Law as a Second-Order Essentially Contested Concept

Since Gallie introduced the notion of essentially contested concepts, it has given rise to considerable debate and confusion. The aim of this paper is to bring clarity to these debates by offering a critical reconstruction of the notion of essential contestedness. I argue that we should understand essentially contestable concepts as concepts that refer to ideals or to concepts and phenomena that can only be fully understood in light of ideals and that are, as a consequence, open to pervasive contestation. Moreover, some concepts are second-order essentially contested: it may be a matter of ongoing dispute as to whether they should be regarded as essentially contested. The concept of law is an example. Some authors argue that it is ideal-oriented, whereas others claim it is not. This insight that ‘ law ’ is a second-order essentially contested concept may explain some of the disagreements between legal positivism and its various opponents.


INTRODUCTION
In 1956, WB Gallie introduced the notion of essentially contested concepts. These were 'concepts the proper use of which inevitably involves endless disputes about their proper uses on the part of their users'. 1 Since then, the notion has become popular in various disciplines of the humanities and social sciences. 2 It helps us understand why certain debates are so intensely and unceasingly fought-while at the same time being neither futile nor merely confused.
Even so, the notion has given rise to considerable debate and confusion. 3 First, there is much confusion and disagreement about how the notion of essentially contested concepts should be defined or identified. 4 Frequently, especially in legal scholarship, the term is used in a very loose way as meaning little more than a concept that is 'very hotly contested, with no resolution in sight'. 5 Second, various authors have argued that the notion cannot explain political disagreement; 6 some have even rejected the notion altogether. 7 Finally, it is often controversial as to whether specific concepts, such as 'law' or 'democracy', should be called essentially contested. 8 This paper has three connected aims. First, I want to bring clarity to these debates by offering a critical reconstruction of the notion of essential contestedness in terms of ideal-orientation. Second, I introduce the notion of second-order essential contestedness; some concepts give rise to controversies about whether they are essentially contested or not, because according to some authors, we must understand them in terms of ideals, whereas others advocate that they can be understood in neutral terms. Third, I argue that the concept of law is a second-order essentially contested concept; acknowledging this may help us better understand the debates between legal positivists and non-positivists.
Gallie has suggested seven conditions for identifying an essentially contested concept, but both the status and the precise meaning of these criteria remain vague. Moreover, many authors simply ignore some of them. 9 Therefore, I will begin this paper with a critical discussion of these conditions. I will argue that we should divide them into four semantic and three pragmatic criteria. The semantic criteria are important characteristics of the notion that are crucial for our understanding of essential contestedness, whereas two of the pragmatic criteria can be omitted, because they do not embody necessary requirements. The fifth criterion refers to the pragmatic requirement that there actually is a contest. It allows us to make a distinction between contested and contestable--essentially contested concepts are essentially contestable concepts that are actually contested. Then I will suggest that the reason certain concepts are essentially contested or contestable is that they refer to an ideal or an ideal-oriented phenomenon. Such an understanding in terms of ideals does justice to the most important ideas in Gallie's work, and offers an illuminating framework for analysing normative debates. Finally, I will introduce the notion of second-order essentially contested or contestable concepts. The most important example of such a concept is the concept of law. The insight that law is a second-order essentially contested concept may clarify some theoretical debates, especially those between legal positivists and non-positivists.
Gallie introduces the notion of essentially contested concepts as an explanation for evident and persistent disagreements as to the proper use of certain terms. His examples include 'art', 'democracy', 'the Christian tradition', 'social justice' and 'good'. 10 However, in my view, it is of special importance to law and politics. Both in law and in politics (at least the democratic versions of politics), the existence and legitimacy of disagreement is institutionalised as essential to the core business, in highly regulated court proceedings and in more open parliamentary and public debates, respectively. Hence, a notion that explains the intractability of some of these disagreements is highly relevant for disciplines that study law and politics. Therefore, in this paper, I will focus on legal and political concepts. divide these conditions or criteria into two clusters. The first cluster is one of semantics; these criteria pertain to the meaning of concrete essential contested concepts. According to Gallie, an essentially contested concept is (1) an appraisive concept, (2) which signifies or accredits an internally complex achievement that is (3) initially variously describable and (4) open. The second cluster is one of pragmatics, the use of essential contested concepts in actual debates. This includes the other three conditions: (5) the concept is used both aggressively and defensively, (6) it derives from an original exemplar that all contestants accept as authoritative, and (7) it is plausible that the continuing dispute enables the original exemplar's achievement to be sustained and/or developed in an optimum fashion.
Although Gallie does not use the terms semantic and pragmatic, he singles out the first four conditions as the 'four most important necessary conditions'. 12 Whereas the first four criteria have been generally accepted in the theoretical literature, the three latter criteria are quite controversial, and are often simply omitted. 13 In addition to these seven criteria, there is another important aspect to Gallie's theory: namely, the function of characterising a concept as essentially contested. What is it that we gain-if anything-if we understand that a certain concept is essentially contested? For Gallie, the central point of introducing the notion is that it may explain why some disputes are endless and yet perfectly genuine. 14 In the remainder of this section, I will critically discuss each of the seven criteria; in Section 3, the function of explaining persistent disagreement. 15 I will provide a charitable reconstruction of each of the criteria, partly in my own words, as Gallie's own presentation is often vague and difficult to understand. I will then discuss the criteria, also in light of the vast critical literature since 1956, to decide which of them is essential and which may be disregarded.

The Concept is Evaluative
The evaluative character is the first condition mentioned by Gallie-he uses the term 'appraisive'. Essentially contested concepts need not be purely evaluative; examples such as 'art' or 'democracy' are both descriptive and evaluative. According to Gallie, the partly evaluative character of the concept implies that we cannot 12 Ibid 172, 174. However, in the book version of the article, he skips this qualification, and merely refers to the first four conditions as the conditions that 'suffice to explain how and why the kind of situation might arise' of persistent disagreement. Gallie, Philosophy and the Historical Understanding (n 1) 161. In this version, he also explicitly clusters the first five conditions together as 'the formal--or in a loose sense the logical--conditions to which apparently the use of any essentially contested concept must conform'. So in one clustering, the fifth condition is included, in the other it is excluded. This ambiguous status is reflected in my suggestion in this section, that the fifth condition is not crucial to understanding essential contestability, but in a weak interpretation is the criterion that distinguishes contestable concepts from contested concepts. 13  appeal merely to empirical or logical criteria to decide which conception is the best.
Purely descriptive concepts such as 'tree' or 'water' cannot be essentially contested, because they are not evaluative. This characteristic is uncontroversial, and it is frequently at stake when someone argues that a concept is not essentially contested. For example, Kenneth Ehrenberg has argued that a concept like 'law' is best not regarded as essentially contested, because it can be understood in a normatively neutral way. 16 (I will discuss the example of 'law' In Section 6.) A similar point could be made by those who believe in a purely empirical concept of 'democracy'. If the concept of 'democracy' could be understood without reference to underlying values, we could merely specify the minimum conditions required to call a political system democratic, and there would be no reason for a persistent disagreement after this stipulative definition had been accepted. However, non-positivist authors with a more substantive, value-laden understanding of democracy might reply that democracy is oriented towards the ideal of democracy, and indirectly towards ideals such as equality and human rights, and that therefore we may debate endlessly about what it precisely implies.

