Stud. Hist. Phil. Sci. 33 (2002) 289–314 www.elsevier.com/locate/shpsa
Testing testimony: toxicology and the law of evidence in early nineteenth-century England Ian A. Burney Wellcome Unit for the History of Medicine, University of Manchester, Maths Tower, Manchester, M13 9PL, UK
Abstract This essay’s principal objective is to examine how, when confronted with a case of possible criminal poisoning, early nineteenth-century English toxicologists sought to generate and to represent their evidence in the courtroom. Its contention is that in both these activities toxicologists were inextricably engaged in a complex communicative exercise. On the one hand, they distanced themselves from the instabilities of language, styling themselves as testifiers to fact alone. But at the same time, they saw themselves as deeply implicated in the difficulties of forging a coherent signifying system out of a disparate collection of signs that in themselves bore no intrinsic meaning. The analysis has three main components: first, to suggest why criminal poisoning featured so prominently in the burgeoning legal literature on evidence which provided the framework for expert testimony in English courts; next, to show that the scientific evidence offered by toxicologists in poisoning cases can be usefully understood as a form of (unstable) language; and finally, to suggest that this recourse to signs informed the toxicologist’s encounter with another type of courtroom expert—the legal advocate—who was equally (though differently) interested in manipulating signs in order to construct (and deconstruct) legally sanctioned proof. 2002 Elsevier Science Ltd. All rights reserved. Keywords: Evidence; Expertise; Poisoning; Testimony; Toxicology
In 1853 the Pharmaceutical Journal issued an editorial setting out in trenchant terms a familiar complaint about the treatment of scientific testimony at legal trials: ‘In the legal profession’, the Journal observed, ‘the all-inspiring ambition is “a verdict.” In the struggle for this prize truth is distorted, fact is so blended with fiction that it appears under false colours, [and] justice is reduced to a mere hypothesis’. E-mail address:
[email protected] (I. A. Burney). 0039-3681/02/$ - see front matter 2002 Elsevier Science Ltd. All rights reserved. PII: S 0 0 3 9 - 3 6 8 1 ( 0 2 ) 0 0 0 0 2 - X
290
I. A. Burney / Stud. Hist. Phil. Sci. 33 (2002) 289–314
Contrast this with the aims of the chemist—the Journal’s exemplary scientific witness: ‘He is not called into the witness-box as an advocate or a logician, but as an authority with reference to a plain matter of fact on which his knowledge and experience enable him to throw some light’.1 Such comments operate on a distinction conventionally drawn between legal and scientific modes of generating proof. In court, plain matters of scientific fact are placed in a context where fiction, false colors and the contingent imperatives of a verdict predominate. There is nothing in this depiction that should surprise us: from the first English manuals on expert witnessing to the present day, commentators have devoted considerable attention to strategies through which science might survive its legal ordeal. John Gordon Smith’s 1825 Analysis of medical evidence, one of the earliest tracts specifically dedicated to this issue, warned the would-be expert of his perilous position when facing the ‘dextrous advocate’, whose principal aim was to ‘trick’ and ‘deceive’. While the scientist might be master of the realm of things, in giving evidence he had to stray into the advocate’s sphere of expertise—that is, into the realm of words. ‘As we can have little evidence as to thoughts, opinions, or intelligence, but through words, which are their proper signs and medium of exhibition’, Smith lamented, ‘a witness can in no other way give testimony’. The expert witness, in this view, is placed in an inescapably inhospitable climate in which he must translate his ‘plain matters of fact’ into a medium over which he is not in full control, and whose primary characteristic is that it yields to manipulation. Smith’s advice is simple, if uninspiring: ‘Caution in speaking, as well as in preparing to speak, should be observed by him’.2 The villain of Smith’s piece is the adversarial basis of trial procedure. This assessment was not only typical of his period, but is one broadly shared by his modernday successors. As a 1993 Royal Commission report argues, the adversarial approach to fact-making runs counter to scientific methods of testing and probing for truth. Adversarialism rests on the principle that truth emerges from a fair fight between two opposing sides. The element of combat is reflected in discussions of the criminal process and the metaphor of battle is never far away: trials are ‘fought’ by advocates (adversaries), who ‘win’ cases or ‘lose’ them. Forensic experts fit uncomfortably into this combative context since the conduct of scientific investigation is, for the most part, characterized by co-operation and collaboration rather than by secrecy and conflict.3 Current analysis thus reinscribes the assumption that representatives of science encounter strategies of ‘manipulation’ only when they leave their own spaces of cooperative and collaborative activity, and enter alien territory governed by antithetical
1 2 3
Anon. (1853–1854), p. 150. Smith (1825), p. 425. Roberts and Willmore (1993), p. 107.
