Z to a scientific text

1 of 11 3rd International Conference on Science and Literature, Paris July 2-4 2018 Dr Ben Toth

Applying S/Z to a scientific text 1 Introduction The epigraph comes from an interview in 2018 with Bruno Latour: How do you cope with telling your grandchildren that you were born in 1947 and had an enormously good time — that you profited from globalization and the process that has led to the sixth extinction. How do you tell this to your grandchildren? If you say, “Well, I had a good life, too bad for you,” you are a moral wreck. So how do you handle this situation? This is fodder for art. ​(Bruno Latour 2018) Never have the sciences and the arts seemed more disconnected, and never, as Bruno Latour suggests, has the need for their reconciliation been greater. I want to investigate the possibility that what we call science and what we call art are in some way complementary rather than solely antagonistic. My specific purpose is to consider the challenge of applying a particular literary technique - the one set out by Roland Barthes in his 1970 study S/Z - to a scientific paper. I should emphasise that I am highlighting challenges today rather than results, and some of these challenges are quite acute. By focussing on texts I am concerned with a small part of what is obviously a much larger issue. But it is an important one. Whatever else science is - and it is many consequential things for the world - it is also a written thing. As Latour observed, a scientific experiment can be seen as a literary technology, first emerging publicly in his words as ‘​a text about a non textual situation, later to tested by others to decide whether or not it is simply a text​’ (Latour 1999:124) 2 The literary dimension of science Analysis of what Peter Dear called the literary dimensions of science (Dear 1991:2) is extensive and difficult to summarise adequately. But roughly speaking it can be divided into two types. The first treats scientific writing as just another sort of writing, and science as just another competing cultural form. The other, at the other end of the spectrum as it were, is based on the idea that the transmission of scientific messages requires a special language, or that a distinctive linguistic form can be detected within whatever language is used for scientific communication. If science is one form of culture, one practice among many, a range of analyses becomes possible, including the role of science in literature and the presence of literary and cultural forms in science. Scientists and scientific ideas are now a significant part of cultural studies, as in Alison Winter’s ​Mesmerized: ​Powers of Mind in Victorian Britain​ (1998), Alice Jenkins’ ​Space and the 'march of mind'​: literature and the physical sciences in Britain, 1815-1850 (​ Jenkins 2007), Sally Shuttleworth’s ​The Mind of the Child​ (Shuttleworth 2010). Frederique Ait Touati’s ​Fictions of the Cosmos​ ​(2011), Susan Squiller’s ​Epigenetic Landscapes​ (Squiller 2017); and Adam Toon’s ​Models and Make Believe​ ​(Toon 2012).

2 of 11 3rd International Conference on Science and Literature, Paris July 2-4 2018 Dr Ben Toth The other approach, based on the idea that vagueness is a feature of language but not a feature of the world, is to attempt to free language of ambiguity. Efforts to do so have a long pedigree. John WIlkins proposed that science be expressed in a universal form in 1668 (Wilkins 1668). In the first half of the twentieth century, as language moved to the centre of philosophical enquiry, one sees Russell’s minimum vocabulary; Wittgenstein’s symbolic logic; the Vienna Circle’s logical positivism; Ramsey sentences, and Tarski’s analysis of the notion of true sentences. The model for these approaches is mathematics, such that the ideal scientific statement would be compressed into mathematical notation. And since this has proved difficult to achieve, a version of this concern is to uncover within ordinary language a distinctive syntax, semantics, or underlying form which characterises scientific expression. Sitting between them, neither linguistic analysis nor cultural critique, or rather, a bit of both, is the rhetorical analysis of science. And being in the middle raises the question put by Alan Gross, namely ‘​can a rhetorical hermeneutic, or a way of reading texts as rhetoric, be anchored in coherent and enabling theory?​ (Gross 1997). Addressing this question is a significant body of recent work in the rhetoric of science by Fahnestock (Fahnestock 1999), Cicarelli (​Ceccarelli, L. 2001)​, Conduit (Conduit 1999), Harris (Harris 1991), and Locke (Locke 1992), in addition to Gross’s work, with Shapin and Schaffer’s Leviathan and the Air Pump ​(Shapin and Shaffer 1985) standing as a distinctive way-marker, combining the social study and rhetoric of science. The work of Roland Barthes probably falls somewhere in this middle ground. But his name appears only once, trivially, in the bibliography of the Society for Literature Science and the Arts 2001-2011 (Rhee 2018), and while the title of Alan Gross’s Starring the Text​ playfully alludes to ​S/Z,​ he makes no further use of it, nor does Locke in ​Science​ ​as Writing(​ Locke 1992). In recent anthologies Barthes is represented better in The Routledge anthology (Clarke and Rossini 2011) than the Cambridge one (Meyer S 2018). And it is perhaps worth noting in passing the limited characterisation of Barthes in Randy Harris’s 2009 commentary on Gross (Harris 2009), and the lack of an entry in the Encyclopaedia of Literature and Science (Gossin 2002) ​This level of under-representation ​is striking because Barthes was, from his earliest appearance, concerned with the relationship between rhetoric, technology, science, and society. African Grammar,​ published in 1957, concerned colonialist rhetoric; the relevance of his ​The Rhetoric of the Image​ and ​The Old Rhetoric an Aide Memoire,​ is obvious; and the extension of rhetoric in a later work on photography is important. And especially in a statement made in 1967 that ​only writing can break the theological image imposed by science​ (Barthes 1967), the potential for his work to re-invigorate the largely American discipline of rhetoric of science is clear.