The Concept is Internally Complex
The concept involves various dimensions or elements. For a full description, we need to include all these elements, but we cannot simply disaggregate them. When discussing the concept of democracy, Gallie further elaborates on this criterion as that it 'admits of a variety of descriptions in which its different aspects are graded in different orders of importance'. 17 For example, elements of democracy are the majority principle, the equality of all citizens, and continuous active participation; each of these elements should be included in our description, but we may grade their importance differently, and we may interpret each of them differently. The British parliamentary system is in many respects different from the German system; yet both embody defensible conceptions of democracy. We cannot reduce 'democracy' to a one-dimensional concept by defining it in terms of merely one of these elements. Gallie takes a holistic approach to this criterion of internal complexity. Although the explanation of the value of the concept must include 16 Ehrenberg (n 9). Like a number of authors, Ehrenberg incorrectly replaces appraisive with normative rather than with evaluative. However, 'normative' has either a different meaning or a broader one than 'appraisive'. On the one hand, it has a different meaning if we distinguish between values and norms, as standards of evaluation and action guides respectively. Then 'normative' interpreted in a narrow sense as action-guiding means something different than 'appraisive' interpreted as referring to standards of evaluation. If we understand 'normative' as 'action-guiding', 'beautiful' is evaluative or appraisive, but not normative. reference to the respective contributions of its various elements, the value is attributed to it as a whole. 18 The whole is greater than the sum of its parts.
William E Connolly has expanded on this idea by suggesting that essentially contested concepts may be understood as cluster concepts. Cluster concepts are those to which a broad and variable range of criteria apply, and which are themselves also relatively complex and open; these criteria are often cluster concepts as well. 19 To make a cluster concept intelligible, we must demonstrate its complex connections with a host of other concepts to which it is related. 20 Connolly argues that conceptual disputes arise for three reasons: namely, people jointly employing a cluster concept weigh the importance of shared criteria differently; they interpret the meaning of these criteria differently; and they add or drop criteria.
There is a substantial consensus regarding the idea that essentially contested concepts are internally complex. 21 Consequently, a frequent claim is that because certain concepts are not internally complex, they cannot be essentially contested. Kenneth Ehrenberg has elaborated on this negative argument. 22 He suggests that concepts that seem to be essentially contested can sometimes be partly disaggregated by dissecting them into various elements, some of which are merely descriptive, while others might be evaluative. Such disaggregated concepts would then no longer be usefully regarded as essentially contested; only those constituent elements that are evaluative would be essentially contested. To take a simple example: if we could define 'law' in terms of two elements--a system of rules and orientation towards the value of legality--we might argue that the element of a system of rules is a descriptive component, and that only the value of legality is an evaluative component and that only legality can therefore be essentially contested. Then, in line with Ehrenberg's argument, it would be better to disaggregate the concept of law, and to say that 'law' itself is not essentially contested, only 'legality' is.
His argument is analytically correct; a concept that is not complex and not evaluative is not essentially contested. However, it will often be controversial as to whether concepts can be disaggregated in the way Ehrenberg suggests. There may not be a neutral answer to this question; in fact, it is precisely the point on which opposing schools differ. With regard to the concept of 'law', most legal positivists defend that descriptive elements-for example, the sources of law-are enough to define it, whereas most non-positivists argue that if we restrict ourselves to descriptive elements, we miss a dimension crucial to our understanding of law, namely that it is oriented towards the value of legality. 23 18 Gallie, 'ECC', 172; Waldron (n 5) 150. 19

The Concept is Variously Describable
According to Gallie, there are 'a number of possible rival descriptions of [the concept's] total worth, one such description setting its component parts or features in one order of importance, a second setting them in a second order, and so on'. 24 This idea has been widely accepted and fruitfully elaborated on in the well-known distinction between concepts and conceptions, introduced by Rawls and made popular by Dworkin. 25 A conception can be defined as an elaborate interpretation of a concept. Essentially contested concepts allow for a plurality of conflicting conceptions, which is the result of the evaluative and internally complex character of the concept. 26 For Dworkin, there may be different conceptions associated with one concept but, in his theory of law, participants to the legal practice are guided by the idea that one conception is better than the others and that therefore one conception may provide the right answer to a legal question. 27 So accepting that a concept admits of a plurality of conceptions does not imply a relativist position for Dworkin; on the contrary. At first sight, the answer is not so clear in the debate about whether Gallie's notion of essentially contested concepts implies a relativistic or even a sceptical position. 28 Gallie argues 'it is quite impossible to find a general principle for deciding which of two contestant uses of an essentially contested concept really "uses it best"'. 29 However, that does not exclude judging 'the rationality of a given individual's continued use--or in the more dramatic case of conversion, his change of use-of the concept in question'. 30 According to Gallie, a certain piece of evidence or an argument put forward by one side can be recognised, even by its opponents, to have a definite logical force. 31 This is not a sceptical position; there are good arguments for certain conceptions, even if they may not be conclusive in the eyes of everyone. The parties to a debate may sincerely believe that their view is the best. 32 For example, in a court case, both attorneys may sincerely believe that their own interpretation of a 24 Gallie, 'ECC', 172. 25  However, her argument presupposes that the concept refers to a common core, and does not hold against Connolly's interpretation in terms of a cluster concept (she does not discuss Connolly in her article). 27 Of course, Dworkin's right answer thesis has been strongly criticised; I will not go into that debate here. I merely use it to show that accepting a reasonable pluralism of conceptions can be combined with a non-relativist position. 28 Eg, Swanton (n 7) 814. 29 Gallie, 'ECC', 189 (italics in original). 30 Ibid 189 (italics in original). 31 Ibid 190. 32 In Philosophy and the Historical Understanding, 188, Gallie (n 1) explicitly argues that the case for a particular interpretation of a given concept may be regarded as 'advanceable by argument--even if by argument that can lead to no final knock-down conclusion'. certain constitutional clause is the best-even though they may have to acknowledge that there is no knockdown argument that compels everyone to accept it. It seems fair to conclude that Gallie's view is one in support of legitimate pluralism rather than relativism. There is a plurality of reasonable conceptions, each of which may be supported by good arguments. Even opponents can accept some arguments and new pieces of evidence as good and as having logical force, but they need not regard them as conclusive. Sometimes, however, they will be convinced by those arguments or the new evidence, and then their conversion may be justifiable and intellectually respectable. 33

The Concept Must be Open in Character
The concept must have an open and dynamic character, it must admit 'of considerable modification in the light of changing circumstances; and such modification cannot be prescribed or predicted in advance'. 34 The condition of openness largely follows from the three previous conditions, but it adds an important aspect: namely, that the set of possible conceptions is not a fixed one, but may change over time. Even if there is a current consensus on one specific conception, this is always a provisional closure-it may be challenged and adapted in the light of changing circumstances. This is an important insight if we want to understand historic changes and the potential for adaptation of concepts in the light of changing circumstances. 'Democracy' now has a meaning different from that in classic Athens.