I. A. Burney / Stud. Hist. Phil. Sci. 33 (2002) 289–314
291
norms. Yet recently, and in response to Anglo-American debate over what critics call ‘junk science’ in court (that is, adversarially motivated scientific opinion entered into evidence despite its lack of specialist credentials), a growing literature within the field of science studies has been examining how an analysis of the contingent foundations of scientific fact-making, and the social, cultural and literary technologies that support any claim to scientific credibility, might be used to call into question the assumed opposition between law and science.4 In these accounts the operations of closure in scientific controversy come under particularly close scrutiny. In her pointedly titled ‘What judges should know about the sociology of science’, Sheila Jasanoff draws on a number of core science studies concepts—notably Harry Collins’s model of the experimenter’s regress and Bruno Latour’s analysis of black-boxing—to argue that legal adversarialism shares several of the strategies employed in everyday scientific practice. Advocates, in challenging the foundations of expert evidence by, for example, seeking to expose experimental indeterminacy, are only doing what scientists (and, for that matter, sociologists of science) themselves do.5 Litigation, in this view, emerges as ‘an especially potent resource for making transparent the values, biases, and social assumptions that are embedded in many expert claims about physical and natural phenomena’.6 Jasanoff’s analysis suggests that in order to break out of the usual binary framework for understanding the place of scientific testimony in legal contexts, we might be well advised to attend to the similarities in legal and scientific fact-finding and fact-making, rather than focusing on their differences. Such is my intention in this article, within a highly specific historical framework and with special regard to the observations on language, fact and adversarial play with which I began. My subject is English toxicology in the first half of the nineteenth century, and my principal object is to examine how, when confronted with a case of possible criminal poisoning, toxicologists sought to generate and to represent their evidence in the courtroom. My contention is that in both of these activities toxicologists were inextricably engaged in a complex communicative exercise. On the one hand, they distanced themselves from the instabilities of language, styling themselves as testifiers to fact alone. But at the same time, they saw themselves as deeply implicated in the difficulties of forging a coherent signifying system out of a disparate collection of signs that in themselves bore no intrinsic meaning. How to make and represent stable meaning, in the laboratory and in the courtroom, lies at the core of my consideration of the discourse of toxicological proof. This analysis has three main components: first, to suggest why criminal poisoning featured so prominently in the burgeoning legal literature on evidence which pro-
4
For a sample of this literature, and guides to further reading, see Golan and Gissis (1999) and Jasanoff and Lynch (1998). 5 Jasanoff (1992), p. 349.. Indeed, as Michael Lynch has noted in the context of the O. J. Simpson trial, advocates act as particularly well funded and motivated players in resisting closure—leaving the comparatively impoverished sociologist of science to marvel on the sidelines (Lynch in Jasanoff and Lynch, 1998, p. 831). 6 Jasanoff (1995), p. 20.
292
I. A. Burney / Stud. Hist. Phil. Sci. 33 (2002) 289–314
vided the framework for expert testimony in English courts; next, to show that the scientific evidence offered by toxicologists in poisoning cases can be usefully understood as a form of (unstable) language—that, to echo John Gordon Smith, even in their retorts and test tubes, chemists had to have recourse to a complex system of signs as the only means of expressing what was ‘there’; and finally, to demonstrate that this recourse to signs informed the toxicologist’s encounter with another type of courtroom expert equally (though differently) interested in manipulating signs in order to construct (and deconstruct) legally sanctioned proof.
1. Poison and the law of evidence Criminal poisoning occupied a conspicuous place in early nineteenth-century English medico-legal treatises, with the best known textbooks of the day consistently devoting more of their pages to it than to any other subject.7 It also drew considerable attention outside specialist circles. It was, for one thing, a crime that exercised a peculiar hold on the public imagination.8 More crucially for our purposes, poisoning was of particular concern to legal scholars and practitioners involved in the development of rules for managing evidence in trial proceedings—in particular those relating to indirect or ‘circumstantial’ evidence. The English law of evidence, modern legal historians agree, was largely a product of an interaction between a growing eighteenth- and early nineteenth-century treatise literature and changing courtroom practices. This was itself part of a broader shift, not only in law but in branches of inquiry (theology and natural philosophy, most importantly), from a focus on direct testimony provided by a recognized authority as the grounds for demonstrative knowledge to an epistemology grounded in the corroborative logic of things.9 In the context of English legal theory and practice, as Barbara Shapiro has forcefully argued, this probabilistic orientation was introduced gradually through a highly specific category of crimes long established in the Continental Romano-canonical legal tradition—that of crimum exceptum. Under normal circumstances, indirect evidence (termed indicia, a term which itself echoed classical jurists’ discussions of probabilistic signs adduced as proof of a crime) could not take the place of ‘full’ proof (either the oath of two independent and direct eyewitnesses or confession), but could only serve as the grounds for subjecting the suspect to legal torture with the aim of eliciting confessional proof. By the fourteenth century, however, Conti-
7
For example, the second edition of Male (1818) devotes nearly half its pages to the topic; it comprises roughly a third of Paris and Fonblanque (1823); and the first 280 pages of Taylor (1844) are taken up with the subject, a focus unmatched by any other single subject. 8 Burney (1999). 9 There is a vast secondary literature on this topic: the fullest overview of the connections between law and other fields of inquiry is Shapiro (1983). Her argument for the primacy of law in the construction of the modern concept of ‘fact’ is developed in Shapiro (2000). The classic philosophical exploration is Hacking (1975). The emergence of probabilistic thinking as it relates to the history of English experimental science is a primary theme in the seminal work of Schaffer and Shapin (1985).
I. A. Burney / Stud. Hist. Phil. Sci. 33 (2002) 289–314
293
nental legal theorists were arguing that crimes marked by their essentially secret nature—crimes such as witchcraft, rape and poisoning—were incapable of full proof, and that in these exceptional circumstances indicia should be allowed to stand as sufficient grounds for conviction without resort to torture. Shapiro argues that sixteenth-century English treatise writers, who in general shared their Continental counterparts’ view of circumstantial evidence as inferior, adopted this conception of crimum exceptum as a way of handling cases of inherently secret crime. This, Shapiro argues, was the start of a gradual process through which the relative value of direct and circumstantial evidence became reversed: initially framed as a strict exception to normal proof, by the eighteenth century circumstantial evidence had become detached from the category of secret crime and had become widely regarded as providing a superior form of (probabilistic) proof. In her account, then, the broad integration of probabilistic thinking in English law eventually made secret crimes a redundant category. As all evidence began to be considered as involving inferential logic, this form of proof no longer required its own distinct rationale. But while it may be true that interest in certain forms of crimum exceptum (notably witchcraft accusations) did decline, and that circumstantial proof became accepted as a means of attaining ‘moral certainty’, it is not the case that eighteenth- and early nineteenth-century legal commentators dispensed with the category altogether. This is because proof by circumstance was still a matter for debate—especially amongst treatise writers keen to qualify the kind of enthusiastic embrace represented by William Paley’s famous assertion that ‘circumstances cannot lie’. The principles of circumstantial evidence, English legal theorists urged, needed careful exposition, a point they illustrated by reference to compelling real-life examples. Poisoning cases seemed particularly well suited for this illustrative objective. As Alexander Welch has shown, the eighteenth- and early nineteenth-century English legal community continued to regard poison as a kind of exceptional crime, due to the peculiar characteristics with which it was commonly associated. Unlike more palpably ‘physical’ forms of violence—fixed for centuries in the legal imagination by Coke’s much cited evocation of a bloody affray, with a stained knife as a visible token of the deed—poisoning was figured on as a quintessentially ‘hidden’ act: it could be accomplished without a direct confrontation between murderer and victim; since it was thought to leave no marks on the surface of the body, poison effected an invisible violation of flesh; and since its recognized physiological and pathological signs often simulated natural disease processes, poison’s murderous course lent itself to interpretative obscurity. For these reasons, in Welch’s words, ‘poisoning trials are bound to depend on the evidence of things not seen’,10 a conclusion that was itself frequently voiced in contemporary legal practice and theory. Two of the judicial opinions most cited as 10 Welch (1992), p. 25. A substantial portion of Welch’s important study of the rise of probabilistic narrative across a range of genres is devoted to poison and theories of legal evidence, and this has served as a starting point for my efforts to develop a sustained comparative analysis of legal and toxicological discourses of proof.
