Why Legal Formalism is Not a Stupid Thing

Legal formalism is the foil for many theories of law. Yet formalism remains controversial, meaning that its critics focus on claims that are not central. This paper sets out a view of formalism using a methodology that embraces one of formalism&#8217;s most distinct claims, that formalism is a scientific theory of law. This naturalistic view of formalism helps to distinguish two distinct types of formalism, &#8220;doctrinal formalism,&#8221; the view that judicial behaviour can be represented using rules, and &#8220;rule formalism,&#8221; the view that judges follow external rules when they are deciding cases. Doctrinal formalism, understood in naturalistic terms, overcomes many of the criticisms that have been levelled at formalism and can also be used to rehabilitate the currently out&#8208;of&#8208;favour &#8220;declaratory theory of law.&#8221; Doctrinal formalism is also a longstanding view of law, reflecting both what the original formalists thought of law, and what many present&#8208;day doctrinal lawyers seem to believe. The naturalistic methodology is used to show that the main dispute between doctrinal formalism and American legal realism can be explained by a difference of assumptions concerning whether the values of judges are relative to society, or relative to other judges.


Introduction
Legal formalism has been called the thesis, to which modern American legal thought is the antithesis (Grey 1983, 3). Yet exactly what formalists believe is unclear. There are many different conceptions of formalism that share very little other than a bad reputation (Simpson 1989, 834;Stone 2004, 167;Weinrib 1988, 950;2013, 22). This paper seeks to identify the central claims of formalism by using a naturalistic methodology. The focus is primarily on the original meaning of formalism, because it is a historical school of thought that deserves recognition on its own merit and because it is difficult to understand the schools of thought that succeed it and the schools that reacted to it without such an understanding. The primary justification for the naturalistic methodology is to develop the original formalists' view that they were practising science.

Doctrinal Formalism (from Behaviour to Rules)
The predominant variant of formalism is the view that the behaviour of judges can be represented using rules. We can call it the predominant view because it most closely reflects the views of the original formalists and it is the most coherent formalist view of adjudication. We will call it "doctrinal formalism" due in part to its focus on doctrine, and in part to the close similarities with the methodology used by present-day doctrinal lawyers.
Characteristic of formalism is the focus not on the behaviour of individual judges, but on that of all judges within a legal system. There is assumed to be a core psychology shared by all judges which they use to decide cases. The formalists called this core psychology the "doctrine" (Beale 1916, 142). The doctrine is what judges purport to represent (accurately or inaccurately) in the form of so-called "black letter" case law. Because it is relatively rare to draw a distinction between behaviour and representations of that behaviour, the term doctrine has often been used indiscriminately to refer to both. However, in this paper we limit the meaning of doctrine to the actual behaviour of judges, not how their behaviour will be represented (though accurate black-letter representations will of course correspond to the doctrine).
Formalists assume that the doctrine is complete in that there is always a right answer to the question of how a judge will respond to a particular factual scenario (even if it is to do nothing). Beale approves of the view Judge Grosscup expressed in Swift v. Philadelphia & Reading Railroad (1894): 1 "No plain or valley, no nook or corner, to which the dominion of man has extended itself, is without some law of the land. Indeed, law is the breath of dominion" (Beale 1916, 154; see also Grey 1983, 7-8). By contrast, the black-letter representations that reflect the doctrine are likely to be incomplete to some degree, even in a mature legal system, because they are made in the light of the doctrine (Beever 2011, 225-6).
Unlike natural lawyers, formalists accept that the doctrine changes, both in time and by place (Beale 1916, 145 n. 1;Grey 1983, 28-9;Langdell 1879, viii;Wambaugh 1894, 93). The doctrine might change where there is a change in social or economic conditions (Beale 1916, 142;Samuels 1975, 295;Grey 1983, 28-9; but see Beever 2011, 226). Doctrine also varies between different legal systems. Thus Beale (1916, 137) writes that "the particular common law of New York and the particular common law of Tennessee must necessarily be different, unless the two states form a single legal unit, which we know not to be the case." It is because of this fluidity of the doctrine that some formalists worry about the ossifying effect on the law of systemising representations (Samuels 1975, 304-5).