Each Party to the Discussion Recognises that its Own Conception is Contested; The Concept is used Both Aggressively and Defensively
Gallie himself seems unclear about this criterion, which is illustrated by the fact that he gives three different formulations. The first formulation states: each party recognizes the fact that its own use of it is contested by those of other parties, and that each party must have at least some appreciation of the different criteria in the light of which the other parties claim to be applying the concept in question. 35 The second formulation states 'to use an essentially contested concept means to use it against other uses and to recognize that one's own use of it has to be maintained against these other uses'. 36 The third and simplest formulation is that the parties use it both aggressively and defensively against other users. 37 The basic idea actually has two elements: the parties to a debate acknowledge that there are alternative conceptions, and they are prepared to defend their own. This condition has often been criticised in discussions of essential 33 Gallie, 'ECC', 190. 34 Ibid 172. 35 Ibid. 36 Ibid. 37 Ibid. contestedness. Even Gallie admits that sometimes this condition need not be present for a concept to be essentially contested. 38 Jeremy Waldron has argued that the required awareness that one's own conception is contested by others is often not present with partisans in actual debates. 39 It is quite unrealistic to expect from ordinary citizens that they have this kind of philosophical reflexivity when they engage in heated political debates. Frequently, they will believe that there is only one valid conception of what 'democracy' or 'justice' entails-namely, their own-and that all others are mistaken. Moreover, even in philosophical debates, let alone in the heat of actual debates, it is not always possible for participants to determine whether indeed the conflict turns on different conceptions rather than on emotional bias, prejudice and lack of education, conflicting interests, or simply confusion. Moreover, if they do not have that awareness yet, but continue to talk past each other, does that mean that the concept cannot be essentially contested? Does it suddenly become essentially contested once the parties become aware that they were using different conceptions?
Gallie does not give clear arguments why precisely this condition is so important. Perhaps he believes he needs this condition to distinguish between essentially contested concepts and other concepts, such as contingently contested concepts or confused concepts, but the condition fails to do so adequately. First, it arbitrarily excludes those debates where the parties are not aware that the core of their disagreement is conceptual rather than merely a conflict of interests or beliefs. Moreover, it makes the distinction variable in an arbitrary way, as a concept could become essentially contested merely because suddenly the parties realise that they have different conceptions-even though the concept itself does not change. Finally, it does not succeed in excluding contingently contested or confused concepts as debates on those concepts may sometimes be raging on in full awareness that the parties have different conceptions. For example, let us assume that I call a certain flower orange and someone else calls it yellow. We discover that we have different conceptions of orange. Initially, we may both be convinced that our conception is the correct one, and thus may be prepared to defend our view aggressively. If after the debate we still disagree, we may agree to submit our differences to the authority of five others. If they all hold that the flower is orange, I should revise my view of what orange means. I may have been confused and, thus, it turned out that the concept was only contingently contested.
However, we should not simply discard the fifth condition altogether. We should skip the first element, the requirement of awareness, which is only included in the two longer formulations mentioned above. The third and shortest version simply states that the parties to a debate use the concept both aggressively and defensively. If this merely means that they appeal to the concept in actual debates, we can accept it-although in a minimal, almost trivial sense. My reconstruction of this requirement is simply that there is actually a debate in which the parties appeal to the 38 Ibid 181. 39 Waldron (n 5) 162; see also Ehrenberg (n 9) 224. concept or to its implications. This makes clear that it is not a semantic criterion, but a functional or pragmatic one: essentially contested concepts are concepts that are used in actual debates. 40 This reconstruction may help us solve an ambiguity in Gallie's work. Not every concept that meets Gallie's semantic criteria is actually contested. For example, there may be a broad consensus about the meaning of 'democracy' within a certain community at a specific time. Gallie merely remarks, 'any proper use of this concept is in the nature of the case contestable, and will, as a rule, be actually contested'. 41 Although Gallie does not elaborate on this distinction between contested and contestable, various authors have done so. 42 My suggestion is that the function of the fifth characteristic is to make the distinction between contested and contestable. Essentially contestable concepts may be defined as those that meet the four semantic criteria. Because of these characteristics, essentially contestable concepts have a clear potential for contestation. However, that potential need not always be realised-for example, because we have reached a temporary closure in the debate, 43 or because open debate is not possible in a society. 44 There can also be-as in law-a generally accepted authority to settle the meaning of concepts. 45 In these instances, although a concept is contestable, it is not seriously contested-at least not at a specific time in the history of a specific community.
I suggest a simple stipulation: namely, that essentially contested concepts are essentially contestable concepts that are actually contested. 46 The reason for this minimal formulation is that it leaves the question open to further empirical analysis as to why, and under which specific conditions, the concept is actually contested. In the remainder of the article, I will use both "contested" and "contestable", depending on the authors discussed and on whether I focus on actual debates or just on the possibility of debates.

There is an Original Exemplar
Gallie introduced this condition and the next one in order to distinguish essentially contested from thoroughly confused concepts. An essentially contested concept is one coherent concept rather than a combination of two separate concepts simply 40 However, the debate may be both at the general level of the concept itself or merely at the level of implications; in the latter case there need not be awareness of the plurality of legitimate conceptions. For example, in debates whether referenda should be included in the democratic system, the focus may be on different empirical assessments of whether referenda would make decision-making more or less efficient and just, whereas the underlying differences in conceptions of democracy are not explicitly addressed although they may influence how important efficiency and justice are perceived to be. In such cases, the concept of democracy is only indirectly or implicitly contested. 41 Ibid 169; see also Gallie, 'Art as an Essentially Contested Concept' (1956) 6 The Philosophical Quarterly 97, 113. 42 Clarke (n 7) 124; Mason (n 32) 89. 43 Collier and others (n 21) 218, 229, argue that concepts can achieve a stable meaning within a given framework. A conceptual debate may then be subject to a temporary closure. 44 Gray (n 20) 336-7. 45 Alasdair Macintyre, 'The Essential Contestability of Some Social Concepts' (1973) 84 Ethics 1. 46 For a slightly different distinction, see Gray (n 20) 338. united under one name. For Gallie, the unity of this concept finds its basis in a common exemplar that all contestants regard as authoritative. All contestants claim to derive their specific conception from a process of imitation and adaption from this exemplar. The exemplar need not merely be one example--it can also be a set of examples or a tradition. 47 According to Jeremy Waldron, condition VI is not necessary. 48 In his view, the unity of a contested concept may also be found in a reference to a common problem or challenge to which different answers are given. This illustrates that there may be different ways in which we could try to distinguish an essentially contested concept from a merely confused concept. Other possibilities could be, for example, referring to a generally accepted formulation of the fundamental idea (eg, democracy is government of the people, for the people, by the people), referring to some common minimum elements (eg, democracy requires at least an equal vote for all citizens and a majority principle) or a combination of these.
Indeed, this condition has been strongly criticised, and has often been simply ignored. Even Gallie does not consider it necessary; he introduces conditions VI and VII as a possibility, as 'conditional in the extreme'. 49 We may safely disregard it, but that leaves us with a double challenge regarding a critical reconstruction. Gallie introduces this criterion for two reasons: to provide a contested concept with a certain unity, and to distinguish it from radically confused concepts. The unity of the concept may partly be found in Connolly's idea of cluster concepts, discussed above. This still leaves us with the challenge as to how to determine whether specific concepts are essentially contested or merely radically confused-a challenge that I will discuss in Sections 4 and 5.