3 S/Z What I want to do today is propose the techniques set out in ​S/Z​ as a way of anchoring the reading of a scientific text in a coherent theory. Doing so suggests that a scientific

3 of 11 3rd International Conference on Science and Literature, Paris July 2-4 2018 Dr Ben Toth text occupies the same epistemic space as a work of fiction, or to put it another way, that a work of science is also always a work of literature because the same force that produces literature produces science. And the purpose of using S/Z is to suggest this can be claimed without losing what is distinctive to each. This is only a preliminary report. It sets out briefly the major elements of the technique in S/Z, and highlights some areas of difficulty in applying it to a scientific text. S/Z, a 200 page essay about a 13 page story, first published in French in 1970, and in English in 1974(Barthes 1970a), has been acclaimed as a turning point in Barthes personal career and as a mark in the transition from structuralism to post-structuralism. It is perhaps the most detailed literary forensic analysis of a literary text ever produced, and consists of three fundamental operations: -

the text of a short story by Honore de Balzac is broken into 561 fragments the presence in each fragment of one or more of five narrative codes is identified the way these codes weave together to produce the text as a story is discussed.

Among other things, S/Z is Barthes’ most systematic account of codes, a term which occured throughout his work, regularly acquiring new meanings, beginning with the codification of myth in language, through analyses of fashion (the vestimentary code), love, and photography. In S/Z code ‘is a perspective of quotations; a mirage of structures’ (Barthes 1970a p20), ‘it is one of the forces that can take over the text’. Narrative codes are constitutive rather than a method of transmission. They are more like DNA code, or moral code - condensed formulations through which visible structures may be generated. The five codes in S/Z are as follows, drawing in part on Barthes interpretation of them in an interview he gave shortly apart publication of S/Z (Barthes 1970b) : The ​Hermeneutic​ (HER) code covers the setting into place of an enigma and the discovery of the truth it conceals. In a general fashion, this code governs all intrigues. The ​Seme​ (SEM) code is usually single words which, in addition to being signifiers as all words are, have the additional property of connotation - invoking associated and secondary meanings. Symbolic ​(SYM) codes are related to semes but exist in a network rather than a single word. They are seen conventionally in rhetoric (for example antithesis, two directly contrasting terms placed together) or tarot (in which the symbolic structuring of cards is used to infer meaning). Symbolic structures are open to substitution and variation, and like semes play a significant role in the multivalence of texts. Proairetic​ (ACT) A term borrowed from Aristotelian rhetoric, which ensures that we read the novella as a story, a succession of actions. Actions can fall into various