Formalists see their work as stemming from the fallibility of judges. Judges are thought to be fallible at discerning the doctrine (i.e., getting the wrong result) or fallible at articulating the doctrine (i.e., getting the right result, but articulating the wrong reasons). Thus Beale (1916, 145) points out that positive law may depart from the doctrine because the tribunal, being human, might err (see also Beever 2011, 225-6, 228-9;Goff 1999, 313-4;Weinrib 1988, 962-3;2013, 13). Likewise, Wambaugh (1894, 46-7, 50, 57-60) writes that judgments might be unreliable due to lack of care, overlooked authority, or bias, or if they are badly argued. A clear recognition of the fallibility of judges is shown by the 1966 House of Lords Practice Statement (House of Lords 1966Lords , 1966. Up until that point, the House of Lords had followed the principle of stare decisis, meaning that the court would always follow its previous precedent. This proved problematic because there was no higher court in the judicial hierarchy to overrule House of Lords' precedent when it proved problematic or unworkable. This lacuna led to the House of Lords giving notice through the practice statement that it might depart from its own precedents, implicitly recognising its own fallibility. Formalists therefore treat cases of "first impression" (where no previous precedent exists) with caution until more courts are able to pronounce on the issue (Beale 1916, 147;Goff 1999, 314;Wambaugh 1894, 56;Weinrib 2013, 13).
It follows that because judges do not always follow the doctrine, there is scope for a normative theory of formalism, to the effect that judges should follow the doctrine. Nonetheless, formalism viewed as a naturalistic theory is necessarily a description of how judges do decide cases, not how they should decide cases. This understanding of law as a corpus autonomous from normative considerations like politics or morality has had a corresponding appeal for the judiciary as it allows them to present their decisions as normatively uncontentious (Posner 1987, 762, 772-73;2008, 42;Simon 1998, 16;Weinrib 2013, 6, 46).
Further scope for formalism arises because there are better and worse ways of representing the doctrine. Even where a judge represents the doctrine accurately, formalists may seek better representations. To this end, sometimes a distinction is drawn between rules and principles (Virgo 2008). Cases may state rules that explain the decision in individual cases, but principles may be capable of providing a unified explanation for multiple cases, or even for larger areas of law (Beever 2011, 227;Goff 1999, 320;Grey 1983, 8-9;Kelman 2011, 206). To take an extreme example, one disciplinary successor of formalism, the variant of Law and Economics promoted at times by Posner and Landes, has been even more ambitious, seeking to explain all doctrine in terms of the single master principle of maximising the total wealth in society (Coase 1960, 22;Posner 1983, 4-5;1993, 40;Rubin 2000, 544;Veljanovski 2006, 40).
Perhaps the key aspect of formalism is a belief that the methodology is scientific (Beale 1916, 135;Kronman 1993, 171;Van Hoecke 2013, vii;Wambaugh 1894, 24-5). Beale, for instance, views it as fruitless to form a priori definitions, thinking that theorising is better when informed by factual observation (Samuels 1975, 282). Though this view of formalism as natural science has been criticised (Posner 1987, 762;Grey 1983, 21-2), the methodology does bear many similarities to other areas of science that rely upon human intuition as data. One such parallel is Euclidian geometry, which involves the formalisation of human intuitions about the physical world (Grey 1983, 16-9;Weinrib 1988, 962). An even better analogy is grammar (Grey 1983, 27-8 n. 98). Linguists may be able to articulate the rules of grammar that speakers use but struggle to formalise (Strawson 1992, 5). Grammar is a particularly apt analogy because language changes over time, just as formalists think doctrine changes.
Similarities between formalism and psychology exists because the outcomes of cases and the reasons judges give for their judgments are undeniably empirical data that evidence judicial behaviour. The outcomes of cases are more reliable (as the cases must be the actual outcome of the process) but are less informative (as a variety of cognitive processes are compatible with the outcome) whereas reasons are less reliable (because the judge may misstate the doctrine), but potentially more illuminating.