The Continuing Debate Promotes a Better Understanding and a Fuller Realisation of the Ideal 50
When discussing the concept of democracy, Gallie formulates the basic idea as follows: the continuous competition between rival conceptions 'seems likely to lead to an optimum development of the vague aims and confused achievements of 47 Gallie, 'ECC', 176. 48 Waldron (n 5) 158. Numerous authors have criticised the sixth and the seventh criterion; see for example John Gray, 'On Liberty, Liberalism and Essential Contestability' (1978) 8 British Journal of Political Science 385. 49 Gallie, 'ECC', 179. In various places he seems to imply that his sketch is ideal typical (without using the term), for example, when he suggests that not even 'art', 'democracy' and 'social justice' conform with perfect precision to the seven conditions (ibid 180) and that 'each of my live examples conforms sufficiently close to my conditions (I)-(VII)' (ibid 188). In the critical literature, hardly anyone has explicitly noticed these remarks: most authors seem to presuppose that Gallie sees the seven conditions as necessary and sufficient conditions. Consequently, many critiques simply miss the mark, for example, Gray (n 48) 389 (calling the seven criteria necessary conditions). For a similar point see Ehrenberg (n 9) 227 (suggesting that Gallie does not provide an 'exhaustive checklist'). 50 It is plausible or probable that 'the continuous competition for acknowledgement as between the contestant users of the concept enables the original exemplar's achievement to be sustained and/or developed in optimum fashion': Gallie, 'ECC', 180.
the democratic tradition'. 51 Again, Gallie only introduces it as a possibility; it is not an essential condition. There may even be situations in which the continuous discussion has strong negative effects: for example, when it leads to savage political cleavages between the various parties. 52 Gallie's suggestion is a version of the famous marketplace of ideas. The marketplace of ideas metaphor suggests that if we continuously discuss what 'democracy' means, we may sift out less defensible conceptions. For Gallie, the contestation in the marketplace is not merely an intellectual exercise; it also has practical consequences. As a result of this critical reflection, our democratic order may be more successfully adapted to changing circumstances. Moreover, contestation may lead to a higher awareness--and perhaps even stronger internalisation--of the values inherent in the concept, and to a fuller realisation of these values. We should note, however, that this fuller realisation is not a conceptual necessity but an empirical possibility, a potential. It will depend on specific circumstances as to whether this potential is realised, for example, on a society being open and welcoming public debate.
We may therefore conclude that, like condition VI, condition VII does not identify a necessary criterion. However, we may treat it as identifying a potential-which is in line with Gallie's own views. In our critical reconstruction of the notion of essentially contested concepts, we should try to take this potential into account, and to see whether we can do justice to the intuition behind it.
I have now discussed in detail each of the seven conditions suggested by Gallie. The conclusion is that the first four criteria, which are semantic, are crucial to a good understanding of essentially contested concepts. Each of them offers part of the explanation as to why conceptual contests are possible, and this explanation is found in specific characteristics of the concept. However, an analysis of Gallie's three pragmatic conditions leads to the conclusion that only one of them should be regarded as an inherent part of the notion of essentially contested concepts, and serves to distinguish them from essentially contestable concepts. Condition V should be reconstructed as the requirement that the concept should be appealed to in actual debates in order to be essentially contested rather than contestable. In light of the criticisms discussed above, condition VI (the reference to the exemplar) should be dispensed with altogether. Condition VII identifies merely a potential, namely, that the contestation may promote a fuller awareness and a better realisation of the values inherent in the concept.

EXPLAINING PERVASIVE DISAGREEMENTS IN LAW AND POLITICS
For Gallie, the primary reason to introduce the notion of essentially contested concepts is to explain the persistent disagreement regarding them. His focus is not on an empirical analysis but on a philosophical understanding of what makes these 51 Ibid 186. 52 Ibid 179. debates intractable. 53 Gallie suggests that there is also a gain for participants in the debates of this 'higher order' recognition. 54 If parties to a disagreement understand that a concept is essentially contested, this may raise the level of quality of the argument. They may see rival conceptions as being of permanent potential critical value to their own interpretation. 55 Gallie's suggestion that the notion of essentially contested concepts may help us understand the persistent nature of debates has been accepted in various disciplines, but it seems to have special traction in the study of law and politics. Both for law and for politics, dealing with clashes of opinions is not merely accidental, but is essential to the practice. Jeremy Waldron suggests that endless disagreements belong to the circumstances of politics, but I suggest that they also belong to what may be called the circumstances of law. 56 Legal processes designed to deal with conflicting views and interests are central to the practice of law. This is not true for other practices such as art, morality or religion; in these practices, clashing views may be a fact of life, but the practices are not organised in order to provide an adequate response to these conflicts. In other words, understanding and dealing with these conflicts is not central to these practices.
However, although law and politics are both institutions that address pluralism and conflict, they do this in diverse ways. These differences may explain why in legal theory the notion of essential contestedness has been widely accepted, whereas in political theory it is much more controversial as an explanation for political disagreements. 57 Legal processes are structured as a clash between prima facie equally legitimate conflicting views on what a specific rule, principle or value implies as applied to the facts of the case. The practice of law in the context of courts is primarily one of normative argument and interpretation in which the interpretation of authoritative texts is central. It is an undeniable fact that reasonable and highly expert parties have conflicting views on the contents of the law. In many cases involving a highest court, attorneys on both sides provide a fully defensible interpretation. In those legal systems where dissenting opinions exist, it is clearly visible that even the judges themselves often respectfully but strongly disagreeand certainly not only in strongly politicised conflicts but also on more technical legal issues.
Many legal debates can be greatly clarified by understanding them as disputes on the meaning of central concepts such as 'privacy' or 'due care'. It is precisely 53 Gallie, 'ECC', 168, argues that the human activities to which these concepts are related belong to the disciplines of aesthetics, political philosophy, philosophy of history, and philosophy of religion. Later in the article, he also discusses ethics. He does not mention any non-philosophical discipline. 54 Ibid 192. 55 Ibid 193. See also Connolly (n 19) 40-1; Waldron (n 5) 151-3. 56 Jeremy Waldron, Law and Disagreement (Clarendon 1999) 102. 57 This is not a claim about how frequent the notion is being used: in both disciplines it is quite popular in the academic literature. It is a claim about how useful the notion is in explaining disagreement. In political theory there are various authors arguing that the notion of essentially contested concepts is radically mistaken or cannot account for explaining political disagreement; for references see notes 57 and 58. In legal theory, I have not found similar arguments.
the point of judicial processes that they restrain conflicts by reducing them to reasonable disagreements about the implications of legal doctrine for concrete cases. Therefore, the notion of an essentially contested concept provides a fitting explanation for pervasive disagreements in law. Moreover, it can also explain and justify why two competing interpretations of a legal doctrine can both be regarded as legitimate and defensible. It is therefore not surprising that the notion has gained a broad popularity among legal scholars-although this popularity is not so widespread when it comes to the concept of 'law' itself, as I will discuss in Section 6. In political theory, however, the notion is more controversial. Steven Lukes and William E Connolly have argued that political core concepts such as 'power' and 'democracy' may be regarded as essentially contested, and that this may explain the persistence of political disagreements. 58 According to Connolly, conceptual contests are central to politics. 59 This is often called the 'essential contestability thesis'. However, many political theorists have attacked the thesis. Some critics deny that political concepts can be essentially contested, 60 while others simply argue that this does not provide a general explanation for disagreement. 61 Or to put it more precisely, analysis in terms of essentially contested concepts is at best largely irrelevant, but usually blinds us to the more important sources of disagreement. 62 It is easy to understand why these critics doubt the explanatory power of the notion of essentially contested concepts. If we study political conflicts, we may see that they are not only--and usually not even primarily--reasonable disagreements about the meaning of shared concepts such as 'justice' or 'democracy'. First, not all concepts playing a role in political debates are shared. 'Solidarity' is a concept much more at home in socialist than in conservative parties; 'sharia' is a concept not shared by secular or Christian-democratic parties; and animal rights are only defended by some parties. Whereas in law the central concepts have been authoritatively laid down, sometimes implicitly, in authoritative sources, this is only partially true for politics. Second, and more importantly, we should notice that apart from ideological differences on how to interpret normative concepts, political conflicts are also about protecting specific interests and about the distribution of power and money. To understand political conflicts, we must also refer to historical, economic, sociological and psychological theories about belief formation, power, interest groups and so on. 63 58 Steven Lukes, Power: A Radical View (MacMillan 1974); Connolly (n 19). 59 Connolly (n 19) 6. 60 See Swanton (n 7); Clarke (n 9) 125: 'the notion of an essentially contested concept is radically mistaken.' 61 Gray (n 48) argues that the notion may have been useful once, but now constitutes an impediment to progress. 62 For an intermediate position arguing that essential contestedness provides important insights but needs to be complemented by a much broader interdisciplinary research perspective, see Mason (n 32). 63 Of course, Legal Realism and Critical Legal Studies have successfully argued that for a full understanding of legal conflicts we should also go beyond legal doctrine. However, legal processes are structured so that, in the end, the conflict is reframed into a doctrinal dispute in light of common authoritative sources.
Consequently, there are many possible foci for understanding incessant disagreement in politics. It is unlikely that an analysis in terms of the semantics of key concepts in political debate will always provide the most powerful explanation regarding disagreement-even if it may provide part of the explanation. The attractiveness of the notion of essentially contested concepts for political theorists will therefore depend on whether in their research design the primary focus is on ideology or on other factors.

