1.1 The allure of counterfactuality and taxonomy

In his first essay, Fuller distinguishes the legal fiction, as a type of proposition, from other types of non-truthful propositions. More generally, he defines the legal fiction as a particular type of language usage. His second essay moves from definition to an exploration of the reasons fictions exist; he proceeds largely by examining the motivations for the use of fictions, and this examination becomes a taxonomy of types of legal fiction. Subsequent theorists have been drawn more to the first than to the second of Fuller's definitions (conceiving the fiction as a type of proposition), and they have focused more on his taxonomy than on his effort to explain the phenomenon. (As discussed further in Section 1.2, theorists have largely remained silent about Fuller's third essay.)

Fuller's differentiation of legal fictions from other types of statement, including ‘truthful statement[s], … lie[s], … [and] erroneous conclusion[s]’, is perhaps the best-known aspect of all of his work on the topic (Fuller, Reference Fuller1930a, p. 366). Positively, Fuller defines a legal fiction as ‘either, (1) a statement propounded with a complete or partial consciousness of its falsity, or (2) a false statement recognized as having utility’ (p. 369). This definition of legal fictions as, centrally, counterfactual propositions has been the nearly universal starting point for later work on legal fictions, even when that work proposes refinements on Fuller's definition.Footnote ¹

Most of those who accept Fuller's identification of legal fictions as ‘statements’ also accept his underlying assumption about the necessarily linguistic nature of legal fictions (usually, however, without discussing this premise in any detail):Footnote ²

Fuller's first essay is full of other references to legal fictions' status as ‘linguistic phenomen[a]’ (see, e.g., pp. 373–75, 377–79, 384–86, 389). In this he follows (without citing or discussing) Bentham, who similarly regarded the legal fiction as a special form of the linguistic fiction, itself a special form of the linguistic abstraction (Ogden, Reference Ogden1932).Footnote ³ But unlike Bentham, Fuller emphasises the constructive function of legal fictions; according to Fuller, they exist at least partly because of ‘the impulse to keep the form of the law persuasive’ (Fuller, Reference Fuller1930a, p. 387).Footnote ⁴ Thus, according to Fuller, fictions as counterfactual propositions are rhetorical devices for persuasion or self-persuasion.

Fuller's second essay more systematically explores the motivations for using legal fictions (Fuller, Reference Fuller1930b, pp. 517–24, 537–38 & n. 147, 539; see also Fuller, Reference Fuller1930c, pp. 887, 903). As Fuller enumerates these motivations, his explanation becomes a typology: he identifies ‘exploratory’ (Fuller, Reference Fuller1930b, pp. 527–28), ‘emotive’ (p. 517), ‘expository’ (p. 518), ‘descriptive’ (p. 537), ‘historical’ (p. 538 n. 147) and ‘apologetic or merciful’ (p. 539) fictions, among others. The classifying focus of this essay has also become a model for the work of later theorists.Footnote ⁵ Virtually every student of the legal fiction since Fuller has emphasised either Fuller's definitional formulations, his enumeration of types of fiction, or both.

The variations among later theoretical discussions of legal fictions owe more to differences in their authors' disciplinary affiliation and shifts in academic fashion than to any fundamental reconsideration of Fuller's study. Work in the 1980s and 1990s tended towards the interdisciplinary, while more recent American legal scholarship has become increasingly doctrinally focused and critical of the legal use of fictions.Footnote ⁶ A good example of scholarship during the earlier period is Kenneth Campbell's philosophically informed 1983 examination of Fuller's essays (Campbell, Reference Campbell1983). Campbell acknowledges that Fuller saw legal fictions as intimately related to the legal recognition of ‘facts’ – also a matter of legal language usage (p. 359)Footnote ⁷ – but explicitly disavows any desire to examine ‘lawyers’ linguistic practices' (p. 341; see also Del Mar, Reference Del Mar2013). Although he does characterise the legal fiction as a type of language usage, Campbell is concerned only with the philosophical implications of this insight.Footnote ⁸ Campbell describes fictions as ‘means of changing the application of the law by relying on a tension between two classifications of fact’ (Campbell, Reference Campbell1983, p. 339) – one of the functions highlighted by Fuller – but like Fuller, he does not examine in detail how such ‘classifications of fact’ come to exist or to differ.

Many later writers have focused on demonstrating the continued operation of legal fictions by identifying examples not considered by Fuller. Up to the mid 2000s, commentators tended to stress the analogies between legal fiction and ‘fiction’ in a more everyday sense, often taking a humanistic approach. In a 1986 essay, for example, Aviam Soifer disavowed any desire to present a definitive taxonomy (Soifer, Reference Soifer1986, p. 873), but followed a catalogue approach anyway, presenting ‘a few leading fictions, both literary and legal, to suggest and assess a continuum in law and literature’ (p. 873).Footnote ⁹ In a 1990 article, Louise Harmon likewise declared her desire to re-establish the importance of the legal fiction as a topic on the ‘jurisprudential agenda’ (Harmon, Reference Harmon1990, p. 1) through analysis of ‘the history of a particular legal fiction, the doctrine of substituted judgment’ (pp. 16, 22). The author of a 2002 Harvard Law Review Note similarly argued that the ‘debate over the legal fiction should be reawakened’; this author aimed ‘to prevent the more modern conceptions of the legal fiction from being unfairly stigmatized and dismissed as were their earlier counterparts’ (Note, 2002, p. 2228).