Formalists infer the doctrine from this empirical evidence. Though rarely acknowledged in the legal context, intuitive inferences pervade both day-to-day human cognition and scientific research (Mercier and Sperber 2011, 58-9;Sperber and Mercier 2012, 372). For example, if there is a knock on my door at eight o'clock in the morning, I might infer that it is the postman. The evidence I have is only the knock at the door. However, I also have tacit information that the postman normally visits around eight o'clock and that it is rare for anybody else to visit this early. This tacit information has been called a warrant or a generalisation (Anderson, Schum, and Twining 2005, 62;Toulmin 1958, 91;Walton 2005, 15). Scientific inferences are similar. Scientists make inferences on the basis of imperfect and incomplete information to novel and untested scenarios. Nobody has yet arrived at a mechanical process for discovery in science, and it is generally recognised that scientific discovery does not result from deduction alone (Carnap 1995, 5;Okasha 2002, 138-9). As a result, everyday inferences and scientific inferences are always probabilistic and conditional (Carnap 1995, 9). The formalist methodology is the same. Formalists infer general principles from instances that are limited (not every conceivable factual scenario is represented among the cases) and unreliable (judges may misidentify the doctrine, or fail to articulate it accurately). Thus the formalist methodology cannot be criticised for relying on probabilistic inference rather than pure deduction (Moore and Sussmann 1932, 572).
Formalists are also empiricists, but they tend not to conduct experiments themselves. Instead they rely on the outcomes of cases as "natural experiments" (Langdell 1887, 124;Samuels 1975, 282). Langdell (1887, 124) believes, for example, that a law library is a sufficient resource for legal scientists, and Wambaugh (1894, 67-8) writes that law students cannot conduct experiments. It is not clear why formalists reject experimentation. It may be that they think that judges in hypothetical scenarios might not behave as they do in real life. Present-day psychology recognises that differences between real life tasks and experimental tasks may undermine the external validity of hypothetical experiments, but these tend to be taken to affect the weight of the evidence, not to undermine it completely (Loewenstein 1999, F26). As such, it is common for researchers to conduct hypothetical research with real life judges (e.g., Aspinwall, Brown, and Tabery 2012;Dhami and Ayton 2001).
Instead of making observations themselves, formalists rely on the facts as recorded by the judge (Bix 2009, 192;Kelman 2011, 208;Langdell 1887, 124). Contemporary psychology distinguishes (somewhat confusingly in a legal context) between judgement and decision-making. Judgement (note the additional e) is what lawyers call fact-finding, whereas decision-making is the response to particular facts and corresponds to doctrine. Formalists ignore fact-finding, treating it as uncontroversial (Grey 1983, 11-2;Simon 2004, 514). Mainstream psychologists may also rely on subjects' self-reports of the facts that were relevant to their behaviour, but there is some risk that psychological subjects and practising judges misidentify the relevant facts, either mistakenly or deliberately. As we shall see, this might jeopardise the scientific quest.

The Declaratory Theory of Law
Doctrinal formalism is closely associated with the declaratory theory of law, the belief that when judges give reasons for their judgments, they are merely articulating the pre-existing doctrine (Blackstone 1827, 69;Hayek 1982, 77-8;Wambaugh 1894, 75;Weinrib 1988, 956, 10; Willis v. Baddeley). 2 Given the tenets of doctrinal formalism set out above, the reasons for this adherence to the declaratory theory are unsurprising (Samuels 1975, 289; but see Wambaugh 1894, 80).
The declaratory theory has fallen out of favour nowadays, and it is treated as something of a mystical belief (Bingham 2000, 26;Goff 1999, 314;Reid 1972, 22;Twining and Miers 2010, 293-4). Beever (2013, 422) points out that it is routinely dismissed as an absurdity, and Lord Reid (1972, 22) claims that "we do not believe in fairy tales any more." But rejecting the declaratory theory of law creates its own problems. Most notably it entails accepting the positivist view that judges make law. Law-making by judges is troublesome because it is necessarily made in the context of contentious litigation, after the dispute has arisen. As a result, it is retrospective, and seemingly unfair to the loser in the litigation who could not have altered his behaviour to avoid liability (Weinrib 1988, 999). Some such as Bentham (1823, 13) accept that judges legislate, but strongly object to it because of this retrospective implication. Others profess that judges make law, but seem to see this retrospectivity as unproblematic (Kleinwort Benson Ltd v. Lincoln City Council;3 Regina v. Governor of Brockhill Prison;4 Twining and Miers 2010, 294).