A CRITICAL RECONSTRUCTION IN TERMS OF IDEALS
Our analysis thus far has identified certain challenges regarding a critical reconstruction of the notion of essentially contested or contestable concepts. First, fit: every critical reconstruction should include the four semantic conditions. Second, function: it should explain how the notion might help us to understand why some disagreements are pervasive. Third, unity: it must show how these four conditions and the function hang together. We have to construct a coherent theory of the notion with a point that can explain the characteristics rather than merely a theory in which a number of contingent characteristics are loosely connected. Fourth, distinctiveness: we must be able to distinguish between essentially contestable concepts and similar concepts such as radically confused concepts. Finally, robustness: a reconstruction should attempt to do justice to the two hypotheses with regard to conditions V and VII.
The two most important challenges are fit and function. Various authors have emphasised the function or the use of the notion of essential contestation. However, they differ in whether and how the function and the semantic or descriptive criteria should be combined. Carver even completely dismisses the semantic criteria and derives essential contestedness purely from how concepts are used in essentially contested arguments. 64 In my view, this turns Gallie's notion upside down, and transforms it into an extremely broad-and thus unproductive-label. Ehrenberg treats all descriptive criteria as subordinate to the function. 65 For Ehrenberg, the function is predominant, and the descriptive criteria are only relevant ('a starting point') in as far as they are helpful to understand the function. I suggest that the semantic criteria and the function are of equal importance. We should find an understanding of the notion that can explain the four semantic criteria in light of the function, and vice versa. The semantic criteria-the descriptive characteristics that a certain category of concepts have in common-can only be fully understood in light of the function, and the function can only be understood in light of these characteristics.
My thesis is that these challenges may best be met by understanding essentially contestable concepts in terms of ideal-orientation. The suggestion that essentially contested concepts refer to ideals has been made by Gallie and various other 64 Carver (n 13). 65 Ehrenberg (n 9) 227. authors. 66 The two core characteristics in Gallie's presentation are that the concept is evaluative and is internally complex. We may reconstruct this to mean that the concept refers to an ideal or to a phenomenon that is oriented towards an ideal. Or to provide a more precise definition: essentially contestable concepts are concepts that refer to ideals or to concepts and phenomena that can only be fully understood in the light of ideals, and that are, as a consequence, open to pervasive contestation.
The understanding of ideals used in this analysis is a philosophical, more precise one than the loose common-sense notion of ideals in which an ideal may refer to ambitious or vague goals. 67 Ideals may be conceived as complex values that are usually not completely realisable, and that are often difficult to formulate exactly. We can never grasp the meaning of ideals completely; there is always a surplus of meaning. Ideals partly transcend contingent, historical formulations and implementations. Therefore, they must be distinguished from mere goals, both because they are more complex and because it is usually impossible to fully formulate or to fully realise them. Although we may loosely say that Michael Phelps' personal ideal was to win Olympic gold, and this certainly is unrealisable for most of us, it is not an ideal in the philosophical sense at stake here. We can formulate precisely what it is, and we know how to determine whether Michael Phelps did succeed. Therefore, winning Olympic gold must be considered a goal rather than an ideal.
Usually the focus in discussions of ideals is on the fact that they cannot be fully realised. 68 However, for understanding essentially contested concepts, it is the fact that their meaning can never be formulated exactly which is the most important. therefore provide additional valuable insights. As a result, every conception is always inherently open to contestation. The reference to ideals in essentially contested concepts may be direct or indirect. For example, concepts such as 'democracy', 'rule of law' and 'justice' directly refer to ideals. 70 On the other hand, concepts such as good care refer indirectly or implicitly to ideals, and it may not always be immediately clear that they do so. Standards of good care can be understood in terms of minimum rules of care, but they also have an aspirational dimension, which refers to an ideal of a good caring relation or, in the context of health care, to the ideal of a good doctor or a good nurse. 71 As I will discuss in Section 5, many legal terms, such as 'negligence' or 'secret elections' only indirectly refer to ideals, and it may depend on the degree of elaboration of those concepts in a specific jurisdiction whether it is useful to regard them as essentially contested.
A reconstruction of essentially contestable concepts in terms of ideals provides a response to the challenges I distinguished in the beginning of this section. Not only does it explain the four most important characteristics and the function, it also provides unity and distinctiveness. Therefore, a reconstruction in terms of ideals presents the most charitable and most useful reconstruction of Gallie's theory. I will show now in more detail that it can meet each of the challenges.
Fit. If we understand Gallie's reference to a valued achievement as a direct or indirect reference to ideals, we can construct a coherent theory of essentially contestable concepts that accounts for all four semantic criteria. Ideals are values, and are therefore evaluative (condition I). They are complex or, in Connolly's terminology, they are cluster concepts (condition II). In order to understand ideals, we must refer to other complex ideals and constitutive values, as well as to intermediate principles and rules. Each of these other elements, in turn, is also related to further ideals with which they are associated, and thus may often be only understood in the light of encompassing theories. 72 For example, the ideal of democracy relies on ideals such as freedom, equality, human rights, and tolerance; it refers to intermediate principles such as the majority principle and one-man, one-vote; it is related to-although it does not fully comprise-the ideal of the rule of law. In their turn, the ideals of freedom, equality, human rights, and tolerance are also complex. Even the more specific ideals (for example, freedom of religion) are complex, though much less complex than general ideals such as the rule of law and democracy.
Gallie's conditions III and IV can also be easily accounted for if we understand essentially contestable concepts in terms of ideals. As I argued above, ideals have a 70 This can also be a negative reference: 'injustice', 'terrorism' and 'rape' have also been regarded as essentially contestable concepts: Collier and others (n 15) 216. 71  surplus of meaning, which implies that there can always be different and conflicting interpretations. A concept referring to an ideal, thus, may give rise to a number of competing conceptions, or, in Gallie's terms, it must be variously describable (condition III). Because ideals always have a surplus of meaning and can never be completely realised, they are also important factors in enabling change. 73 In Gallie's terms, they are open and admit of variation in changing circumstances (condition IV). Depending on variations in context, new dimensions of an ideal may become appreciated, and some dimensions may be seen as more important than others in the specific social conditions. For example, since the initial formulation describing privacy as 'the right to be left alone', 74 our understanding of the ideal of privacy has changed drastically in response to social change and, more in particular, to technological developments.
Function. Ideal-orientation may also account for the function, as it explains why pervasive disagreement is possible. Ideals have a surplus of meaning, and can never be completely formulated. Every attempt to precisely formulate the meaning of an ideal is thus open for contestation, because it inevitably has left out some aspects of the ideal, or because it weighs the different values and principles comprised in the ideal in a manner that is at variance with that of some critics. Although each of the parties may believe her own conception to be the best, there is not one conception that in abstracto is superior. 75 The parties may have good reasons to prefer their own conception, for example, that it emphasises precisely those dimensions of the ideal that are most important for dealing with topical problems, or that it provides the fullest account of the ideal in a particular context. For instance, a theory of democracy in terms of equality-operationalised in terms of one-man, one-vote-may be more attractive in the context of a state than in the context of industrial democracy; in the latter, elements such as accountability, participation, and critical discussion may be more important. However, there are no conclusive arguments as to why one conception of democracy is the best one in general. Thus, debates about the meaning can at best result in a provisional closure, but these closures are always open to new contestations.
Unity. A certain degree of unity and distinctiveness is already provided by the qualification of the four most important criteria as semantic, and by the insight that essentially contestable concepts are cluster concepts rather than merely 73  -namely, that there is always one right answer to a dispute. I suggest that a better interpretation of this thesis is that the parties to a legal dispute sincerely believe that there is one answer for which there are better arguments than for the alternatives, at least in the specific context of that legal dispute. However, as Dworkin interpretations are notoriously controversial, I will not discuss the question here whether this is the best interpretation of his work. I merely want to put my interpretation forward as the best way to understand why ideals may give rise to pervasive disagreement in line with Gallie's theory. For Gallie's view see footnotes 29-32 and the accompanying text.
based on family resemblance. However, the insight that essentially contestable concepts refer to one common ideal that can give rise to different and even mutually exclusive conceptions of the ideal, as explained above, provides a more substantive explanation of the concept's unity.
Distinctiveness. In order to be useful, the notion of essential contestability should mean more than merely hotly contested. Not all concepts that admit of competing and incompatible conceptions should be qualified as essentially contestable. Definition in terms of ideals makes it possible to distinguish essentially contestable concepts from various other types of contested concepts, such as open-textured and interpretive concepts.
First, essentially contestable concepts should be distinguished from those having an open texture. 76 One reason is that open textured concepts can be descriptive, whereas essentially contestable concepts are always evaluative. The notion of the open texture of ordinary language, as discussed by HLA Hart, primarily provides an explanation for disagreement at the penumbra of descriptive concepts like 'vehicle'. 77 We may debate about whether a bicycle and roller skates qualify as vehicles, but in principle this conflict can be solved by accepting a stipulative definition. After that authoritative settlement, there is no reason for contestation left -although new penumbra cases may always arise that ask for further specification, such as skateboards. The notion of open-texture of language is especially helpful for understanding debates on what Dworkin calls vague criterial concepts. 78 We may yet be in search of a more precise demarcation criterion, but there is nothing that makes a more precise criterion categorically impossible. Hart's famous 'no vehicles in the park' rule contains a vague criterial concept, but this vague concept of vehicle does not refer to ideals, and after we have settled the vagueness, there is a clear rule.
A second reason why essentially contestable concepts and open textured concepts are not identical is provided by Jeremy Waldron. He remarks that the contestation in the case of essentially contested concepts is not merely at the penumbra but is at the core of the concept. 79 The disagreement is not about marginal or borderline cases but about paradigm or core cases. Some may hold that the United States is a core case of democracy; others may hold that it lacks important features of social democracy. According to Waldron, this disagreement is about the core or essence of democracy. Similar points can be made with regard to the concept of 'privacy': whether homosexual conduct between consenting adults in one's bedroom is protected under this right is not a matter of the penumbra. 80 So an analysis in terms 76 On the difference between open textured and essentially contestable concepts, see MacIntyre (n 45). 77 Hart (n 25) Ch 7, s 1. 78 Ronald Dworkin, Justice in Robes (Belknap Press 2006) 9. On the difference between interpretive, criterial, and natural kind concepts, see also his Justice for Hedgehogs (n 72) 158-60. Natural kind concepts can have an open texture too, but in the context of our discussion, this is irrelevant, since natural kind concepts are descriptive and therefore can never be essentially contested. 79 Waldron (n 5) 149. 80  of core and penumbra associated with the notion of open texture of law is different from that of an analysis in terms of essentially contestable concepts, which often concerns the core meaning of the concept. This contestation at the core can be explained in terms of the surplus of meaning of ideals.
Second, essentially contestable concepts should not be identified with interpretive concepts as defined by Dworkin; they constitute only a subclass of the latter. Dworkin suggests that interpretive and essentially contested are synonyms, but, in my view, this is a mistake. 81 In the class of interpretive concepts, Dworkin includes some criterial or natural kind concepts, namely those that become temporarily contested because they are embedded in law in a specific way. An example is the concept of 'book' in a new sales tax, which, if left undefined, might give rise to interpretive conflicts (in fact, this example is similar to the 'no vehicles in the park' example). 82 However, the concept of 'book' is not evaluative, so it cannot be essentially contested according to Gallie's analysis. There are many concepts in need of interpretation because they are controversial or vague, but calling all these concepts essentially contested would erase the distinction between simply contested and essentially contested. The mere fact that a concept is interpretive and therefore contested does not make it essentially contested. 83 Robustness. The reconstruction of essentially contestable concepts in terms of ideals is successful in terms of fit, function, unity, and distinctiveness. This is enough to accept it as a critical reconstruction. However, it also offers interesting perspectives on Gallie's condition VII, which I reconstructed as merely identifying a potential, namely, that the contestation may promote a fuller awareness and a better realisation of the values inherent in the concept.
If we rephrase it in terms of an ideal rather than of an original exemplar, this condition identifies an important insight. When participants or external observers acknowledge that every formulation of an ideal, every conception, is only partly adequate, they may regard alternative formulations as providing important insights and helping to keep the full implications of ideals on the table. Continuous contestation may prevent a permanent reduction of the meaning of the ideal to merely one conception or to one ideological interpretation. The conception that has been implemented through legislation by a majority in the past may have gained 81 Dworkin (n 78) 221. 82 Dworkin (n 72) 164-5. 83 My argument can also be made differently. I argue that there is an inconsistency in Dworkin's work.
According to Dworkin, interpretive concepts are essentially contested concepts; according to Gallie, essentially contested concepts are appraisive or evaluative; concepts such as 'book', 'bald' or 'lion' are not evaluative, and thus cannot be essentially contested; therefore Dworkin's suggestion that these concepts can be temporarily interpretive cannot be correct--at least not if he wants to stick to Gallie's theory. In the text, I solve this inconsistency by denying his identification of essentially contested and interpretive concepts. Another alternative would be to restrict Dworkin's definition of interpretive concepts, and exclude descriptive concepts that have become controversial merely because they are embedded in a normative context. This would make it possible to regard his examples of 'book', 'bald' and 'lion' as examples of criterial and natural-kinds concepts that can temporarily be contested, but not essentially contested. I believe this would be a better solution, but defending this claim would require a much more elaborate exegesis of Dworkin's work than is possible in the context of this article.
institutional authority, but it is never the last word. With its surplus of meaning, the ideal always provides a source of criticism and renewal. However, the ideal merely presents a potential for contestation and for an improvement of reality. It is very well possible that a specific interpretation of the ideal has become so widely accepted, and is so strongly engrained in law and other institutions, that it is almost impossible to contest it, let alone that a contest leads to an effective challenge of the dominant conception. There is no guarantee that a continuous contest will lead to an optimal realisation of ideals. Nevertheless, the insight that ideals--and therefore essentially contestable concepts as well--contain a potential for criticism and improvement is an illuminating one, both for social activism by participants in the contestation, and for scholars analysing the development of certain debates and social practices.