4 of 11 3rd International Conference on Science and Literature, Paris July 2-4 2018 Dr Ben Toth sequences which should be indicated merely by listing them, since the proairetic sequence is never more than the result of an artifice of reading; furthermore, since these actions produce effects, each effect will have a generic name giving a kind of title to the sequence, and we shall number each of the terms which constitute it, as they appear (ACT. "To be deep in": 1: to be absorbed) (Barthes 1970a:18) . Cultural​ (REF) codes ‘are references to a science or a body of knowledge; in drawing attention to them we merely indicate the type of knowledge referred to, without going so far as to construct (or reconstruct) the culture they express. Taken together, Barthes explains, ‘the five codes create a kind of network, a topos through which the entire text passes...each code is one of the voices out of which the text is woven’ ...’the convergence of the voices (of the codes) becomes writing’. (p21) Despite the detailed step-by step formulation it contains, and Barthes’ own view that the narrative codes could be applied outside of literature, they have not been applied to many literary works (none according to John Sturrock in 1998 Sturrock 1998 p33) and not at all outside literary studies. 4 Applying Barthes’ narrative codes to a scientific text ‘Without doubt’ wrote Gianno Frazzeto in 2004, ‘science has put on a new face in the past century. It has come to occupy a central role in society and now enjoys a privileged position among the knowledge-producing disciplines’ (Frazzeto 2004) We know quite a lot about scientific papers, which are added to the world knowledgebank at the rate of about 2M per year (AJE 2018). They are dense, meticulous, and modest; they seek to avoid ambiguity, avoid connotation in favour of denotation - on the side of truth or at least of accuracy. But objectivity is not the only goal. As the authors of the Manual of Scientific Style write, and as Shapin and Shaffer identified in seventeenth century scientific writing, ‘​there must also be a profound sense of reality—a connection to the genuine human thought processes that gave rise to theories; to the details and vicissitudes of the experiments that support one contention or another; to the real life circumstances of science ​(Rabinowitz and Vogel 2009). As this excerpt shows the characteristic scientific style of having no style (Locke 1992:120) is ambitious and not easily achieved. Adding to their literary quality, scientific papers are subject to a kind of forensic literary analysis before and after publication - peer review. Is the paper written coherently, are its claims clear, are they plausible, are they likely to be of interest to other people in the field? So whatever aspiration a paper has to transparency, to be a text that isn’t answerable to any of the laws of textuality, it is always subject to some of its laws; it is always subject to interpretation. Scientific texts are, to use Barthes’ phrase, moderately plural (Barthes 1970a:6), and are therefore potentially subject to the same analysis as Sarrasine in S/Z.

5 of 11 3rd International Conference on Science and Literature, Paris July 2-4 2018 Dr Ben Toth It is important to note that the analysis in S/Z is radically different from conventional literary analysis. It is not concerned with the meaning of Sarrasine, or whether it is a good text. It is up to other types of criticism to establish this. The purpose of Barthes’ analysis is to describe, as a commentary, the mechanism by which a text produces meaning, which is the legitimate substance of other types of analysis. The overarching question is can the narrative codes and the associated analysis be applied at all to a scientific text? The answer must be postponed because the only really adequate answer is a completed analysis. But it can at least be broken into a number of subsidiary questions: 4.1 Which text to choose? Barthes noted that the tutor text for S/Z came to him without any great deliberation, part of his rhetorical strategy for asserting that his method would apply to any literary text. And similarly I would like to say that any scientific text might be chosen for analysis. We take this to be case, but in practice there are choices to be made. Firstly, what type of text? I would like to choose a scientific paper rather than a textbook because this seems most amenable to systematic analysis. Second, which science? Foucault observed that his analyses of epistemes dealt mostly with sciences of uncertain status (criminology, psychology) rather than those of certain status such as physics, and of course mathematics itself. The same issue here. And a diachronic consideration - is it easier, or better, to use a current text, or one from the recent past, or even one from earlier centuries? And finally a question about the status of a text. Is it more productive to apply the analysis to a paper that has been accepted into the scientific canon, or are retracted or discredited papers equally suitable? Perhaps even fabricated research, truly a kind of science fiction, which accounts for around 1 in 10 papers depending on definitions and reporting methods (Fanelli D 2009). I do not know the answer to these questions. But the aspiration is to theorise the text of science not to undertake a sort of literary peer-review. The analysis is not about the rightness or wrongness of the text, not an evaluation of its contents, but an analysis of how it functions, leaving largely to one side the issue of whether or not it is a good text. 4.2 Do Barthes’ narrative codes apply? There is no particular numerology in there being five narrative codes. Barthes said somewhere that he planned a sixth code, about the author, but dropped it; and says, in The Grain of the Voice​ that he doesn’t know if his division has any theoretical stability. And yet, he emphasised, as if the number 5 had just this theoretical stability, ‘the first three lexias - the title and the first sentence of the story - have already provided us with the five major codes under which all the textual signifiers can be grouped: without straining a point there will be no other codes throughout the story but these five, (S/Z p 18-19) ​For the purpose of the analysis then, it seems important to see if the five narrative codes cover a scientific paper in the same way as they do a literary text . A preliminary conclusion is that they all apply:

6 of 11 3rd International Conference on Science and Literature, Paris July 2-4 2018 Dr Ben Toth ●

A scientific paper usually presents as a mystery, a question. In formal terms it sets out a hypothesis which it will eventually answer. The ​hermeneutic​ code runs through a scientific paper as its organising principle.