A further problem with the belief that judges make law is how the general public get to know this law. Judicial decisions are so numerous, it is difficult even for specialist practitioners to keep up to date with the state of the law in their area of expertise, even with the benefit of specialist services that systemise and update the authorities for them. The general public rarely have the time or facilities to stay abreast of caselaw (Ellickson 1986, 628), save in the case of disputes that can be predicted well in advance (Hertwig 2006, 394). This has led Hertwig (2006, 394) to point out that the route from black-letter law to behaviour is "bewilderingly difficult." Yet another issue with the view that judges create law is what "creation" really means. Positivists who object to the declaratory theory of law reject the view that judges are merely articulating the pre-existing doctrinal principles. Instead they profess that law is "created" at the point it is articulated by the judge. If creation is taken to mean a biblical genesis from a void, then this (from a naturalistic perspective at least) would be even more mystical than the characterisation of formalism. Such a view sits ill with a scientific perspective because science presumes that every event has a causal history, for instance the behaviour of neurons in a judge's brain. Thus the view that judges legislate must mean something less extreme.
A better account of what creation means in this context may lie in the difference between the formalist view of law that judges are articulating a single doctrine and other naturalistic theories of law that recognise variation between judges, meaning that different judges articulate different doctrines. A formalist believes in a single doctrine, meaning that it is easier to predict how a judge will decide. Formalists therefore believe it possible to predict the outcome of a case with the assistance of a good lawyer (or a good formalist). Thus Beale writes: "In order that law may help rather than hinder the carrying on of the work of society it must be possible for every person, of his own knowledge or by the help of others' knowledge, to discover the application of the law to any contemplated act" (Beale 1916, 155; see also Weinrib 1988, 962;2013, 9). By contrast, those who reject formalism might argue that in cases of first impression, it is difficult to predict the outcome because there is variation between judges. However, once the doctrine is articulated, other judges will tend to follow it, at least because of the principle of stare decisis, meaning that the uncertainty is reduced. Once this uncertainty is reduced, consequences follow such as that people can plan their affairs with more certainty.
Thus from a naturalistic perspective, neither formalist nor positivist views are mystical. Formalists believe that judges articulate a common doctrine, meaning that from the perspective of other members of society (advised by lawyers), there is limited uncertainty about how a judge will decide a case. Positivists believe that judges articulate different doctrines, meaning that there is considerable uncertainty before a judge determines a case, but this is reduced once a judge gives his decision and reasons.

Representational Formalism (from Rules to Behaviour)
A subtly different, but popular, understanding of formalism is that judges reach decisions by following explicit black-letter rules, such as those articulated by judges in previous analogous cases (Weinrib 1988(Weinrib , 1001. We can call this "rule formalism." For example, Moore claims that formalism asserts that there is a unique result deducible "from the rules and the facts" (Moore 1981, 159; emphasis added). Posner explains it as a belief that judicial decisions are a product of a syllogism where the black-letter law provides the major premise and the facts the minor premise (Posner 2008, 41; see also Scalia 1998, 25). Schauer (1988, 510) believes the heart of formalism is the concept of decision-making according to rules. Leiter (1997, 277-8) summarises formalism as meaning that judges are "rule responsive" rather than "fact responsive." To show why rule formalism is so different to doctrinal formalism, it helps to distinguish between behaviour that is determined by following explicit rules, and behaviour that can be described using rules (Pylyshyn 1990;Sloman 1996, 5;Smith, Langston, and Nisbett 1992, 2-3;Strawson 1992, 5). For example, a computer follows rules. The "rules" in the software are "read" by the hardware to determine the behaviour of the system. Thus a judge applying a statute (i.e., externally represented rules) is following a rule in this sense. By contrast, a ball falling to the ground is an example of behaviour that can be described by rules such as those of gravity. But the ball itself does not follow rules to determine its behaviour (Pylyshyn 1990, 6). The two understandings of formalism correspond to this distinction. Rule formalism assumes that judges follow explicit (black-letter) rules, whereas doctrinal formalism does not. Doctrinal formalism thinks it possible to describe (and predict) the behaviour of judges using rules, without being committed to the idea that judges follow explicit rules (save in the context of stare decisis, a point that will be considered further below). 5 While formalists are interested in representations, they do not see these representations as necessary to determine or limit the behaviour of judges. Langdell sees representation in the form of rules as primarily useful to students of the law. In an address to the Harvard Law School, he explains his method as "of material service to all who desire to study that branch of law systematically and in its original sources," and "[t]he vast majority [of cases] are useless, and worse than useless, for any purpose of systematic study" (Langdell 1879, ix and viii;emphasis added). His focus is squarely on "teaching and learning law" rather than describing the internal process by which judges actually determine cases (Langdell 1887, 124).