294
I. A. Burney / Stud. Hist. Phil. Sci. 33 (2002) 289–314
precedents for the growing acceptance of circumstantial evidence, for instance, were delivered at cases involving the charge of poisoning.11 These and other poisoning trials were explicitly mentioned in the leading general evidentiary treatises of the day,12 while poisoning figured more prominently still in the first treatises dedicated explicitly to explaining and critically assessing the value of circumstantial evidence, W. M. Best’s Treatise on presumptions of law and fact, with the theory and rules of presumptive or circumstantial proof in criminal cases (1845) and William Wills’s Essay on the principles of circumstantial evidence (1838, revised edition 1862). ‘It is in cases of supposed poisoning, Best observes, ‘that the nicest questions arise relative to the proof of a corpus delicti’.13 Wills supports this view by introducing his core discussion of ‘The essential characteristics of circumstantial evidence’ with the following contrast: A witness deposes that he saw A. inflict on B. a wound, of which he instantly died; this is a case of direct evidence. B. dies of poisoning; A. is proved to have had malice and uttered threats against him, and to have clandestinely purchased poison, wrapped in a particular paper, and of the same kind as that which has caused the death; the paper is found in his secret drawer, and the poison gone. The evidence of these facts is direct; the facts themselves are indirect and circumstantial, as applicable to the inquiry whether a murder has been committed, and whether it was committed by A.14 Wills presents this distinction between evidence of wounds—again, the classic legal exemplar of visible and physical violence—and evidence of poisoning as his sole (and thus paradigmatic) example of what distinguishes proof by circumstance. But while poisoning might serve as a particularly instructive illustration of this branch of evidentiary law, this did not mean that poisoning cases should be judged by extraordinary standards. ‘The processes and results of chemical analysis in application to the discovery or reproduction of poison’, Wills insists, ‘are subordinated to the control of those general principles of law which, in all other cases, govern
11 The first case is R. v. Blandy (1752), in which Baron Legge instructed the jury that where a ‘presumption necessarily arises from circumstances, they are more convincing and satisfactory, than any other kind of evidence, because facts cannot lie’ (Howell, 1809–1826, Vol. 18, p. 1187). The other case is R. v. Donnellan, which will be discussed in detail below. 12 In what by common historical consensus is the first English legal treatise to explicitly mention the place of expert testimony, Lofft (1791), the section ‘Of proof by experts’ is principally occupied with the circumstances of one contemporary poisoning trial. Subsequent treatises make more room for poison: MacNally (1802) includes a chapter dedicated to ‘Evidence of medical men in cases of poison’, while Bentham’s five-volume contemporary critique of existing rules of evidence (Bentham, 1827) contains extended discussions of several poisoning cases. 13 Best (1845), p. 166. 14 Wills (1862), p. 17; emphasis original. Though the first edition discussed the connection between circumstantial evidence and poisoning, that provided in the later edition was more fully elaborated, and is the edition that will be cited throughout this essay.