●

A feature of scientific paper is that ​semantically​ it will contain as little ambiguity or connotation as possible. The style of writing is purposefully, dry, direct, mono-semantic. Yet the semantic code is still present in a scientific paper, because unless one yields to the dream of a language of pure signifieds, as John Wilkins did in the early days of the Royal Society, and Borges famously dissected (Borges 1952), language will always contain the capacity to signify more than its putative signifieds. If science ​is​ a type of literature, it is a fiction of declarative sentences, and so participates in the myth of innocence, of language as nature. As Barthes wrote, ‘doesn't a sentence, whatever meaning it releases, subsequent to its utterance, it would seem, appear to be telling us something simple, literal, primitive: something true, in relation to which all the rest ... is literature?’ (Barthes 1970:9).

●

Does a scientific paper include a ​symbolic​ network of the sort Barthes detected in Sarrasine? The network constitutes rhetoric, so the question might be put does science contain rhetoric? In the sense of containing verbal deceit, no. In the sense of being organised to be persuasive, yes, as Jeanne Fahnestock demonstrates in Rhetorical Figures in Science (Fahnestock 1999) . To take one example of rhetoric that she highlights, ​ploche​ - the precise repetition of a term across several sentences in order to stabilize the terms of an argument. The repetitive use of the term QALY (Quality Adjusted Life Year) in health research is an outstanding example of rhetorical ploche, in which a complex and contestable concept becomes stabilized, immune from analysis, amenable to measurement and statistical analysis, through repetition.

●

It is clear that the ​proairetic​ code forms a major part of a scientific text, by describing the sequence of actions taken to perform the experiment, to collect the and then analyse the data. The basic design of an experiment, as taught in schools - the IMRaD format - Introduction, Methods, Results, and Discussion is found widely in scientific papers. It may be, as Peter Medawar suggested in 1963 (Medawar 1963), largely a rational reconstruction of the actual events of a scientific experiment, and certainly scientists are finding it difficult to replicate experiments from given accounts (Barss 2018) . But the basic design provides a telling example of proariesis - the arcing logic inscribed in words describing human activity.

●

In Barthes’ analysis science formed part of the ​cultural​ code, part of the background of knowledge to which the narrative could draw upon. But if a scientific paper becomes the object of analysis then the cultural code informing it becomes slightly different. On one level the network of references is the cultural code for the scientific paper. But like a story a scientific text must draw

7 of 11 3rd International Conference on Science and Literature, Paris July 2-4 2018 Dr Ben Toth on the culture in which it exists, and we can provisionally say that both types of text contain the cultural narrative code to roughly the same extent. To summarise: Barthes’ five narrative codes do apply well to scientific texts. The symbolic , proairetic, and cultural codes are present in the same way and to the same extent as in a work of fiction. The hermeneutic code is present, probably to a greater degree than in most non-scientific texts. The semantic code is present but probably in a reduced quantity. 4.3 Are 5 codes sufficient? But ​is there a remainder in a scientific text that isn’t covered by these narrative codes? There are at least three types of object in a scientific paper which on first encounter do not appear in literary texts, and so may not be covered by the five codes: tables, illustrations, and mathematics. Of the three, tables are the least problematic because they are the closest to being text. They are also clearly part of rhetoric - as practical strategies for organising and advancing arguments. Provisionally, it seems reasonable to imagine that a table can be considered in whole or part as lexias principally assigned to symbolic narrative codes. Illustrations, which can be called scientific visualisation, appear frequently in scientific texts. Historian and philosophers of science tended to overlook illustration at one time, given the ascendency of words over images. That is no longer the case. At least since Leviathan and the Air Pump​ (Shapin and Shaffer 1985) the importance of images and imagery in texts has been recognised, to the extent that Shapin and Shaffer considered the whole of a text to be a visual device, a mimetic technology. The same approach is followed here. Images are included somewhat indifferently in the same analytic framework as text. In the analyses of scientific texts the provisional solution for illustrations is to account for them principally as fragments containing the symbolic narrative code, not requiring any special identification as illustration. This is not to diminish the importance of illustration, but rather the opposite - to set it on a par with the text. The question of mathematics is more difficult to deal with, for several reasons: Firstly, almost every scientific text will include some mathematics, while few literary texts do. Literary analysis has little need to encounter mathematics, even if there are stories based on mathematical ideas, and mathematical structures are latent in works of fiction ​(Drozdz 2015), a finding which would have pleased ​the Oulipo Group of 1960s Paris, which constructed literature using mathematical constraints (Motte 2007). Secondly, ​because maths is a language spoken mainly by specialists, which cannot be translated, and so remains largely inaccessible to cultural reading.