Over the past decade, work on legal fictions has become more doctrinal and critical, but it continues to identify Fuller as a starting point and to stress definition and classification. In a 2007 discussion of ‘new legal fictions’, for instance, Peter Smith emphasises the same counterfactuality that Fuller identified: ‘A judge deploys a new legal fiction when he relies in crafting a legal rule on a factual premise that is false or inaccurate’ (Smith, Reference Smith2007, p. 1437). But Smith argues that Fuller's list was incomplete; a taxonomy of ‘new legal fictions’ (pp. 1439–40), Smith contends, is necessary to prevent abuse of these practices of accepted departure from factually accurate statement (pp. 1441).Footnote ¹⁰ In a response to Smith, Nancy Knauer has defended the utility of at least some legal fictions. Emphasising definition rather than taxonomy, she concludes that most of the new legal fictions Smith identifies do not merit the label under Fuller's counterfactual definition: ‘Empirical legal errors, discredited regimes, and complex statutory schemes [Smith's “new legal fictions”] are not classic legal fictions’ (Knauer, Reference Knauer2010, p. 49).Footnote ¹¹

Resolution of the debate over the status of the devices described by Smith is impossible without analysis of the assumptions underlying Fuller's definition of legal fictions as counterfactual propositions. These assumptions include the notion that there is a sense in which propositions can be true about the world, or ‘factual’, as well as counterfactual; they thus also include assumptions about the nature of legal communication, and related assumptions about the extent to which propositions generated within a legal system can be factual in the same sense as non-legal propositions. No post-Fuller theorist of legal fictions has directly considered these issues. Their neglect is consistent with a broader tendency in legal scholarship to avoid inquiring too directly into what makes a legal ‘fact’ and the relationship of such ‘facts’ to other aspects of legal discourse (e.g. Twining, Reference Twining and Twining2006, p. 15; Allen and Pardo, Reference Allen and Pardo2002, p. 1770). The same tendency probably explains commentators' relative inattention to the subject of Fuller's third essay: the roles of fictions in science and law.

1.2 Fuller and Vaihinger: fictions in science and law

Fuller's third essay focuses on Hans Vaihinger's general philosophy of fictions and on the functions of fictions in science (Vaihinger's main focus) and law (Fuller's).Footnote ¹² Fuller's decision to devote a third of his overall discussion to these issues suggests that he considered it necessary to look beyond legal concepts and vocabularies, including to philosophical, scientific and everyday discourse, to understand the dynamics, function and potential risks of specifically legal fictions. Yet only one post-Fuller theorist has acknowledged the space Fuller devoted to these issues in his trilogy (Samek, Reference Samek1981).Footnote ¹³ As a result, the work of Vaihinger, as well as Fuller's position on the parallels between scientific and legal fictions, is little known to contemporary English-language theorists of the legal fiction.

Vaihinger was a German neo-Kantian philosopher who wrote the work for which he is best known, The Philosophy of ‘As-If’, in the late nineteenth century; the linguistic philosopher C. K. Ogden published an English translation in 1924,Footnote ¹⁴ but apart from Fuller's treatment, the work has since its translation received little English-language discussion.Footnote ¹⁵ Typical of his generation, Vaihinger constructed a syncretic, systematic theory drawing on contemporary trends in biological science, psychology and transcendental philosophy; he called his theory ‘positivist idealism’ or ‘idealist positivism’ (Vaihinger, Reference Vaihinger and Ogden1924, p. xli). Vaihinger's central idea, building on Bentham's theory of fictions and paralleling some of the insights of then-contemporary American pragmatism,Footnote ¹⁶ was that human cognitive operations originate in the biological necessity of coping with an environment but are nevertheless inevitably ‘emancipated’ from this instrumental function to become themselves the objects of cognitive operations (p. 15).Footnote ¹⁷ This process generates both abstraction and ‘fiction’ (pp. xxix–xxx), which Vaihinger calls ‘“As if”, i.e. appearance, the consciously-false, [which] plays an enormous part in science, in world-philosophies and in life’ (p. xli).

Vaihinger's exploration of these ideas runs to nearly 400 pages in the English translation. Half of the book focuses on the operation of fictionalising in various spheres of human activity, particularly mathematics and the sciences (pp. 1–177). Vaihinger explains fictions in scientific and technical fields as devices for helping us to discover facts about reality.Footnote ¹⁸ And he describes ‘juristic fictions’ as precisely analogous to such mathematical and scientific fictions as the concepts of zero, negative numbers and infinity.Footnote ¹⁹ Fuller summarises much of this information in his third essay on legal fictions. In particular, Fuller reiterates Vaihinger's point regarding the parallels between scientific and legal fictions:

But, Fuller continues, at least those legal fictions that ‘attempt to describe the “nature of the law in general”’, or ‘jurisprudential fictions’, ‘represent a precise parallel to the methods of the physical scientist’, in that ‘[t]hey are attempts to reduce a complex reality to a simple formula, and thus render it tractable to calculation’ (p. 903). Moreover, Fuller also describes the more workaday ‘fictions of legal technique’, such as ‘[t]he notions of a legal right and duty, of legal powers and liabilities, of legal personality’, as ‘all … in the Vaihingerian sense, fictions, and … essentially similar to the concepts of the physical sciences’ (p. 905).Footnote ²⁰