I. A. Burney / Stud. Hist. Phil. Sci. 33 (2002) 289–314
295
the admissibility of evidence and the estimation of its weight and effect’.15 The inherent difficulties presented by poisoning instead meant that the law had to be especially careful of the ‘weight and effect’ of evidence admitted as proof of such crimes. Yet it was precisely this kind of care that S. M. Phillipps, another leading authority on the development of early nineteenth-century law of evidence, argued was markedly absent in poisoning cases. Close scrutiny of case law, Phillipps reported in his anonymously published Theory of presumptive proof, revealed numerous instances of false convictions from ill-judged circumstance or, in his terms, ‘presumption’.16 Phillipps’s intervention is important here for two reasons. In the first (and most straightforward) sense, it stands as a prominent critique of circumstantial evidence in which poisoning served a problematizing rather than an illustrative function. Moreover, careful consideration of his account of how ‘presumptions’ helped to generate courtroom facts suggests a number of striking parallels with the model described by chemists for fact production. This second point will take us from the realm of legal theory and practice and into the toxicological laboratory itself. Phillipps defines legal presumptions as ‘consequences drawn from a fact that is known to serve for the discovery of the truth of a fact that is uncertain, and which one seeks to prove’.17 The law relies on presumptions in matters that afford no direct proof—where a crime was committed in deliberate secrecy, or where the crime by its very nature (both in cause and effect) is hidden from view. Presumptions are thus tools of legal art, accepted pathways for reaching agreement about something unknown. These pathways are not capricious, but are instead framed within legal norms of experience, reason and justice, the most fundamental of which, in Phillipps’s view, is that a presumption must be based on a fact already established as real.18 ‘In laws’, he insists, ‘the arguments should be drawn from one reality to another, and not from reality to figure, or from figure to reality’.19 To demonstrate the dangers of arguing from ‘figure’ rather than ‘reality’, Phillipps turns to a detailed analysis of one of the best known (and most contentious) poisoning cases of the late eighteenth century, the 1781 trial of Captain John Donnellan for the murder of his brother-in-law, Sir Theodosius Boughton. Boughton, the twentyyear-old first born of a wealthy Warwickshire family, who on his next birthday was to come into a per annum income of £2000, died in July 1780 after taking ‘the 15
Ibid. p. 231. Phillipps (1815). Contemporary legal and medico-legal writers commonly attributed this treatise to Phillipps, as do modern scholars such as Shapiro and Welch. As Welch also observes, presumption, a long-standing technical term in English law, had by the mid eighteenth century become more or less synonymous in legal usage with the modern term ‘circumstance’ (Welch, 1992, pp. 20–21). 17 Phillipps (1815), p. 17. 18 ‘They are not just, because they are rules of law’, Phillipps insists, ‘but they are rules of law, because they are just and reasonable’. He continues: ‘It has been found, from common observation, that certain circumstances warrant certain presumptions. Thus, that a mother shall feel an affection for her child,— that a man shall be influenced by his interest,—that youth shall be susceptible of the passion of love,— are laws of our general nature, and grounds of evidence in every country’ (Ibid., p. 12). 19 Ibid., p. 18. 16
296
I. A. Burney / Stud. Hist. Phil. Sci. 33 (2002) 289–314
most gentle and innocent’ medicinal draught prescribed for a venereal condition. Donnellan’s wife was next in line to the Boughton fortune, and Donnellan’s actions following his brother-in-law’s death—most notably his precipitous rinsing of the medicinal phial—led several to suspect him of having introduced poison into Boughton’s draught. Further investigation by the family and its medical advisers intensified suspicion, and Donnellan was eventually brought to trial before the assize court of Sir Francis Buller in March 1781. In his opening charge to the jury, a statement which itself became a classic formulation of the doctrine of circumstantial evidence, Buller made it clear that he considered this an exemplary case of secret crime: ‘You are not to expect visible proofs in a work of darkness’, Buller advised. ‘You are to collect the truth from circumstances, and little collateral facts, which taken singly afford no proof, yet put together, so tally with, and confirm each other, that they are as strong and convincing evidence, as facts that appear in the broad face of day’. The key source for these ‘little facts’ was the deceased’s mother, Lady Anna Maria Boughton. It was she who administered the suspected draught to her son, and because any remaining traces of the substance were washed away immediately after use, her capacity for recalling and distinguishing smell took on a critical evidentiary significance. From the witness stand, Lady Boughton stated that at the time she administered the draught to her son she thought it smelled ‘very strongly like bitter almonds’. She was then asked to smell a bottle containing the mixture that Boughton’s apothecary had intended for his patient, and to declare ‘whether that smells at all like the medicine Sir Theodosius took’. When she replied in the negative, she was presented with a phial containing the same mixture into which laurel water (prussic acid) had been added. ‘This’, Lady Boughton declared, ‘smells very like the smell of the medicine which I gave him’.20 In the prosecution’s view, this was proof that Donnellan had adulterated Boughton’s medicine with a deadly poison, and was guilty of his murder. Lady Boughton was of course not the only witness examined about the phial’s contents. The court heard from five medically trained witnesses who, since none of the suspected poison survived Donnellan’s scrupulous attention to housekeeping and since no chemical analysis of the contents of Boughton’s body was undertaken (the body being too decomposed for successful analysis), testified principally on the basis of clinical and post-mortem observations. The court was particularly interested in whether they could confirm Lady Boughton’s sensory evidence, and four of them made strenuous efforts to do just that.21 The Coventry physician Dr. Rattray, upon smelling the laurel water mixture prepared by the prosecution, testified that he knew of no corresponding smell in the pharmacopoeia. Rattray was then asked whether
20
Paris and Fonblanque (1823), Vol. 3, p. 244. This text includes an extended appendix of extracts from shorthand notes taken at several poisoning trials, including the published notes of J. Gurney in the Donnellan trial. All the following quotes from this trial are taken from this source. 21 The most illustrious of these witnesses was John Hunter, whose lone dissent to the conclusion that Boughton’s death was attributable to poison was in subsequent decades the source of dispute in the legal and medico-legal literature as to whether he represented the face of an ill-prepared or a properly ‘modest’ expertise.