The assumption of rule formalism that judges follow rules to reach their decisions is also problematically circular. In the first place, at an early stage of the development of a common-law system, or where "cases of first impression" arise in a mature common-law system, no black-letter rules exist. In those circumstances, judges cannot make their decisions by following rules. The rules are obviously incomplete, which does not sit well with the formalist view that everything is either permitted or forbidden by law (Beale 1916, 154). For doctrinal formalists, it is the doctrine that is complete, not the rules. The rules only become more complete with more judgments and reliable accounts of these judgments.
Even where precedent exists, doctrinal formalists do not believe that the subsequent court reaches the same outcome simply because the later judge follows the earlier court's precedent. Rather, both courts converge on the same rule because they are both following the same doctrine. The two precedents flow from the same source doctrine, not one from one precedent to the other.
The view that judges decide cases by applying rules also sits uncomfortably with the formalist view of the changing nature of the doctrine. Black-letter representations cannot change, so if the rules cannot change, they will not always reflect the doctrine (Samuels 1975, 304-5).
There is one sense in which doctrinal formalists believe that judges follow black-letter rules, but this is very much an exception to the general principle. This exception is the principle of stare decisis, where judges follow black-letter rules for the sake of consistency and certainty. Stare decisis recognises both the fallibility of judges and the importance of certainty in particular areas such as property law (Baker 2002, 200;Wambaugh 1894, 52, 104). Given the acknowledged fallibility of judges, it is foreseeable that judges erroneously state the doctrine from time to time (Beale 1916, 145;Weinrib 1988, 962-3). It is also likely that people arrange their affairs in reliance on these erroneous pronouncements. Without some means of accommodating these two factors, their conjunction could lead to considerable unfairness. Thus formalists recognise that there are circumstances where judges might abide by the principle of stare decisis, where important matters were at stake and predictability was important. Yet stare decisis is very much an exception to the general principles of formalism.
Admittedly, some formalists recognise that once the rules are articulated, this makes the judge's task easier in that the rules can then be followed (Hirsch 2003(Hirsch , 1345Wambaugh 1894, 73). Nonetheless, it is unlikely that formalists believe that judges mechanically apply the rules once they have been discovered. Cases are rarely identical to a previous case, meaning that a judge needs to decide whether to apply the rule in the previous case or apply a different one. Even where judges feel bound to follow the rules for their own sake (as where stare decisis applies), they retain the ability to identify when the rules fail to reflect the doctrine, either because the rules are wrong in the first place, or because the doctrine has changed. In these cases, they can decline to follow the rule, or they can grant permission to appeal so that the precedent can be reviewed by a higher court.

Doctrinal Law
Few present-day theorists explicitly identify themselves as formalists, but it seems that the concept of doctrinal formalism outlined above bears many similarities to the methodology of doctrinal law, the predominant methodology in common-law academia, at least in English law schools. It has historically been relatively rare that doctrinal lawyers spell out their methodology, but due to the increasing pressure on academia to justify their funding, there have been some attempts to expound doctrinal law's methodology. These expositions seem uncannily like those of doctrinal formalism.