Fuller uses Vaihinger, then, to argue for the existence of basic similarities between legal fictions and other linguistic devices developed outside the law to manage social relations and interactions with a complex non-human world. Fuller does not see fictions as unique to the law; although they might take some legally specific forms, for Fuller, any special characteristics of legal fictions indicate not legal pathologies (as Bentham had maintained) (see Samek, Reference Samek1981, pp. 298–99; see also Quinn, Reference Quinn2013), but just the functions of those fictions in particular social and institutional settings. Fuller adds to Vaihinger's framework as well: Fuller's account is slightly more sensitive than Vaihinger's to the importance of communities of convention and practice for the existence, maintenance and utility of fictions, as well as to the need for those within such communities to justify their activities to others, a need Fuller presents as a matter of ‘persuasion’ in his first essay. In Fuller's view, legal fictions differ from other fictions mainly due to their generation within a specific community, that of legal practice and commentary; scientific and technical fictions result from parallel processes within different communities.

Unfortunately, post-Fuller theorists have not made much use of these points. Instead, they have virtually without exception adopted legal-doctrinal or analytic philosophical vocabularies to discuss the topic. They have, that is, participated ‘internally’ in processes of conventionalisation that Fuller's trilogy suggested could be fully grasped only from a partly ‘external’ perspective. This self-limitation is not necessarily vicious. Among other things, it has kept the commentary on legal fictions comprehensible to others within the same discourse communities. But it has also left important implications of Fuller's theory unexplored.

The rest of this paper pursues some of these implications. Section II turns to a survey of some post-Fuller developments in the study of communities of linguistic convention, describing the approaches and findings of major late-twentieth-century work on the analysis of academic and professional discourse. Section III applies these approaches to two bodies of relevant professional discursive practice: first, the roughly 100 Supreme Court opinions that have used the ‘legal fiction’ label, with a focus on the normative and persuasive function of the label in these sources; and, second, the 50-plus most-cited American law review articles on the subject of the legal use of scientific evidence.Footnote ²¹ These discussions make available insights inaccessible to Fuller's previous commentators, suggesting, among other things, the limits of Fuller's counterfactual definition and some of the reasons for the necessarily ever-shifting shape and location of legal fictions in legal discourse.

2.1 The study of discourse patterns

Over the past several decades, rhetoricians who teach and study writing across the curriculum or for professionals-in-training began to produce a cohesive body of studies of discourse patterns in non-fiction texts (e.g. Bazerman, Reference Bazerman1988; MacDonald, Reference MacDonald1994; Halliday, Reference Halliday1973; Swales, Reference Swales1990). These scholars' pedagogical and scholarly interests drew them to practical, detailed analysis of the differences in discourse patterns between different academic and professional communities. As Charles Bazerman, a pioneer in the area, put it, the driving assumption in this work is that ‘[r]hetoric is ultimately a practical study offering people greater control over their symbolic activity’ (1988, p. 6 n.7).

The work tends to be highly interdisciplinary. Qualitatively, it continues the traditions of close reading and analysis developed in literary criticism, hermeneutics, cultural anthropology, social and intellectual history, the philosophies of language and science, and phenomenology. But it also examines material features of bodies of texts (as opposed to the single texts that tend to be the focus of close reading); sometimes this undertaking supports more quantitative approaches drawing on bibliometrics and citation analysis (techniques originally developed in information science fields), all considered in the light of more general theories about the social and material setting of such practices, drawn from the sociology of knowledge and the professions, as well as psycho- and sociolinguistics and psychology.Footnote ²²

A few legal scholars have adopted similar approaches, especially to citation analysis, but legal scholars have not paid much attention to most of this work (see Petroski, Reference Petroski2012, pp. 368, 371, 374–78). The most analogous legal scholarship seeks, rather, to assess the influence of legal scholarship on legal decision-making outside the academy (e.g. Petherbridge and Schwartz, Reference Petherbridge and Schwartz2011; Harner and Cantone, Reference Harner and Cantone2011) or to advance the legal realist project of demystifying legal discourse (the most recent work on legal fictions discussed above tends to fall into this category). But the findings of discourse analysis also yield other possible paths of inquiry. The rest of Section 2.1 outlines some of these possibilities suggested by two influential works in this area: Charles Bazerman's Reference Bazerman1988 study of the historical development and features of the scientific ‘experimental article’ and Susan Peck MacDonald's Reference MacDonald1994 extension of this work to the study of academic writing in the humanities and social sciences.