A CONTROVERSIAL NOTION: SECOND-ORDER ESSENTIAL CONTESTABILITY
It is often a matter of controversy as to whether a concept is essentially contestable. A trivial explanation regarding some of these controversies has to do with the loose way the notion is often used. Almost every concept that plays a role in debates has been called essentially contested. Jeremy Waldron rightly remarks, 'The idea is clearly vulnerable to overuse'. 84 Another explanation for the controversies is that a concept may be contestable but not actually contested. My critical reconstruction offers two more fundamental explanations as to why there is sometimes not merely a first-order contest about the meaning of the concept in light of certain ideals, but also a second-order contest about whether a concept is essentially contestable. These explanations are associated with the central role of ideals in understanding essential contestability. The crucial point is that it may be strongly controversial whether a concept should be partly understood in terms of ideals. Moreover, as ideal orientation is a gradual characteristic, essential contestability and essential contestedness are also gradual rather than digital characteristics.
Some authors understand social phenomena in terms of functions or intentions of actors; this usually involves referring to the values or aspirations inherent in the phenomena. For example, Philip Selznick has argued that we can only understand social phenomena such as law, family, or friendship in the light of their 'master ideals'. 85 Other authors try to understand and operationalise social phenomena in terms of empirically observable characteristics. A reference to intentions, let alone to ideals with their surplus of meaning, would undermine this attempt. Consequently, most positivists tend to prefer a value-neutral understanding of social concepts, whereas different types of non-positivists hold that this is impossible. 84 Waldron (n 5) 148. 85 Selznick (n 23).
For the first group, those concepts are not essentially contested, but for the latter they are. 86 Concepts that give rise to controversies about whether they are essentially contested may be named second-order essentially contested concepts. 87 A second-order essentially contested concept may be defined as a concept that is reasonably regarded as essentially contested by some, whereas others reasonably deny that it is essentially contested. A second-order essentially contestable concept may be defined as a concept that can reasonably be regarded as essentially contestable by some, whereas others can reasonably deny that it is essentially contestable.
Every first-order essentially contested or contestable concept is also secondorder essentially contestable. For every concept that is treated by some authors as referring, directly or indirectly, to an ideal, someone might come up with a conception that does not refer to ideals. Even a concept as 'justice' could be interpreted in clear descriptive terms like 'giving everyone exactly the same'. Of course, it is highly unlikely, and most people would say that this does not fully grasp the meaning of the concept-and, as a philosopher, I might add that the reason is that this conception fails to do justice to the ideal-oriented character of the concept. Even so, it is not unthinkable that someone might come up with a valueneutral, purely descriptive conception of the concept. This theoretical possibility exists for all essentially contestable concepts.
This shows that every essentially contested concept is also second-order essentially contestable. However, not every such concept will actually be second-order contested. For some concepts, the value-laden character is difficult to deny. The evaluative dimension is central to our understanding of the concept, and this evaluation refers to a complex achievement rather than one that is simple and measurable. For such concepts, the ideal-orientation of the concept cannot reasonably be denied, and therefore the essential contestability cannot reasonably be denied. Examples are 'justice' or 'fairness'. These are essentially contested concepts that are not second-order essentially contested.
The notion of second-order contestability is thus quite trivial because all essentially contested/contestable concepts are also second-order contestable. However, the notion of second-order contestedness is an important one. I suggest that second-order contested concepts can be part of the explanation for some theoretical debates in the humanities and social sciences, but in the context of this article, I cannot explore that suggestion. I will only illustrate it with one debate, namely that between the various schools in legal philosophy, in the next Section. 86 An exception in the positivist tradition is Radbruch who holds that law is oriented towards ideals or values, but combines that in his early work with a legal positivist theory. Gustav Radbruch, Rechtsphilosophie (8ste Aufl, Hrsg von E Wolf und H-P Schneider) (KF Koehler 1974). 87 The possibility of a second-order contestability of the claim that a concept is essentially contested or contestable has been mentioned by various authors; for instance, by Gray (n 20) 339. I briefly mentioned the idea that law is a second-order essentially contested concept in Wibren van der Burg, The Dynamics of Law and Morality: A Pluralist Account of Legal Interactionism (Ashgate 2014) 47; this article is an attempt to elaborate this tentative suggestion.
Before doing that, however, there is one further observation to be made. Essential contestability and contestedness are gradual notions. 88 The degree of contestability may vary with the complexity and with the aspirational character of the concept. Ideals can be more or less complex; the more complex a notion, the more it is open to be essentially contested. 'Justice' and 'the good society' are more complex than 'freedom of religion'; hence there is a greater contestability.
A concept can also be more strongly contestable if its aspirational character is stronger. Concepts that directly refer to ideals are more strongly contestable than partly descriptive concepts that more indirectly refer to ideals. In some concepts, like 'democracy', the ideal dimension is clearly visible. In other concepts, like 'secrecy of elections', it is only very indirect. The secrecy of elections can be described as an element of the complex ideal of democracy, yet in most jurisdictions it has been legally elaborated upon in very precise formulations. Thus, the legally enforced conception of secret elections only indirectly refers to the ideal of democracy, and may sometimes be scarcely contested. As a result, the ideal of democracy only plays an indirect and marginal role in our understanding of the secrecy of elections, and thus the concept of 'secrecy of elections' is only marginally essentially contested.
For practical purposes, in these marginal cases we may analyse the concept adequately without reference to underlying ideals. It only makes matters unnecessarily complex if we nevertheless call it essentially contestable-even if it can be shown theoretically that it is. This will especially be true for many legal concepts that have been elaborated upon in statutes and case law, resulting in a relatively complete and coherent legal doctrine that is generally accepted. Examples may be concepts like 'negligence' or 'medical malpractice'. Although they indirectly refer to ideals such as good care or good medical practice, in positive law they have been so robustly elaborated upon that a reference to these underlying ideals is usually not useful in practice. Still, there may be occasions when we do have to refer to those ideals, in order to re-examine and reconstruct our conceptions: for example, when social views change regarding responsibility for harm to future generations, or when new medical technologies are discovered. It may also be true for moral concepts or political concepts on which a strong overlapping consensus exists -at least in the relevant forum. For example, within a specific jurisdiction, there may be a broad consensus both on what 'democracy' implies and on the judgment that this society is democratic.
The gradual character of essential contestedness may lead to disagreements on whether a concept is in fact essentially contested. An example is provided by Arnold Heidenheimer, who argues that, since the erosion of communism, 'democracy'-one of the central examples in Gallie's work-has become less essentially contested, and that it now should be considered a formerly essentially contested concept. 89 It may be doubted whether this judgment is empirically correct. Compare views on democracy in the US with those in Denmark, and it is clear that very different conceptions still exist. Even so, we can understand Heidenheimer's suggestion-it is not completely unintelligible 88 Ehrenberg (n 9) 214, 227. 89 Heidenheimer (n 8). or absurd. This illustrates the theoretical possibility that a concept may become decontested. 90 Usually, however, while some authors believe that the meaning of concepts such as 'democracy' or 'rule of law' has been conclusively settled in positive law, or is firmly embedded in an overlapping consensus, others will disagree.

