.

If the earlier passage does not convince you, reflect on certain other aspects of Dworkin's presentation of conventionalism. The whole point of conventionalism is to deny that moral principles feature among the grounds of law. This conception of law, in other words, is expressly designed to prevent those principles from having that role in a constitutive explanation of legal standards. If it did conceive such a role for principles, then its guiding idea -that legality demands the determination of legal rights and duties to be an explicit and non- contestable affair- could not get started. This is precisely why conventionalism envisages the possibility of there being gaps in the law, and therefore encourages judges to make all-things-considered judgements about how the law should be developed when the explicit extension of the relevant conventions has run out. If principles had a constitutive role to play in its account of what makes propositions of law true, then conventionalism would not, I take it, be so concerned about gaps in the law. Finally, notice how Dworkin often speaks of what the relevant conventions 'stipulate as law'.²³ This matters because it indicates what the conventions do, according to conventionalism, as Dworkin understands it. The conventions determine who counts as a legal authority in a given jurisdiction, and how that authority is to be exercised. As Dworkin says himself, the conventions are principally concerned with: 'who has the power to legislate and how that power is to be exercised and how doubts created by the language are to be settled'.²⁴ So, it is taken as granted here that conventionalism proceeds on the understanding that legal institutions create legal norms by communicating a characteristic intention to do so. And it is precisely for this reason why when Dworkin discusses whether conventionalism justifies our practice his focus is on the instrumental goods that such a conception of law is likely to yield, chief among them of course the securing of valuable uniformity of action in the face of coordination problems.

Having begun with this understanding of conventionalism as an interpretive thesis, it is relatively easy for Dworkin to criticize it in the way that he does. If that claim sounds too strong, weaken it to read that I find those criticisms convincing and am prepared to accept them for the purposes of this paper. I make those concessions, of course, because my view is that Dworkin has only captured one way of elaborating the root idea underlying the project of legal conventionalism, from an interpretivist perspective. There is, it seems to me, and as I shall now go on to explain, another way of proceeding that does not take the orthodox view or standard picture for granted together with its non-moralized account of social conventions.

The subtle but crucial difference between this alternative view, which I favour, and conventionalism, as Dworkin understands it, which I do not, is its commitment to a one-step rather than two-step approach to the constitutive question. Drawing on our moralized account of social conventions, the thought is that moral principles are pivotal in a constitutive explanation of why certain conventions obtain in the legal domain. Unlike conventionalism, we do not take the view that the existence and content of those conventions exclusively depends on facts about the existence of a convergent practice of judges and other officials, as well as facts about their motivating reasons for thus conforming to that practice, and only then raise the question whether there are any good reasons to conform. Such facts do not constitute a conventional standard all on their own, so principles are needed to make it the case that those conventions obtain in the first place. Understood as considerations external to those descriptive facts about convergent behaviour and attendant mental states, the principles identify how, why, in what way, and which of these facts make it the case that the law includes a particular requirement. In this way, the principles do not merely explain why judges have reason to conform to their conform practice. They rather set the standard: they constitutively determine what the practice is, and what it requires on a particular issue.

Perhaps the most natural objection to raise at this point is that I have failed to propose anything other than Dworkin's own answer to the constitutive question. Even if we agree that proceeding in this fashion has the potential to avoid the shortcomings of conventionalism, this is only because I have given up on anything distinctive about the thesis that there are social conventions at the foundations of law. Indeed, there really is no point in pursuing the project of legal conventionalism unless we retain a theoretically interesting role for the convergent practice of judges and other officials in a constitutive explanation of legal standards. As we have seen, this is especially hard to do from an interpretivist perspective. Among other reasons, this is because there does not seem to be anything other than a contingent connection between, on the one hand, the principles or other standards that judges are disposed to enforce in dealing with matters that come before them in their official capacity and, on the other hand, those which actually do, as a substantive matter of political morality, justify the imposition of coercive force: namely, the legal rights and duties that obtain in the instant case.

But I wonder whether the connection is so contingent. As I said earlier, my appeal to moralized conventions in the legal domain is not meant to displace the thesis that citizens are entitled to principled consistency in the coercive enforcement of their legal rights and duties. The thought is that a moralized approach to the project of legal conventionalism helps to put that thesis on its strongest footing, and in so doing may suggest a requisite turn and development in the positive conception of law as integrity.

The basic idea is that legality has the expressive value of enabling a political community to speak with one voice. A legitimate question that we can raise about law as integrity, then, is whether it properly enables such a community so to speak. Compare the following types of principles. On the one hand, those which are implicit in certain political decisions and practice and moreover presupposed by them by way of justification. On the other hand, those which constitutively explain why judges have good reasons to conform to their by hypothesis conventional practice of recognizing certain acts or events as having legal significance in a given jurisdiction. The former are the principles in play on law as integrity. The latter are the principles in play on my view. Both are genuine moral principles, to be sure. But the scope of the latter is considerably narrower.

A long-standing criticism of law as integrity is that it threatens to render law esoteric. In other words, it entails that 'there can be law -lots of law- that no one has ever heard of'.²⁵ Another way of putting this point, which is altogether more germane to my purposes, is that on law as integrity the principles which by hypothesis justify past political decisions and practice are not sufficiently tied to the common practice, propositional attitudes, in short the agency of those whose task it is to identify and enforce certain standards as law in a given jurisdiction. True, legality demands that a political community speaks with one voice. But whose voice? Is it right for a political community to be governed by principles that conceivably have little, if anything, to do with the agency of those persons or institutions charged with the task of developing the law on a daily basis?

In any case, the indications are that no such problems arise on my account, because the principles in play are specifically those which constitutively explain why judges have good reasons to conform to their conventional practice of recognizing and enforcing certain standards as law. If this is right, we may yet have a new take on the project of legal conventionalism: an alternative perspective on why it is that the existence and content of legal standards constitutively depends on social conventions followed by judges and other officials. That this account merits further scrutiny is evident in view of its demonstration of how we can reconcile the thought that there are social conventions at the foundations of law with the thesis that certain moral principles make propositions of law true. Indeed, once we eschew the notion that legal conventions are constituted by facts that fall short of providing judges and other participants in these common practices with genuine normative reasons to conform to them, it becomes much easier to account for the normativity of the conventional practices that form our concern, explain why it is exactly that they have some particular content, and moreover show how their existence is compatible with the presence of substantial controversy among lawyers and judges about what those conventions require in the instant case. The upshot is that we do not, pace Dworkin, have to avoid giving social conventions pride of place in our conceptions of law in order to capture his positive thesis that certain moral principles feature in a constitutive explanation of legal standards. We can have both. We do better, I suggest, to integrate these two competing perspectives on the nature of law.