Bazerman's pioneering book characterises academic scientific discourse as the primary product of a particular inherently social activity.Footnote ²³ To Bazerman, ‘a[ny] community constitutes itself in developing its modes of regular discourse’ (1988, p. 79); this is true of both scientific and academic communities, whose members are devoted to the generation of discourse and in turn find their communities shaped by features of that discourse.Footnote ²⁴ Historically, Bazerman traces how the social and publication structures of mid eighteenth-century English scientific practice coalesced around a strategy of ‘exclusivity’ rather than ‘inclusivity’ (p. 136); how ‘empiricism’ then developed as ‘a mode of persuasion’ by which the private witnessing of empirical facts began to require special training to translate into explanations of experimental results (p. 140); how scientific articles subsequently increased in length and began citing other articles at higher rates, especially as article writers started integrating citations into their arguments as a kind of shorthand reference to complex concepts (p. 167); how the conventional grammar and syntax of experimental prose likewise changed in ways ‘suggest[ing the] increasing intellectual complexity’ of the claims being made (p. 169); and, ultimately, how writing in social-scientific fields began to ‘imitat[e] the forms of argument developed within the natural sciences’ (p. 257). In part, Bazerman bases these conclusions on analysis of the microstructures of scientific writing, including sentence structure as well as argument structure and content; these features affect and restrict what can be said and how, even though the acknowledged purpose of scientific discourse is ‘just simply to record the natural facts’ (p. 14). We see scientific discourse as unproblematically fulfilling that purpose, Bazerman contends, partly because these microstructures are so familiar as to be invisible to us – but they were not always so. Bazerman does not explore the implications of his approach for legal discourse, but his concerns and general conclusions are completely consistent with Fuller's.

Although Bazerman's focus is on scientific writing, he does suggest that scholarly writing in the social sciences (1988, pp. 257–88) and humanities (pp. 39–45) might differ from scientific writing in meaningful ways. Susan MacDonald has further explored some of these differences (MacDonald, Reference MacDonald1994, pp. 4–5, 7). Continuing Bazerman's inquiry into microstructures of communication, MacDonald examines ‘sentence-level differences in disciplinary knowledge-making’ (pp. 147–69). The centrepiece of her analysis is a classification of grammatical sentence subjectsFootnote ²⁵ in academic writing from three disciplinary subfields of psychology, social history, and literary criticism and theory (pp. 56–70, 75–107, 109–44). This study shows, according to MacDonald, that writing in disciplines conventionally considered scientific or quasi-scientific, like psychology, contains a relatively greater number of abstract, ‘epistemic’ sentence subjects (p. 160), suggesting ‘that psychologists place greater importance on cooperative disciplinary knowledge making than historians and literary academics’ (p. 160) and that disciplines of this sort tend towards ‘disciplinary compactness’, i.e. ‘progressive[]’ ‘build[ing] upon the work of … predecessors’, rather than reinvention of problems and solutions anew with each scholarly effort (pp. 25, 64–65). Literature scholarship, in contrast, has a ‘less epistemic focus’ (p. 164) than academic writing from the other two subfields; literature scholarship includes more concrete, ‘phenomenal’ grammatical subjects (p. 159). MacDonald surmises that this difference might be related both to literature scholars' almost exclusively textual subject matterFootnote ²⁶ and to these scholars' relative lack of disciplinary community after the erosion of an early twentieth-century ‘gentlemanly tradition of politeness in [literary] scholarship’ (p. 121).

Bazerman and MacDonald are just two among many academics with similar interests in language usage within specialised communities dedicated to the production of written material. The professional and academic legal communities share this dedication, as Fuller recognised. Many of the core features of academic writing, such as its presumed ‘transparency’,Footnote ²⁷ are also present in professional legal writing (but see Mellinkoff, Reference Melinkoff1963). And like academic writing, legal writing is a central product of a professional community largely defined by its members' relationship to a tradition of publicly disseminated texts. More generally, the legal community shares academic communities' goals of engaging in ‘socially negotiated inquiry’ (MacDonald, Reference MacDonald1994, p. 7) and the making of ‘knowledge claims’ (p. 9). It therefore makes sense to use discourse analysis to study the discourse of both judicial opinions and legal scholarship. Yet the types of discourse analysis done by Bazerman and MacDonald have been little noted by legal academics to date.Footnote ²⁸

2.2 How studying discourse patterns can advance Fuller's project

One reason for the relative scholarly inattention to the microstructures of legal discourse might be the difficulty of locating this discourse precisely along the continuum of discourse modes proposed by Bazerman and MacDonald, which runs from the natural-scientific at one extreme, through the social-scientific, to the humanistic at the other extreme. Modern judicial opinions and legal briefs involve and depend on considerable ‘disciplinary compactness’; parenthetical cites to precedential opinions, as well as ‘string cites’,Footnote ²⁹ epitomise this discursive characteristic. This feature of opinions and briefs makes them resemble academic writing in natural-science disciplines. On the other hand, legal opinions and briefs, like humanistic work, focus largely on texts considered as texts. Legal scholarship shares both of these features of opinions and briefs, and thus also resembles both natural-scientific and humanistic scholarship. So it is not immediately obvious where legal discourse fits within the schemes developed by discourse analysis.

This difficulty, however, does not make it impossible to study the microstructures of legal discourse. On the contrary, it suggests that such study might usefully advance both our understanding of the kinds of issue that concerned Fuller and our understanding of other specialised discourse communities. For example, we can test both Fuller's understanding of legal fiction as a species of linguistic convention and the prevailing understanding of the phrase as referring to counterfactual propositions by examining how and when judicial opinions have actually referred to parties' or their own statements as legal fictions. As Section 3.1 explains, over the past two hundred years or so, justices of the United States Supreme Court have changed their patterns of such reference in some surprising ways. These changes suggest that the prevailing equation of legal fictions with counterfactual propositions may be too narrow; instead, it seems that judges most commonly use the phrase when they become aware of the futility of further justification to non-legal audiences.