LAW AS A SECOND-ORDER ESSENTIALLY CONTESTED CONCEPT
Few legal theorists would deny that at least some persistent legal debates, for example, on 'justice' or 'fairness' might be fruitfully analysed in terms of essentially contested concepts. However, although there is a broad consensus that some legal concepts can be essentially contested, there is certainly no consensus among legal philosophers on whether the concept of 'law' itself is essentially contested. 91 This is a good example of second-order essential contestedness. I suggest that this idea can help us understand the debate in legal philosophy between positivists and their opponents. Certainly, I don't want to claim that this is the only issue that divides the parties-and sometimes also the groups within one camp. 92 For example, the various traditions are also divided on issues as whether law can be described as separate from morality, and whether law is typically connected to the state or to force, and on various other themes. Moreover, I have claimed elsewhere that there is a much more fundamental difference between the competing traditions in legal philosophy. 93 This more fundamental explanation is that law can be seen as an essentially ambiguous concept-a concept that can be modelled in at least two incompatible ways, as a product or doctrine, and as a practice. 94 I have argued that this epistemological difference is the most fundamental one in many philosophical debates. However, I will not repeat that analysis here. In this article, I cannot do justice to all nuances in the debate between positivists and non-positivists, I just want to draw attention to one aspect of it.
Gallie is unclear about whether 'law' is an essentially contested concept. In the book version, he groups the concepts of 'science' and 'law' together as possible candidates, although he doubts whether they fully fit the framework. 95 Earlier, he shows no doubt about the concept of 'science'; it is essentially contested. 96 Therefore, it is plausible that Gallie regards 'law' as essentially contested as well, as it belongs to the same group of concepts as 'science'. 97 Many legal positivists like Hart claim that 'law' is a normatively neutral concept. 98 Therefore, for those who accept that claim, 'law' cannot be an essentially contested concept. 99 Non-positivists, however, usually argue that we cannot understand law if we do not focus on the master ideal, the purpose, or the leading values of law. 100 Different conceptions of these ideals, purposes, or values are legitimate. Consequently, for those non-positivists, 'law' is an essentially contested concept. In the view of these authors, positivists are merely blind to this unavoidable normative dimension of 'law', but their position would be understood and defended better if they simply acknowledged it. An example of this is Ronald Dworkin's attempt to describe jurisprudential debates in terms of different conceptions of 'legality'. 101 However, chances are slim that Dworkin will be able to convince his positivist opponents that this is the best way to reconstruct the debate. They will probably deny the basic presupposition that law is essentially oriented towards a normative ideal or value. 102 Positivists like Hart regard 'law' as a normatively neutral concept that, therefore, cannot be essentially contested, whereas Dworkin regards it as an essentially contested concept. This is a second-order contest on whether 'law' is an essentially contested concept. In other words, 'law' is a second-order essentially contested concept.
If we regard 'law' as a second-order essentially contested concept, this also has implications for whether it is possible to construe a generally accepted definition of 'law'. Many authors have sought to identify the specific characteristics of law, the properties that distinguish law from other normative orders. They have been 95 Gallie, Philosophy and the Historical Understanding (n1) 190. 96 Ibid 156. 97 For a different interpretation, see Ehrenberg (n 9) 220 who argues that Gallie himself was in doubt about the concept of law. However, from the text we can only infer that 'law' and 'science' do not meet the criteria fully, but that they are in most relevant respects similar to the concepts that are explicitly analysed as being essentially contested. As I discussed earlier (and as Ehrenberg himself accepts as well), the fact that a concept does not fully meet all criteria need not be a reason for not calling it essentially contested, because Gallie presents the seven conditions as ideal typical rather than as necessary and sufficient conditions. A similar mistake is made by Leslie Green (n 91) 18, when he incorrectly deduces from Gallie's acknowledgment that 'law' is not in all respects similar to the clear examples of essentially contested concepts that he 'explicitly excluded' the concept of law. This is an extreme interpretation that is clearly at odds with the fact that Gallie called the concept of law a possible candidate. 98 Hart (n 25) 240. 99 See Ehrenberg (n 9) 223; Green (n 91) 19 (both authors arguing that the general concept of law need not always be evaluative). 100 See Selznick (n 23); Fuller (n 23); Dworkin (n 78). 101 See Dworkin (n 78) 171ff. 102 Many positivists have understood law in terms of its point or function, but that does not necessarily require reference to ideals.
looking for the 'distinctively legal'. Some authors even argue that this is the central task of legal philosophy. Indeed, according to Julie Dickson, the core business of analytical jurisprudence is 'to isolate and explain those features which make law into what it is'. 103 However, if we accept that 'law' is a second-order essentially contested concept, this implies that there is a substantial problem in defining the distinctively legal. At the second-order level, it is contested whether law is inherently oriented towards certain values, such as legality or justice. Moreover, among those who hold that law is oriented towards certain values, it is essentially contested which values they are, and how they should be interpreted. Is legality the master ideal of law, as Selznick and Fuller hold, or are there other ideals such as justice, legal security, and purposiveness, as Radbruch holds? 104 If these ideals are the basis for some form of substantive natural law, what is its content? Is it merely a minimum content as held by Hart, or a more substantive set as defended by Finnis? 105 Of course, this is not a conclusive argument that we cannot find other characteristics that are distinctively legal. After all, there could be certain minimal descriptive elements like the suggestion of a combination of primary and secondary rules (Hart) or of institutionalisation (Raz). I will not go into that debate here. Suffice it to say that there is also broad disagreement on these other suggested characteristics of law. 106 Even if it were possible among positivists to find a broad consensus on certain characteristics, however, the result would only be a list of very minimal necessary conditions. Together, they could never constitute the necessary and sufficient conditions to characterise the distinctively legal. The reason is simple; for many non-positivists, these characteristics cannot be sufficient conditions, because in a complete list of them the orientation to specific legal values or ideals should have a central place. In other words, one reason-but again: one reason among many others-that positivists and non-positivists cannot agree is that 'law' is a second-order essentially contested concept.