To begin with, many doctrinal lawyers see their work as science. Virgo, for example, writes that "the methodology of legal research does have much in common with scientific research. Indeed, the study of law is sometimes described as legal science" (Virgo 2008, 339). Van Hoecke (2013, vii) also includes legal research within the meaning of scientific research. Echoing these views, the Council of Australian Law Deans' Statement on the Nature of Research states that "it is the doctrinal aspect of law that [...] provides an often under-recognised parallel to 'discovery' in the physical sciences" (CALD 2005, 3).
Virgo also shares the characteristic belief of formalists that lawyers cannot conduct experiments, giving the justification that "scientists tend to research by experiment and investigation to explain and predict how the natural world works. But law is artificially constructed and cannot be researched in the same way." As such, he adopts a characteristically Langdellian view (updated for the modern age) that "[t]he raw materials are cases and statutes rather than chemicals and elements and there are no laboratories, just libraries and the internet" (Virgo 2008, 339).
According to Virgo, the goal of a doctrinal lawyer is "to identify the principles underpinning a mass of rules to explain the function of those rules. This is done through the recognition of maxims of the law which are generalizations and distillations of complex bodies of law and are known as principles" (ibid., 340). He also compares the process of distilling the rules to that of mathematicians identifying general principles to explain phenomena (ibid.). These views very much echo the analogy with Euclidian geometry discussed above.
Finally, Minow, occupying the same position as Langdell of dean of the Harvard Law School more than a century later, provides a "field guide" to types of legal scholarship. Top of her list are "doctrinal restatement" and "recasting." Restatement includes organising and reorganising caselaw into coherent elements, categories, and concepts. Recasting includes gathering more than one line of cases across doctrinal fields or categories and showing why they belong together or exposing unjustified discrepancies (Minow 2013).
Though many modern-day doctrinal lawyers rarely identify themselves as formalist, the notable exception is Weinrib (2013, 22-55). Over a series of articles and books, Weinrib has spelled out the philosophical basis of his formalism, which he and others have then applied to diverse areas of private law (Weinrib 1988;Beever 2011). Weinrib has developed formalism in his own direction, but there are many similarities with doctrinal formalism. He approves, for instance, of Unger's depiction of formalism as a belief that the black-letter law imperfectly reflects an intelligible moral order (Unger 1983, 565;Weinrib 1988, 953) and considers the underlying doctrine to be coherent and unitary (Weinrib 1988, 971-2).

The Realist Critique
The antithesis to formalism is generally considered to be realism (Grey 1983, 3;Leiter 1999, 276;2010;Posner 1986, 185), and realism provides a helpful foil to the views of formalism. Realism offers arguments against both doctrinal formalism and rule formalism, but we will look at the reasons realism doubted doctrinal formalism.
Understanding the divisions between formalism and realism is more straightforward if we return to the distinction between fact-finding and decision-making. 6 As we have seen, the formalists treated fact-finding as uncontroversial, which on one level it is. In a court case, a fact finder needs to make inferences about what happened from the evidence. The evidence is invariably incomplete, and necessarily controversial. The parties may disagree over the evidence, and what inferences can be drawn from the evidence. Yet science assumes that there is a single right answer, even if it is difficult or impossible to be certain what that right answer is in practice.
At the first, fact-finding, stage, both formalists and realists necessarily assume that the judge is aiming to make the correct inference to the facts. This is because any decision-making is dependent on accurate factual inference. If the inference to the facts is wrong, the chances are that a decision that depends on those facts will not further the judge's aims (whatever they might be).