The received ‘counterfactual’ conception of the legal fiction suggests a dichotomy between properly factual and fictional propositions. On this understanding, legal discourse that relies on fictions would seem to be the antithesis of factual scientific discourse. Notably, however, Fuller did not make this argument. Rather, like Vaihinger, and anticipating the implications of Bazerman's work, Fuller argued that in certain ways the parallels between legal fictions and scientific discourse outnumber the differences. But both Vaihinger and Fuller considered legal and scientific discourse as distinct, non-overlapping systems of communication. They did not directly consider what might happen when members of these different communities try to communicate with members of others, or to make use of the products of other communities' efforts. In legal writing, we can see this intersection of discourses most dramatically in the treatment of scientific information for purposes of dispute resolution. To investigate the legal treatment of ‘fact’ in this sense, Section 3.2 looks at features of legal scholarship on the relations between scientific and legal discourse, analysing shifts over time in the topics addressed, the assumed knowledge of readers, charges to readers, and grammatical subjects (see Bazerman, Reference Bazerman1988, p. 25). These shifts, like the changes in use of the term ‘legal fiction’, show that the recognised boundaries of what is communicable in legal texts – even to scholarly audiences – have not only altered but become more indefinite over time.

3.1 The functions of ‘legal fiction’ in United States Supreme Court opinions

Writers of United States Supreme Court opinions have consistently used the ‘legal fiction’ term to mark claims that lie on the boundaries of what the writers perceive as justifiable to those outside the legal community's conventions. Their usage of the term also suggests that the types of claim perceived as difficult to justify to the non-specialist have themselves shifted over time. These conclusions are based on a study of 120 Supreme Court opinions in which the phrase ‘legal fiction’ has appeared (a total of 142 uses of the term).Footnote ³⁰ The study traced shifts in the frequency with which the term has been used, as well as shifts in what it has been used to refer to – its semantic content. In addition, the study identified shifts in the term's syntactic function, that is, in the frequency with which the term or its referent has coincided with the grammatical subject of the clause in question. The purpose of focusing on the semantic and syntactic functions of the label was to examine how the term has functioned to hook legal justifications up to the semantically indicated ‘real world’ of non-legal fact beyond legal authority.Footnote ³¹

One limitation of the study is that opinions referring to ‘fiction’ or ‘fictions’ alone, rather than to ‘legal fiction’ or ‘legal fictions’, were excluded. The group of opinions studied is therefore underinclusive. But it is not necessarily unrepresentative, since it still includes a wide variety of usages of the term. Moreover, a narrower focus on references to ‘legal fiction’ usefully singles out particularly ambivalent references to the concept. The bare noun ‘fiction’ can connote insubstantiality or illegitimacy (especially but not only when contrasted with ‘fact’). The modifier ‘legal’, however, counters that implication, asserting the effectiveness of the fiction and underlining its transformation into a kind of social fact. More directly than references to ‘fiction’ alone, references to ‘legal fictions’ in legal rationales mark moments of ambivalence about the task of legal justification.Footnote ³²

This ambivalence seems to have grown with time. References to ‘legal fictions’ in Supreme Court opinions increased significantly, starting in the 1970s, following an earlier, isolated high point during the 1930s.Footnote ³³ The increase cannot be fully explained by an increase in the number of opinions issued. This increase challenges the assertion made by some theorists that legal fictions are being overlooked. The Supreme Court, at least, seems to be increasingly alert to the possibility of legal fictions' existence and operation. This fact does not, of course, invalidate Smith's argument that courts, including the Supreme Court, are recognising and relying on new legal fictions (Smith, Reference Smith2007). But, over time, the number of references to legal fictions in concurring and dissenting opinions, as opposed to majority opinions, has increased.Footnote ³⁴ These separate opinions are usually less concerned with orthodox legal justification and the articulation of doctrine (Smith's focus) than with criticism of the authoritative justifications advanced in majority opinions.Footnote ³⁵ The function of such separate opinions is to assert, and thereby create, a weakness in the justificatory authority supplied by the majority opinion (and this is true even when the separate opinions thereby contribute to the Court's overall institutional legitimacy). Both the increased frequency of concurrence and dissent and the increased use of the ‘legal fiction’ term in such separate opinions indicate increased ambivalence about the defensibility of legal justification in general and certain justifications in particular.

Patterns in the semantic content of the term ‘legal fiction’ tell a similar story. To examine these patterns, I classified the referents of the term ‘legal fiction’ in the opinions using a scheme based on MacDonald's classification of grammatical subjects.Footnote ³⁶ She used two general categories – ‘phenomenal’ (referring to the real world) and ‘epistemic’ (referring to knowledge and/or discourse) – and several subcategories within each. Many referents of the term ‘legal fiction’ in Supreme Court opinions, however, do not seem to fall into any of MacDonald's subcategories, which she developed for the analysis of scholarly writing. I therefore created two new subcategories of epistemic–phenomenal referents: one for legal status, rights, relations and duties (e.g. the legal personhood of a corporation,Footnote ³⁷ legal equality,Footnote ³⁸ the party status of an organisation in litigationFootnote ³⁹), and the other for explicitly counterfactual propositions (e.g. the assumption that Congress is collectively aware of judicial interpretations of a statute,Footnote ⁴⁰ the notion that public figures have consented to make their lives publicFootnote ⁴¹). These two categories are effectively epistemic–phenomenal hybrids; they refer to (epistemic) legal utterances, acts or assumptions that link the extra-legal (phenomenal) world of individuals and groups with the legal (epistemic) world of, among other things, rights, duties, litigation, legislation, imputed mental states and judgments.