CONCLUSION
In this paper, I have presented a critical reconstruction of the notion of essentially contested and essentially contestable concepts. As a first step, I have argued that Gallie's seven conditions for essential contestedness should be divided into four semantic criteria, which are central to our understanding of the notion of essential contestability, and three pragmatic criteria, which mostly should be disregarded as defining criteria. Moreover, we should make a distinction between contested and contestable-essentially contested concepts are essentially contestable concepts that are actually contested. As a second step, I have argued that we should understand essentially contestable concepts in terms of ideal-orientation. Essentially contestable concepts may be defined as concepts that refer to ideals or to concepts and phenomena that can only be fully understood in light of ideals and that are, as a consequence, open to pervasive contestation. Understanding contestability in terms of ideals provides a reconstruction that can be illuminating for our analysis of pervasive conceptual disagreements.
However, we should accept that it might be a matter of endless dispute as to whether a concept is essentially contestable. Because the thesis that a phenomenon can only be understood in terms of an inherently ideal-oriented character is often strongly contested, the related concept may be called second-order essentially contested. Moreover, both essential contestability and essential contestedness may be a matter of degree.
I have illustrated this possibility of second-order essential contestedness with the concept of law. Although legal ideals such as the rule of law are quite broadly accepted as being essentially contested, this is different for the concept of law. Some authors argue that it is ideal-oriented, whereas others claim it is not. This insight that 'law' is a second-order essentially contested concept may explain some of the disagreements between the school of legal positivism and its various opponents. Moreover, it may explain why the search for the 'distinctively legal' is often futile.

FUNDING
The research for this paper was partly funded by the Netherlands Organisation for Scientific Research NWO.