I. A. Burney / Stud. Hist. Phil. Sci. 33 (2002) 289–314
297
‘the smell described by Lady Boughton, something like bitter almonds, convey[s] to you an idea of that mixture?’. ‘It does’, he replied, adding that he had ‘given the laurel water to many people to smell to, and they always described the smell to be something like bitter almonds. I do not exactly know how they expressed themselves’, he continued enigmatically, ‘but they meant to say that’.22 Dr. Parson, professor of anatomy at Oxford, sought a widening of scope by inviting the jury to smell some laurel water, so that ‘they may judge how well it agrees with the description that Lady Boughton has given of the supposed physick’.23 Though when further questioned Parson contradicted Rattray’s testimony by conceding that many medicinal preparations might produce a similar smell, he nonetheless insisted that the court could be confident in the evidence secured from Lady Boughton. This assemblage of indicators adduced by the prosecution, in Phillipps’s critical analysis, represented in both its form and its source all that was dangerous about proofs derived from circumstance. Lady Boughton’s testimony that the smell of the phial as she remembered it was similar to a substance whose smell was in turn known to be like that of the poison imputed to Donnellan, in Phillipps’s view, rested on layers of mediation which could not be accepted as grounds for proof. Lady Boughton had had no prior knowledge of the smell of laurel water, and had only identified it on the basis of its purported resemblance to bitter almonds, a smell she did know: ‘One circumstance was supposed from another, equally suppositious’, Phillipps maintained, ‘and from two fictions united a third was produced’.24 That the material evidence against Donnellan was founded on a merely presumed resemblance was bad enough. Worse still was the transient and inherently unstable source of this evidence. Since the phial had been washed out immediately after the draught had been consumed, Lady Boughton had only a ‘momentary’ experience of its contents, and even that of a highly suspect sort. ‘What’, Phillips demanded, ‘is so uncertain as the sense of smell? Of all the human senses, it is the most uncertain, the most variable, and fallacious’. This was of even greater importance given the peculiar circumstances of the testifying sensory agent: in this case, proof depended ‘upon the breath, the smell of a woman, distracted at the moment, with the loss of her son, and ready to ascribe that evil to the first thing that came in her way’. The Donnellan prosecution, arguing not ‘from one reality to another’ but from ‘inference [to] inference’, thus represented a contravention of fundamental evidentiary principle. ‘A simile’, Phillipps pointedly concluded, ‘is no argument’.25 To summarize the above discussion: for early nineteenth-century legal commen-
22
Paris & Fonblanque (1823), Vol. 3, p. 252. Rattray supplemented this testimony with an account of his own physiological reaction to Boughton’s body at the post-mortem examination: ‘one could not expect any smell but partaking of that general putrefaction of the body; but I had a particular taste in my mouth at that time, a kind of biting acrimony upon my tongue. And I have in all the experiments I have made with laurel-water, always had the same taste, from breathing over the water, a biting upon my tongue, and sometimes a bitter taste upon the upper part of the fauces’, Ibid., p. 260. 23 Ibid., p. 267. 24 Phillipps (1815), p. 36. 25 Ibid., pp. 36, 37.
298
I. A. Burney / Stud. Hist. Phil. Sci. 33 (2002) 289–314
tators, poisoning trials presented a uniquely instructive illustration of proof by circumstantial evidence. Such proofs relied on arriving at the truth of an unknown by agreed-upon indicative means—‘presumptions’. But there was a danger of admitting—through lapses in logic and lapses in perception—ungrounded presumptions (‘figures’ or ‘similes’) in place of the ‘real’. These, therefore, were legal tools that required close monitoring, that were open to dispute, and that gained credibility by being tested within the conventionalized arena of the courtroom. This brings us to the second, and more oblique, use of Phillipps’s critique: a comparison between the legal use of presumption and the analytical activities of the toxicological laboratory— principally the use of ‘reagents’ as tools for probing the chemical unknown. Accordingly, the discussion now turns from tools used in law to establish hidden facts to strategies adopted by the early nineteenth-century chemist and toxicologist to do the very same thing—that is, to gain access to things unseen, and to present their findings not as ‘figure’ but as ‘real’. By reading toxicology through the lens of legal presumption I will be putting into practice the methodological agenda I sketched at the outset, in order to suggest ways of seeing commonalities in the ostensibly distinct activities of advocates and chemical experts.
2. Languages of the laboratory In law, as we have seen, a great deal of emphasis was placed on questions of how proofs should be constructed and communicated. Medico-legal writers, seeking ways to secure expert testimony in the courtroom, were equally concerned with these questions. Their interest in presentational issues should come as no surprise, since whatever their techniques for generating evidence in their own domains, experts have to communicate their findings to the constituencies that they wish to interest. But what is perhaps less obvious (and what certainly has been occluded in the literature on expertise that assumes a co-operative and unmediated process of scientific factfinding) is that by the time the expert witness appears in court, he has already gone through an intricate set of representational procedures through which his ‘plain matter of fact’ is produced. At the core of these operations lies a familiar and self-consciously expressed activity—the ‘manipulation’ of conventionally agreed signs for anterior realities the proof of whose existence is sought.26 To demonstrate this, let us consider the conditions under which early nineteenthcentury chemists in general, and toxicologists in particular, derived evidence of the composition of matter they had under analysis, focusing primarily on one key substance of toxicological interest—arsenic. There are several good pragmatic and historical reasons for focusing on arsenic: for the better part of the century arsenic was 26
‘Manipulation’ was an essential component of chemists’ representation of their activities in this period. Observing that nine-tenths of chemical facts are derived from ‘artificial means’, Michael Faraday embraces the image of the chemical lab as a site of ‘manipulation’ (Faraday, 1829, pp. ii-iii). For more on chemical manipulation, see Knight (1999). For a pathbreaking analysis of the broader epistemological culture of chemistry at this time, see Golinski (1992).