These two epistemic–phenomenal referent subcategories are by far the most common types of referent of the term ‘legal fiction’ in the opinions analysed.Footnote ⁴² Nearly half of the 142 occurrences of the term refer to legal rights and status, and nearly another quarter refer to counterfactual propositions. The only other referent types occurring more than a handful of times fall into MacDonald's fourth subcategory (the epistemic subcategory she calls ‘reasons’, which I broadened to include arguments, grounds, justifications, bases and positions, and which accounted for 14 of the 142 referents) and her seventh subcategory (the epistemic category she calls ‘-isms’, which I broadened to include doctrines and classes of legal rules, and which accounted for 18 of the 142 referents).

This distribution – particularly the disproportionate frequency of references to legal rights and status, which outnumber references to counterfactual propositions – shows that the ‘legal fiction’ label has been most often applied, in Supreme Court opinions, not to postulates or propositions acknowledged to be untrue or inaccurate, but rather to postulates and propositions that give some phenomenal reality a legal meaning, applying to that reality a label that users of legal discourse can work with. Interestingly, it was only in the twentieth century – shortly before Fuller published his trilogy and around the time Vaihinger was writing – that the Supreme Court began using the phrase ‘legal fiction’ to refer to counterfactual propositions. Before the 1920s, opinions used the phrase to refer to legal rights and status far more often.Footnote ⁴³ Despite the stress on counterfactuality in defining ‘legal fiction’ since Fuller, then, legal writers were using the term long before this to refer merely to artificial legal designations, legal terms and concepts describing and effectuating the interface between the phenomenal world and the world of legal language.Footnote ⁴⁴ And they have continued to do so.

The syntactic function of the term in Supreme Court opinions supports similar conclusions about its role in justification.Footnote ⁴⁵ Rhetorical analysis often focuses on grammatical subjects because they receive the primary emphasis in English propositions; subjects are usually either the agent or recipient of the action described by the clause in question. When clauses are structured indirectly, so that the grammatical agent is not the rhetorical theme of the clause, this indirection may itself be significant; it often suggests that the writer is de-emphasising or avoiding full commitment to the rhetorical theme.Footnote ⁴⁶ In almost every instance in the present study – all but 12 of the 142 referent clauses analysed – the referent of ‘legal fiction’ differs from the grammatical subject. If we assume that ‘legal fiction’ is the theme of most of the clauses in which the term appears, this lack of coincidence between subject and theme in Supreme Court opinions indicates that the writers of these sentences have lacked confidence that they could persuasively describe legal fictions as affecting the world in any meaningful way – or simply did not want to do so.Footnote ⁴⁷ When identifying legal fictions, that is, opinion writers have grammatically emphasised not the fictions themselves, but the grounding of the writers' propositions in phenomenal reality or in some further legal reason or evidence.Footnote ⁴⁸ Sometimes this emphasis is a matter of dismissal: the writer is seeking to discredit another opinion-writer's position by arguing that the position rests on a legal fiction. But at other times, opinion writers have apparently used the term ‘legal fiction’ to concede a point to sceptics while arguing for an admittedly fictional position.Footnote ⁴⁹

Overall, the location, semantic content and syntactic function of these references tell a consistent, if complex, story. Supreme Court opinion writers have consistently used the term ‘legal fiction’ to mark assertions of specifically legal ‘fact’: relations and powers created and recognised by humans in the legal system by means of legal language, not just counterfactual propositions in the broadest sense. When such a legal ‘fact’ gets tagged as a ‘legal fiction’, it is a sign that the explanation in question is coming close to the limit of what the writer perceives as justifiable to readers who are outsiders to the legal community's conventions. The term functions as a label for other legal terms, or for concepts named by legal terms, that cannot be succinctly justified in lay language, but are nevertheless accepted by judges and lawyers as ingredients of conventional forms of legal reasoning. Those who are legally trained understand these terms and concepts as meaningful, but non-lawyers generally do not. Justices' use of the term also acknowledges the public role of judicial reasoning in the United States; Supreme Court opinions are not addressed only to legal insiders, even though they are most understandable to insiders (Short, Reference Short2012; Staszewski, Reference Staszewski2009). It is in part through their publicity, and their presumed acceptability by non-legal actors (if not their actual acceptance), that these opinions claim legitimacy and declare their effects on social ordering (Cover, Reference Cover1986; Schauer, Reference Schauer1995). The ‘legal fiction’ label thus combines an assertion that legal language transforms real-world relations with a defensive concession that the non-legally trained might not understand exactly how or why.