8 of 11 3rd International Conference on Science and Literature, Paris July 2-4 2018 Dr Ben Toth Thirdly, because any appearance of mathematics in a text is linked to something fundamental in our culture - to the idea of mathesis, that the world may be understood in an orderly way. So maths is at once outside the bounds of literary analysis, un-readable, and fundamental to culture. Mathematics puts in a brief appearance in S/Z as ‘an ideally pure message’ (p145), and as ‘the violence ...of ‘pre-demonstrative nomination’ (p130) In other words, Barthes is no help ​because he is speaking about maths from the other side, from the side which sees maths as pure message in counterpoint to the plural messages of literary texts. But this formulation is inadequate. Mathematical arguments may have a literary structure, as Netz showed in his double reading of Archimedes Spiral Lines, w ​ hich finds a ​certain homology of style between the exact sciences and poetry in the Hellenistic world​ (Netz 2009:230) (Netz 2018). More generally,​ as Henk Bos has argued, mathematics is incompletely understood by its practitioners; its concepts are fluid; and the truth of its primary objects is not safeguarded by logic (Bos quoted in Cifoletti 2006), all of which opens maths in principle to cultural analysis. Since ​mathematics is culture’s way of asserting the possibility of certitude, precision, and objectivity in the world, ​a first approach to analysis would be to note an occurence of mathematics in a fragment by assigning the cultural narrative code. Whilst a coding such as REF-Mathesis would not be incorrect for a mathematical statement, this approach seems to limit the potential function of maths within the text. To take one example, a mathematical statement might form part of a hypothesis and hence be assigned to the hermeneutic code. The more one looks, the more tempting it is to conclude that a mathematical statement might be assigned to any of the narrative codes. Or to more than one simultaneously, which could be a recognition of the role played by maths in scientific papers. Dealt with in this way rather than solely a background cultural resource, the analyst has to comprehend the maths. And this creates a really significant issue for which I have no ready answer. As long as one can treat maths simply as a cultural signifier of objectivity one can leave its detail opaque. But this makes no sense ultimately. The analyst has either to become fluent in maths, or has to choose the sort of maths sh/e can understand.

5 Conclusion In conclusion I must apologise for presenting a half finished piece of work because the codes have yet to be applied to a scientific text. But I am looking forward to the challenge of finding out what the application of the codes brings. I can summarise where I have reached as follows. Barthes was pleased with the turn his work took when he published S/Z. He thought his narrative codes gave him the framework he was looking for, and he thought the framework might have wider applicability. In terms of scientific papers, the five narrative codes have full applicability; they cover such things as table of data, visualisations, and maths. There is a serious issue in

9 of 11 3rd International Conference on Science and Literature, Paris July 2-4 2018 Dr Ben Toth knowing how to indicate the functional value of the mathematical language as it occurs in scientific papers, and this must be resolved before an analysis can take place. Thank you 25 June 2018 4126 words References American Journal Experts 2018. Scholarly Publishing in 2016: A Look Back at Global and National Trends in Research Publication Aje.com. N. p., 2018. Web. 20 June 2018. Barss, P 2018 Barss, Patchen. ​Ghost Effects​ University Of Toronto Magazine. Magazine.utoronto.ca. N. p., 2018. Web. 20 June 2018. Barthes R. 1957 Grammaire africaine in Mythologies. Paris: Éditions du Seuil. Available in English as African Grammar in The Eiffel Tower and other mythologies. University of California Press 1997. Barthes R 1970a. S/Z. New York: Hill and Wang. Barthes 1970b. On S/Z and Empire of Signs. Les Lettres Francais. Translated in The Grain of the Voice . Hill and Wang 1984 [[Barthes R . The Reality Effect. Reprinted in 1984 in The Rustle of Language. Hill and Wang]] Borges, Jorge Luis (1964). Other Inquisitions (1937–1952). Translated by Ruth L. C. Simms. Austin, TX: University of Texas Press. Originally published in 1952 Buehl 2016 Assembling arguments: multimodal rhetoric and scientific discourse. University of South Carolina 2016 Ceccarelli, L. Shaping Science with Rhetoric: The Cases of Dobzhansky, Schrödinger, and Wilson. Chicago: University of Chicago Press, 2001. Cifoletti, Giovanna 2006. “Mathematics and Rhetoric. Introduction.” Early Science and Medicine, vol. 11, no. 4, pp. 369–389. JSTOR, ​www.jstor.org/stable/4130139​. Clark B and Rossini M (eds) 2011. The Routledge companion to literature and science. Routledge. Condit, C. M 1999. The Meanings of the Gene: Public Debates about Human Heredity. Madison: University of Wisconsin Press.