I. A. Burney / Stud. Hist. Phil. Sci. 33 (2002) 289–314
299
regarded as the English poisoner’s agent of choice; it consistently outstripped all other substances in the published statistics on criminal poisoning; discussion of arsenic in medico-legal textbooks overwhelmed all other topics, toxicological and otherwise; and arsenic was the centerpiece of legislative debates on the control of poison, in 1851 becoming the first poison to be made the subject of legal control.27 Little wonder, then, that by the early nineteenth century chemists had developed an extensive battery of tests designed to detect arsenic, each of which relied upon its own sign system for indicating arsenic’s presence. For centuries, arsenic had been signalled through taste and smell, indicators that were toxicology’s inheritance from an empiricist tradition of chemistry privileging bodily experience as the basis of reliable knowledge. Taste and smell as tests for arsenic still featured in the early nineteenth-century literature: as William Brande explained to a Royal Institution audience in 1827, when correctly heated, arsenic conveys a peculiar ‘alliaceious odour’, while its taste ‘is singularly nauseous; it creates a peculiar astringent sensation about the mouth and fauces, a great flow of saliva, and a painful feeling in the mouth which can never be forgotten by those who have made the experiment’.28 Yet the reliability of this incorporated knowledge was by no means universally conceded. Indeed, in the same year of Brande’s insistence on the unmistakable taste of arsenic, this very point was the subject of fierce debate in the pages of the Edinburgh Medical Journal: for some arsenic had an acrid taste, for others it was distinctly sweetish, while a third view maintained its lack of taste altogether. In the face of this controversy, an exasperated Times editorial declared itself ‘astounded’ that so simple a matter should have become the subject of public dispute. By the next decade, Anthony Todd Thompson, professor of medical jurisprudence at the University of London, was advising his students that ‘every test depending on the senses of taste, or of smelling, should be viewed with suspicion, as much of the accuracy of the judgement that they enable us to pronounce, must depend on the condition of the health of the organs of these senses, and on many other circumstances’.29 Contemporary toxicologists could also seek proof of arsenic in a seemingly more tangible form—a form more akin to what Phillipps and other legal theorists would describe as ‘real’ as opposed to ‘figure’. This process, described as the arsenic hunter’s ‘crucial experiment’, involved reducing, in a heated retort, raw matter extracted from the deceased’s body into a metallic arsenical residue. Of all the substances in the poisoner’s toolkit, arsenic was thought especially compliant to this ‘reduction
27
For more on the special status of arsenic in this period, see Bartrip (1992). Brande (1827–1828), p. 67. Brande, who in 1813 succeeded Davy as professor of chemistry at the Royal Institution, was a prominent chemical practitioner and author. 29 Thompson (1836–1837), p. 452. Recall that it was for this reliance on sense-based indicators conventionally and consensually described that Phillipps criticized the Donnellan evidence. In that particular case it was a distraught woman anxious to place the blame somewhere for the death of her beloved son that was at the center of the suspect circuit of proof. But Phillipps’s identification of the limits of this form of proof forms a part of a broader trend in chemistry towards what Lissa Roberts has described as the ‘decline of the sensual chemist’. For this important discussion, see Roberts (1995). 28
300
I. A. Burney / Stud. Hist. Phil. Sci. 33 (2002) 289–314
test’. As a mineral poison, it was easier to isolate and extract than vegetable poisons (which could merge with and get lost in the organic constituents of the body). Moreover, as a ‘thing out of place’, a substance not found naturally in the body, the discovery of arsenic suggested its introduction from an external source.30 The reduction of arsenic thus seemed to present the toxicologist with the opportunity to achieve what his audience in the courtroom most wanted in response to the threat of secret poisoning—to translate this most ephemeral of crimes into a more conventional form of violence. By reproducing the equivalent of the bloodied dagger in his tubes and retorts, the toxicologist promised to establish the corpus delicti (or proof of the commission of the crime) through direct evidence, ostensibly providing a route out of the dependence on the unseen and on the inferential logic of circumstance. It was precisely this capability that impressed legal writers. As far as W. M. Best was concerned, it was when ‘poison is extracted from the dead body by means of chemical analysis’ that science serves law in its ‘most legitimate, valuable, and wonderful application’. ‘Of the various chemical tests’, William Wills agreed, ‘unquestionably those which, applied to the human body or its contents or excreta, reproduce the particular poison which has been employed, are the most satisfactory’.31 Medico-legal commentators also recognized the value of the reduction test, not least because of its capacity to fulfil legal expectation. Robert Christison, the doyen of Scottish medical jurisprudence and the first major British figure in toxicology, observed that reducing a suspected substance to a tangible emblem of itself ‘must obviously be much more satisfactory to the mind of an unpracticed operator, and still more to the unscientific minds of a criminal court and jury—an object which every medical jurist should keep in view’. Its apparent materiality, Christison added, allowed the toxicologist to play to the evidentiary theatrics of the courtroom: the reduction test, he observed, ‘may often be of signal utility, by enabling the medical inspector to lodge in evidence a portion of the arsenic as administered’.32 Yet from the perspective of toxicological practice, this embrace of reduction carried potential dangers. Christison worried that the legal bias in favor of ‘demonstration’ might cast unwarranted doubt on other seemingly less tangible indicators.
30 The importance of maintaining arsenic’s status as an irreducibly ‘external’ agent can be gauged by the outcry inspired by the world-renowned toxicologist Matthieu Orfila’s spectacular (and soon recanted) pronouncement in 1839 that arsenic was a normal constituent of the human body. This controversy is reviewed in Danger and Flandin (1841), and in a series of articles in the London Medical Gazette for 1840–1841. 31 Best (1845), p. 388; Wills (1862), p. 232. It should be noted that any equation between the end result of the reduction test—a metallic-looking residue on a glass plate—and ‘the poison arsenic’ would itself have to rely on a number of mediating layers (linking the non-poisonous arsenical ore to its poisonous arsenious acid form, for example, or the visible characteristics of the metallic residue to arsenic alone). 32 Christison (1829), p. 195. Though this essay is concerned with poison in relation to English legal practice, Christison’s inclusion is justified because of his significant influence on the theory and practice of toxicology across the Scottish/English legal divide. He testified in several of the most important English poisoning trials of the period, and was a frequent commentator on the production of expert evidence in court.