Occasionally, opinion writers have made something like this acknowledgement explicit. In a 1948 majority opinion, for instance, Justice Frankfurter argued against ‘call[ing] commerce [that is] in fact interstate “local commerce” [simply] because under a given set of circumstances’ it may be regulated without violating the dormant commerce clause; this labelling, Justice Frankfurter argued, would be ‘to indulge in a fiction’, and ‘[e]specially in the disposition of constitutional issues are legal fictions hazardous, because of the risk of confounding users and not merely readers’.Footnote ⁵⁰ In part, Justice Frankfurter is arguing against misuse of a conventional legal term, ‘local commerce’. But in making that argument, he also more generally cautions judges to use legal fictions with care and only when necessary, and he distinguishes between the effects of fictions on ‘users’ and ‘readers’. In so doing, he expresses unusual self-awareness about the close and somewhat self-defeating relationship between the ‘legal fiction’ label and the task of legal justification.

3.2 The boundaries of communication in scholarship on science and law

It thus seems that the term ‘legal fiction’ has a broader semantic compass than post-Fuller theorists have suggested in their emphasis on counterfactuality. Consistently, writers of Supreme Court opinions have applied the label to aspects of their legal justifications that provoke ambivalence because further defence or explanation seems difficult. They use the term and the concept to mark some of the points at which legal language directly impinges on the extra-legal world of what MacDonald calls phenomenal fact.

Legal language is, of course, not the only specialised communication system that impinges on the phenomenal world. Bazerman's history of the experimental article focuses on another such system, that of experimental science. Bazerman describes this system as one that identifies and delineates in words – and not only for initiates – the factual world of ‘passive constraints’ on what can be said and understood.Footnote ⁵¹ Legal writing performs an analogous function, and it also interacts with the systems analysed by Bazerman. Scholarship on the legal treatment of scientific evidence directly addresses that interaction. From the presentation of the claims made in this scholarship, we can tell how science has been described to make sense to lawyers and where legal vocabularies have been seen to need alteration to accommodate issues considered significant by scientific and technical communities. We can assume that scholarship on these topics that has gained wide readership in the legal community – measured by the imperfect yardstick of citation counts – has been most influential in negotiating and shaping this boundary between discourse systems.

To explore how this shaping has occurred, I analysed features of the fifty-two most-cited articles published in American law reviews on the legal use of scientific and technical expert evidence.Footnote ⁵² The historical scope of these materials is slightly narrower than that of those discussed above: publication dates range from 1901 (Hand, Reference Hand1901) to 2000 (Risinger, Reference Risinger2000). Overall, the materials indicate that over this period, anxieties about the integration of scientific and legal discourse have increased. Article authors have grown less confident that their readers share a common understanding of how the phenomenal world should be perceived and described, but more confident that such a shared understanding is desired or expected. They have thus also apparently felt greater pressure to map out specific prescriptions for that understanding and to stress the capacity of legal discourse to absorb divergent vocabularies.

These articles map the boundaries of legal discourse in part through the articles' own boundaries: their opening and closing sentences. As Bazerman notes, the ‘attention’ an academic article pays to its readers ‘can be seen in the knowledge and attitudes the text assumes that the readers will have, in the types of persuasion attempted … and in the charge given by the author to the readers (i.e., what the author would like the readers to do after being convinced by the article)’ (Bazerman, Reference Bazerman1988, p. 25). An article's opening sentence is the first, most basic statement of the ‘knowledge and attitudes’ the author assumes readers have; its closing sentence gives a ‘charge’ to readers. My analysis therefore focused on these features of the articles.

Some opening sentences are declaratory statements, asserting information that author and audience (the legal academic or practitioner reading the article) are assumed to share. The assumption may be relatively explicit. A 1939 article on blood-alcohol testing techniques, for example, opened with the following: ‘The great number of injuries and deaths resulting from motor vehicle accidents caused by drunken drivers is a matter of common knowledge with which the public is vitally concerned’ (Ladd and Gibson, Reference Ladd and Gibson1939, p. 191). But not all articles open with such a blunt declaration. Others pose or describe a puzzle or problem, an area where consensus is acknowledged to be lacking, as in this opening sentence taken from a 1943 article: ‘We can think of no more vexed or vexatious law-medicine problem than the proper appraisal and just compensation of so-called “traumatic neurosis”’ (Smith and Solomon, Reference Smith and Solomon1943, p. 87). An opening sentence of this kind states a puzzle: it registers insecurity about the area of consensus. It establishes the need for a particular contribution, identifying how the contribution will complete existing understandings. Over the period covered by this study, ‘puzzle’ opening sentences have become more common, but not markedly so. Only during the 1980s did sentences of this kind outnumber declaratory, propositional opening sentences.Footnote ⁵³

Other trends in opening sentences are more pronounced. Some opening sentences describe concrete, phenomenal events or states; others describe more abstract propositions such as relations or concepts.Footnote ⁵⁴ No opening sentences with concrete referents appeared in any articles in this study published before 1970. We might explain this by noting that concrete opening sentences presuppose a shared sense of what counts as ‘phenomenal’, so they are a feasible rhetorical approach only in the wake of a process of factualisation like the one described by Bazerman in scientific communities. The increase in concrete opening sentences could therefore indicate an increase in the perceived stability and authority of scientific discourse and knowledge, both absolutely and relative to legal discourse. Together, the increased rates of puzzle and concrete opening sentences might suggest an increasing desire to draw on the stability of those extra-legal specialised vocabularies; in turn, and in combination with the patterns described below, this suggests a corresponding increase in uncertainty about the precise boundaries of shared legal vocabularies.