10 of 11 3rd International Conference on Science and Literature, Paris July 2-4 2018 Dr Ben Toth Dear P (1991) The literary structure of scientific argument. University of Pennsylvania Press. Drozdz S 2015 Quantifying origin and character of long-range correlations in narrative texts ​https://arxiv.org/abs/1412.8319v2​ [cs.CL] Fahnestock J 1999. Rhetorical Figure in Science. Oxford University Press. Fanelli D (2009) How Many Scientists Fabricate and Falsify Research? A Systematic Review and Meta-Analysis of Survey Data. PLoS ONE 4(5): e5738. https://doi.org/10.1371/journal.pone.0005738 Frazzetto, G. (2004). The changing identity of the scientist. EMBO Reports, 5(1), 18-20. Gossin P (ed) 2002 Encyclopaedia of Literature ad Science.Greenwood Press. Greenland, Sander et al. “Statistical Tests, P Values, Confidence Intervals, and Power: A Guide to Misinterpretations.” European Journal of Epidemiology 31 (2016): 337–350. PMC. Web. 13 June 2018. Gross A (1996) Rhetorical Hermeneutics. SUNY Press. Gross A (2006 ) Starring the Text . Southern Illinois University Press. Harris R (1991). ​Rhetoric of science. 1991. College English​ 53.3, 282-307. Harris, R (2014) "Alan G. Gross: Floppy Eared Rhetorical Rabbit, Redux." Poroi 10, Iss. 2: Article 7. ​https://doi.org/10.13008/2151-2957.1191 Jenkins, A. (2007) Space and the 'march of mind': literature and the physical sciences in Britain, 1815-1850. Oxford University Press: Oxford, UK. Latour B (1999) Pandora’s Hope. Harvard University Press. Latour B 2018 "The Critical Zone Of Science And Politics: An Interview With Bruno Latour - Los Angeles Review Of Books." ​Los Angeles Review of Books. N. p., 2018​. Web. 19 June 2018. Locke D 1992 Science as Writing. Yale University Press. Medawar P 1963. Is the scientific paper a fraud? The Listener ;70:377–378. Meyer S (ed) 2018. The Cambridge companion to literature and science. Cambridge University Press.

11 of 11 3rd International Conference on Science and Literature, Paris July 2-4 2018 Dr Ben Toth Motte, Warren 2007 Oulipo: a primer of potential literature 2nd edition Champaign: Dalkey Archive Press Netz R. 2009 Ludic proof: Greek Mathematics and the Alexandrian Aesthetic. Cambridge University Press. Netz R (2018) The long history of cognitive practices: literacy, numeracy, aesthetics. In Meyer, Steven (editor) The Cambridge Companion to Literature and Science. Cambridge University Press. Rabinowitz H and Vogel S 2009 The Manual of Scientific Style. Academic Press. Rhee Jennifer 2018 Society for Literature and Arts Bibliography https://litsciarts.org/bibliography/​ accessed June 3 2018. Shapin S and Shaffer S (1985) Leviathan and the Air Pump. Harvard University Press Sturrock John (1998) The Word from Paris: Essays on Modern French Thinkers and Writers. London: Verso Toon A (2012) Models as Make-Believe: Imagination, Fiction and Scientific Representation. Palgrave. [[Walsh L 2015. WIREs Clim Change 2015, 6:361–368. doi: 10.1002/wcc.342 - unused reference regarding illustration as rhetoric]]