I. A. Burney / Stud. Hist. Phil. Sci. 33 (2002) 289–314
301
John Gordon Smith had earlier voiced similar concerns. Tribunals, he complained, insist on ‘a minute and elaborate process, first of investigation, and then of testimony, ending in a certain arbitrary conclusion, which it is possible may, after all, be at variance with science and truth!’. This, in his view, constituted evidence of the overdependence of scientific experts on legal form, a subordination of science and truth to legal prejudice.33 Protests such as these were grounded in practical considerations. True, the reduction test was reckoned to be a relatively straightforward matter for early nineteenth-century chemists, but only providing one basic requirement was met: there had to be a considerable amount of arsenic present in the substance subjected to reduction. ‘Unless the quantity of metal be considerable’, as Paris and Fonblanque’s medico-legal textbook observed, ‘its metallic splendour and appearance are often very ambiguous and questionable’.34 Yet in most cases of suspected criminal poisoning, commentators generally admitted, only scant material was available. In light of this obstacle, medico-legal writers sought to curb legal expectation of successful reduction, and to promote an alternative group of tests based on color reaction. These four processes, collectively referred to as the ‘liquid tests’, operated on the principle that when arsenic came into contact with one of four chemical solutions (either ammoniacal sulphate of copper, ammoniacal nitrate of silver, lime water, or sulphuretted hydrogen) it would yield a corresponding and characteristic color pattern. These tests, in Paris and Fonblanque’s estimation, were not only ‘capable, under proper management and precaution, of furnishing striking and infallible indications’, but they were often ‘even more satisfactory in their results than the metallic reproduction, on which so much stress has been laid’.35 From a chemical point of view, there was nothing peculiar about the methods employed in the liquid tests for arsenic. Indeed, ‘reagents’, or substances known to react characteristically when placed in contact with another designated substance, were the principal tools by which chemists gained knowledge of the composition of unknown matter. ‘Chemical Tests or Re-agents’, in the words of the popular early nineteenth-century chemical writer Frederick Accum, ‘are called those substances, which, when applied to other bodies, the nature, or composition of which are unknown, quickly act upon them, and produce such changes as are sufficiently striking to the senses, and from which the quality or constitution of the unknown body may be inferred’.36 They are, he continues, ‘the compass by which the chemist steers’, the means by which he can use ostensibly fixed co-ordinates to navigate the chemical unknown.37 This model of chemical navigation, moreover, is one that is founded on experience and consensus—the relationship between signifier and signified (reagent and chemical body) is guaranteed by reference to rules laid down by a community of prac33 34 35 36 37
Smith (1825), pp. 136, 137. Paris and Fonblanque (1823), Vol. 2, p. 251. Ibid,. Accum (p. 49); emphasis original. Ibid., p. 50.
302
I. A. Burney / Stud. Hist. Phil. Sci. 33 (2002) 289–314
titioners. Reagents can perform this function because of an agreement among chemists that their actions are reliable and legible. Indeed, the recognized lexicon of chemical analysis is bounded only by instrumental needs of its users: ‘It may be readily conceived’, Accum observes, ‘that a vast variety of substances, provided their chemical actions be well established, may serve as Re-agents or tests ... By long search and experience, we have, however, learnt to make a choice of some particular bodies only, the effects of which are rapid, and the application of which requires no skill; and to these bodies the name of Re-agents or tests has been given by mutual consent’.38 The units of meaning of this consensually derived discourse, moreover, addressed themselves principally to sense perceptions. Reagents signified by ‘occasioning either a precipitate, a sensible cloudiness, a change of colour, and effervescence, or such other obvious alterations of properties, as experience has proved denote the presence, or absence, of certain bodies’.39 Accum thus underlines the conventional language of chemical demonstration, a collective form of communication by which agreed-upon tools are deployed for unveiling an occluded truth that is sought. Reagents are, in effect, the chemist’s equivalent of indirect, or circumstantial, evidence. In embracing color processes as a means of freeing themselves from the constraints of legal expectation, toxicologists were in another sense merely re-situating their evidence within another contested legal realm, one governed by the contestable rules by which ‘presumption’ was made to stand in for ‘fact’. They were therefore responsible not for producing a plain matter of fact, but for justifying a more complex system of evidentiary signs, one based, moreover, on indicators which, like smells in the Donnellan trial were, themselves subject to objection. Color, for instance, in the view of the eminent medical jurist and toxicologist Alfred Swaine Taylor, was notoriously slippery: ‘nothing is so deceitful as an absolute reliance upon colour in testing’, he warned in his canonical treatise On poisons. ‘Four persons may look at the same coloured product, and it will be found to present to each a different shade or tint’.40 In the face of these difficulties, chemists and toxicologists were acutely aware of the need to forge a medium of communication through which the sense experience upon which their proofs depended might be securely translated across individual observers. One method for achieving this in practice was to describe the chemical reactions that were in Accum’s words ‘sufficiently striking to the senses’ in what was itself a strikingly qualitative language. Descriptions depended on the recognition 38
Ibid., p. 53–54. Ibid., p. 52. 40 Taylor (1859), p. 206. Color, it should be noted, was a core feature in a much broader contemporary discussion about the problematic relationship between subjective and objective knowledge of things. In this discussion, grounded in the Lockean strand of British empiricism that also dominated early nineteenthcentury texts on legal evidence, color figured as an exemplary instance of (unstable) secondary sense perception. John Stuart Mill, for one, used color in his System of logic to illustrate his contention that what we commonly take to be real attributes of things are in fact less external realities than individually derived impressions. For Mill, testifying on the basis of sense perception was inescapably an exercise in conventional classification, one that ultimately relied upon forging a common descriptive language through which individual perception of difference might be regulated. 39
I. A. Burney / Stud. Hist. Phil. Sci. 33 (2002) 289–314
303