Closing sentences indicate similar pressures. Two types of closing sentence correspond to the proposition and the puzzle, respectively: a closing sentence may either describe a state of affairs (perhaps making a prediction)Footnote ⁵⁵ or summon readers to action.Footnote ⁵⁶ Over time, call-to-action closing sentences have become more common; in fact, since the 1960s, they have consistently outnumbered descriptive closing sentences (with the exception of the period since 2000, for which the study included only one article).Footnote ⁵⁷ Such call-to-action closing sentences began to appear in highly cited articles around the same time, in the mid twentieth century, as puzzle opening sentences.Footnote ⁵⁸ This might not be entirely coincidental; both devices imply that the current treatment of scientific information by law is in some way dysfunctional, posing a problem and requiring changed behaviour. Given the forum in which these articles have circulated – journals addressed to legal academics and professionals – it is unsurprising that closing sentences, in contrast to opening sentences, have not become more frequently concrete over the past century. Virtually all of the closing sentences in the study are abstract.Footnote ⁵⁹ Even if the relative authority and perceived stability of scientific, phenomenal discourse have increased, that is, and even if that authority increasingly requires accommodation by legal discourse, no analogous process of factualisation seems to have occurred in this area of legal discourse. More and more, legal discourse is being asked to account for facts – whether agreed-upon or contested – without itself being able to generate anything similar.

The grammatical subjects of opening and closing sentences also support this understanding. The subjects of opening sentences are fairly evenly distributed among the ten subcategories used for analysis but fall mostly into the first, second and seventh: phenomenal particulars; phenomenal groups; and epistemic ‘-isms’, schools of thought, and doctrines.Footnote ⁶⁰ Closing sentence subjects are also fairly evenly distributed, with a slight majority falling into the second and seventh subcategories, groups and ‘-isms’.Footnote ⁶¹ (In the study of Supreme Court opinions discussed above, grammatical subjects also often fell into category four, reasons.) These patterns support Bazerman's claim that opening sentences indicate areas of assumed consensus and/or authorial goals: authors of scholarship on scientific evidence have consistently sought to anchor their discussions in a shared reality by using phenomenal subjects and to state their objective of shaping legal vocabularies to accommodate such reality by using ‘-ism’ subjects (the most common type of subject for articles' opening sentences and the second most common for closing sentences). But in more than half of the opening sentences using epistemic subjects referring to schools of thought or disciplines (seven out of thirteen), those subjects refer to non-legal ‘-isms’, such as general scientific advancements (McCormick, Reference McCormick1982, p. 879), laboratory methods (Skolnick, Reference Skolnick1961, p. 694), technologies for determining truthfulness (Wicker, Reference Wicker1953, p. 711), multiple regression analysis (Fisher, Reference Fisher1980, p. 702), social science data (Fineman and Opie, Reference Fineman and Opie1987, p. 107), psychology (McCord, Reference McCord1987, p. 19), psychiatry (Hall, Reference Hall1956, p. 761) and teratology (Note, 1962, p. 554). In contrast, in more than three-quarters of closing sentences using these types of epistemic subjects (seven of ten), the subjects refer to legal schools of thought or doctrines, such as rules of relevancy (McCormick, Reference McCormick1982, p. 916), standards for the admission of expert testimony (Black, Reference Black1988, p. 695), the legislation of such standards (Hand, Reference Hand1901, p. 58; Slobogin, Reference Slobogin1984, p. 174), a proposed ‘defendant-first’ approach to evidentiary standards (Ladd and Gibson, Reference Ladd and Gibson1939, p. 267), the trial process in mass toxic-exposure litigation (Sanders, Reference Sanders1993, p. 86) and employment discrimination law (Note, 1975, p. 421). That is, when they refer to settled epistemic areas, the articles refer mostly to non-legal areas at the beginning and legal areas in closing. This pattern is not surprising, given the common goal of all of the articles to digest the ‘-isms’ of non-legal epistemic communities for use by the legal epistemic community. Like the other opening- and closing-sentence patterns described above, patterns in the subject category suggest that authors have felt greater confidence over time about the need for legal discourse to accommodate these non-legal ‘-isms’, without correspondingly increasing confidence that it will do so successfully.

We can see these articles as records of efforts to correct for the kinds of gaps in the justificatory capacities of legal vocabulary memorialised by the uses of the term ‘legal fiction’ discussed above. A reference to a ‘legal fiction’ signals an opinion writer's sense that additional explanation of an appellate court's decision to a non-specialist audience is not feasible for one reason or another. Legal scholarship on scientific evidence, in contrast, proposes new resources for the decision and justification of individual issues and cases. Those proposing such resources seem to have been growing more and more confident that they are needed and that scientific discourse is an appropriate place to look for them.Footnote ⁶² But even as this has been occurring, many of the same writers seem to have been growing less and less confident that their legal readers share a common understanding of how the phenomenal world should be described or how legal discourse should be changed to accommodate any such understanding. Of course, the new resources these writers have proposed would not all, as such, be directly available to appellate courts like the Supreme Court, and the authors of this scholarship are not explicitly addressing those areas of justificatory weakness tagged ‘legal fictions’ by the Supreme Court. But these materials do tell complementary stories about at least two crucial kinds of limit on the capacity of legal language to form an epistemic community consistent with its members' aims, and they indicate the continued validity and suggestiveness of Fuller's observations on the linguistic nature of fictions and their relation to scientific fact.