While judges necessarily aim to be accurate in their factual inferences, there may still be some diversity in their fact-finding. The generalisations that judges use to infer what happened may be based in part on their life experiences. Different judges with different experiences may rely on slightly different generalisations. For example, Pennington and Hastie found that decision-makers who had to infer whether carrying a knife to a bar was evidence of premeditated killing varied according to their background: Those from poorer backgrounds thought that it was normal to carry a knife (and thus not evidence of premeditation) whereas those from more wealthy backgrounds found it abnormal (and thus consistent with premeditation) (Pennington and Hastie 1991, 525, 556; see also Schum and Martin 1982, 124;Simon, Snow, and Read 2004, 817). 7 Though we can reasonably assume that judges aim for accuracy in fact-finding, we should take care not to assume that the facts that the judges report will be veridical. This is a key distinction between formalism and realism. To see why realists doubt that judges' reports of their fact-finding are not always veridical, we need to consider the stage that follows fact-finding, that of decision-making.
Formalists and realists have very different views about decision-making. Decision-making embodies values. Once facts are inferred, a decision has to be taken how to respond to those facts. In the most general sense, this covers whether an individual is found guilty, or whether a party is found liable. But it also covers the particular offence or basis of liability, and the consequences that flow from the finding, such as the nature of the penalty or damages. As we have seen, formalists believed that values were relative to the society at the time, which meant that all judges' decisions would embody the same values. By contrast, realists are more sceptical, thinking that judges' decisions sometimes embody individual values (Bix 2009, 193;Cohen 1935, 845-6;Leiter 1997, 268;Llewellyn 1930, 444;Moore 1923, 613;Moore and Sussmann 1932, 576).
Another way of highlighting these differences between formalists and realists is by considering where they lie on a spectrum of metaethical views about judging. At one extreme are natural lawyers, who are objectivists about values, believing that values are always consistent through time and regardless of society. Next on the spectrum come formalists, who are metaethical relativists in that they believe that values are relative to the time and society. They are followed by what might be called "moderate" realists, who agree with formalists some of the time, but also believe that there are some circumstances where judges' decisions manifest their individual values (Baum 2006, 8-9). Finally, an "extreme" realist might believe that judges' values always manifest their individual values all of the time and would thus fall at the opposite end of the spectrum (Leiter 1997, 279).
It is this single metaethical difference between formalism and realism that seems to cause the two camps to hold differing views about the veridicality of reasons. We noted above that there is no reason for a formalist or a realist to differ on their assumption that judges aim to accurately infer facts, but there may be reasons why judges would represent their findings inaccurately if their values differ from other judges. Given that formalists believe that judges have common values, there is no reason for a formalist to think that judges would misrepresent their values. 7 In the context of fact-finding in a case, we should recognise that there are two different things that we may be seeking to identify. We may be seeking to identify what really happened given the evidence, or we may be seeking to identify what adjudicators infer happened given the evidence. Formalists treated these two as one and the same, but it is likely that there are slight divergences between the two. Adjudicators' inferences may diverge from the real world to the challenges of finding out what really happened when we only have limited and unreliable information. Adjudicators inferences' may turn out to be consistently imperfect. It is helpful to return to the analogy of Euclidian geometry. We know that the intuitions of people that are the axioms of Euclidian geometry are consistent. Nonetheless, we also know that these intuitions fail to approximate how we believe the physical universe to be in certain exceptional circumstances. Thus it is a possibility that adjudicators' fact-finding is similarly imperfect.
By contrast, realists think that there are differences between judges. Holmes, for example, pointed out that "[b]ehind the logical form lies a judgment as to the relative worth and importance of competing legislative grounds" (Holmes 1897, 466). Consequently, realists were open to the idea that the values embodied in decisions might be controversial: "It is the great disservice of the classical conception of law that it hides from judicial eyes the ethical character of every judicial question, and thus serves to perpetuate class prejudices and uncritical moral assumptions which could not survive the sunlight of free ethical controversy" (Cohen 1935, 840). If judges have different values, then this gives them a motive to misrepresent their reasons. In the simplest case, this would involve misrepresenting their factual inferences so that the judge can purport to be applying values that would be acceptable to other members of society (and would thus not be susceptible to being overturned). There are a variety of means by which a judge might misrepresent his reasoning to achieve the outcome he wanted. Judicial reasons are supposed to reflect the inference from facts to the outcome. Because the reasons for the outcome have to be acceptable to others, and the outcome cannot be misrepresented, the area most vulnerable to misrepresentation is the facts (Llewellyn 1930, 444;Pound 1910, 15;Simon 1998, 85-6, 92). A realist believes that misrepresenting the facts, such as by leaving out relevant facts, misrepresenting facts, or adding in facts, allows a judge to arrive at the outcome he prefers in a way that is less vulnerable to being overturned. Thus Oliphant wrote of the formalist preoccupation with black-letter law as the "study [of] dead bodies from which the life we would know has departed" (Oliphant 1926, 224).
All the reasons for preferring different forms of value relativism are beyond the scope of this article, but it is illustrative to focus on some of the strongest arguments for each side here.
Perhaps the strongest argument for the individual relativism favoured by realists is empirical. Looking objectively at the behaviour of judges and individuals in wider society, there seems to be considerable diversity of values. An example from the formalist heyday that illustrates the different stance towards decision-making between formalists and realists was the postal acceptance rule. This was a novel challenge to the law of contract caused by advances in communication. For business certainty it was imperative that a consistent practice be adopted, but there did not seem to be overwhelming reasons for preferring one solution over the other. Different courts had taken different approaches to the rule, some holding that an acceptance was binding when it was posted (e.g., England and New York), others (e.g., Massachusetts) holding that it was binding when received (Grey 1983, 3-4). There were therefore precedents both ways. The arch-formalist Langdell thought that there was only one right answer, which was that an acceptance was binding when received. Inferring this rule from the equivocal precedent necessitated the rejection of a colossal weight of precedent. In response to the arguments relying on what was just, Langdell (1880, 20-1) famously declared these to be "irrelevant." Whereas a formalist would assume that there was a single right answer, realists were more pragmatic, thinking instead that there were two imperfect solutions and no overwhelming reason to choose one over the other. From this empirical perspective, the arguments seem to favour realism.
By contrast, from a philosophical perspective, the formalist's metaethical view that values are relative to society seems more satisfying than the realist alternative. This is because to debate conflicts of values, societal relativism is the minimum that must be assumed. If one assumes that values are relative to the individual, then this discourse has no traction because it seems there is no means of resolving differences. To the extent that realism is treated as a naturalistic account of how judges actually behave, this is not, in itself, fatal. But when a realist discusses conflicting values (as realists want to), they must assume societal relativism as a minimum. Thus realists are forced to take different metaethical positions depending on the task they are engaged in. To many, this "facing both ways" is dissatisfying.

Conclusion
A naturalistic view of formalism shows that what we have called doctrinal formalism is a respectable theory of law rather than a jurisprudential joke. Central to doctrinal formalism is the belief that doctrine exists as a value framework common to all judges within a legal system that they draw upon to adjudicate. The black-letter case-law is an imperfect representation of the doctrine which can be used to infer the doctrine. Doctrinal formalism and the very much out-of-fashion declaratory theory of law explain the relationship between the doctrine and the black-letter law in a way that avoids many of the contradictions that trouble other explanations. Many of the criticisms of formalism (such as the complaint that formalists do not rely on pure deduction to reach conclusions) are unproblematic from a naturalistic perspective.
By contrast, the more widespread present-day conception of formalism is what we have been calling rule formalism, the belief that adjudication is the product of the application of black-letter law to facts. Rule formalism suffers from fundamental shortcomings, which may explain why formalism is held in such low regard. The confusion between the two is unfortunate because it has led to the neglect of formalism and has made it more difficult to understand doctrinal formalism's progeny.
Faced with this confusion, a naturalistic analysis is both true to the aspirations of the original formalists and brings much-needed clarity. Stating the assumptions of formalism clearly allows constructive discussion to focus on the real points of disputes between formalism and other theories. A naturalistic approach can be used to distil the dispute between doctrinal formalism and realism to dispute about a single (though large and contentious) issue. The resolution of the metaethical debate about the relativism of judicial values is a challenging one, but a naturalistic analysis makes the problem more transparent.

University College London Gower St WC1E 6BT London
United Kingdom E-mail: paul.troop@ucl.ac.uk