of highly specific shades of color—a purple comparable to the ‘bloom of an Orleans plum’, an orange similar to ‘the peel of a sweet orange’ or a ‘lively’ (as distinct from a ‘sad’) grass green, for example.41 These were often accompanied by an insistence that the effect was inescapably universal: reference to the ‘brilliance’ of a given reaction, which ‘once seen, is never forgotten’, was commonplace in contemporary chemical works. This recourse to a register of sheer intensity seems to follow from the conventional foundations of chemical language. If chemical communication is the result of a negotiation of signs that were in themselves empty, an emphatic form of description might enhance the quest for clarity and stability of expression. Thus the first set of problems that toxicologists faced when adducing proof through color reagents was at the level of recognition and representation: how could discriminations be made, and through what means could they be conveyed? There was a further problem. The ‘brilliance’ attributed to the various color tests, and its capacity to ground visual evidence in an emphatic language of effect, was predicated on an assumption of ‘purity’. In the discourse of chemical demonstration, that is, signs were crystal clear—beautiful, never murky, cloudy or muddy. It was this quality that seemed to guarantee the unmistakability of the tests, their capacity to circumvent subjective difference and appeal to a universal eye. But, as many toxicologists pointed out, the stuff of medico-legal inquiry was anything but ‘pure’. Robert Christison addressed himself to this issue in a lengthy contribution to the Edinburgh Medical Journal in 1827. In it he asked how, and to what extent, ideal laboratory conditions could apply to medico-legal realities. His answer, in short, was not very well. The difficulty, in Christison’s view, was that toxicologists in cases of suspected arsenic poisoning were most often given samples which were not only small in quantity but impure in composition. ‘It is well known’, Christison noted, ‘that the presence of various mineral vegetable and animal substances in fluids that contain arsenic, alters considerably the action of its liquid tests’.42 By way of illustrating this observation, Christison subjected the four liquid tests to the trials of the typical English stomach: broth, coffee and tea with sugar and cream, port and porter were all mixed with arsenic. The results took him far away from the language of stable signification and purity. In many instances, no precipitate was formed where arsenic was present, and conversely precipitate was formed where no arsenic existed. When precipitates formed, they were often ‘distorted’ in color, either by the effect of the color of the organic medium in which they were suspended, or by undergoing a chemical reaction with the medium. The language of purity thus fades into murkiness and confusion— colors being described as ‘very faint’, seen through ‘a haze’, or ‘pale and dirty’. Christison was also critical of recently announced laboratory techniques purporting to overcome these problems by chemically simulating purity. ‘It has been thought by some’, he observed, ‘that the inconveniences now mentioned may be removed by destroying the colour of the liquid in which the poison is dissolved; and in this expectation several decolorizing processes have been proposed by chemists, and
41 42
These descriptions are taken from Guy (1860–1861), p. 604. Christison (1824), p. 60.
304
I. A. Burney / Stud. Hist. Phil. Sci. 33 (2002) 289–314
adopted in all works of Medical Jurisprudence’.43 But these processes entailed their own distortions. For example, that proposed by the eminent French toxicologist Matthieu Orfila, involving the decolorization of the animal and vegetable infusions by the introduction of chlorine, failed on several grounds. The color, Christison warned, is for one ‘very seldom destroyed so entirely but that the precipitate caused by some of the tests still deviate to a certain degree from their characteristic tints; and although the colour of the fluid be even destroyed entirely, it often re-appears in the precipitates’. On the basis of such distortion, Christison concluded, processes such as Orfila’s were ‘generally useless, often detrimental, nay, sometimes even dangerous’.44 In 1836, impediments to the detection of small amounts of arsenic in mixed organic fluids (which resulted in this tension between color and reduction) were widely thought to have been resolved by the innovation of James Marsh, a chemist at the Woolwich Royal Arsenal. The famous ‘Marsh process’, together with the Reinsch process some five years later, seemed to mark a watershed in the regime of arsenic testing. Marsh’s process was based on the chemical interactions between arsenic, zinc and sulphuric or hydrochloric acid. These acids, when mixed with zinc, produced hydrogen gas, and this gas in turn exhibited a strong affinity for arsenic. Thus, when faced with a compound substance suspected of containing arsenic, Marsh recommended that the analyst mix it with the acid, then let the mixture act upon pure zinc. This would produce hydrogen gas to which the arsenic, if present even in minuscule quantities, would be transferred. Having collected the gas, the analyst was then to heat it, and let it stream onto a cold glass surface, the result of which would be the deposit of a thin film of metallic arsenic. Marsh’s process was immediately hailed as ‘beautiful’, and he was awarded the Society of Art’s purest accolade—its gold medal. This response was in a sense entirely appropriate, since through his method Marsh was in effect promising to reintroduce purity into the toxicologist’s lab. Yet if Marsh’s process seemed to overcome the challenge to a stable system of chemical demonstration posed by impure matter, commentators were soon warning about another challenge to toxicological signification. Henry Hough Watson, a Bolton chemist speaking before a meeting of the Manchester Literary and Philosophical Society in February 1841, argued that notwithstanding Marsh’s valuable contribution to the detection of arsenic, the process was itself subject to error. ‘Had it been the fact that arsenic is the only metal which enters into combination with hydrogen, and which is capable of being deposited upon cold surfaces, when the gas is allowed to undergo combination’, Watson observed, ‘we might with propriety have concluded that when, on adopting Mr Marsh’s plan, we happened to get a metallic deposit or crust, arsenic was present in the matter under examination’. But subsequent researches, Watson continued, revealed that other metals, most notably antimony, produced a gas which left deposits with strikingly similar characteristics to its arsenical counterpart, especially when the quantity of the metal in question was small. ‘Although a practised eye may
43 44
Ibid., p. 61. Ibid., p. 72.
I. A. Burney / Stud. Hist. Phil. Sci. 33 (2002) 289–314
305
discern some difference between the crusts, that from antimony being more silvery and metallic, yet the line of demarcation is not easily drawn; for a thin film of antimony looks like arsenic, and a thick crust of arsenic has the metallic appearance of antimony’.45 Marsh’s process was, in Watson’s view, underdetermined, the relationship it orchestrated between sign and thing being insufficiently secure to guarantee stable meaning.46 There was, to be sure, nothing unique in Watson’s emphasis on the problem of underdetermination as a general, structural feature of chemical language. Frederick Accum had himself warned that reagents were not altogether faithful to their conventionally allotted chemical partners. Closer to Watson’s period, the Guy’s toxicologist G. O. Rees observed that ‘chemistry abounds in instances in which the reactions of a single test are common to many elements’. This, in Rees’s view, placed a considerable but necessary burden on the toxicologist. ‘It is not only in relation to our positive knowledge that we must regard this question’, he concluded. ‘We must look to the chances of error from probable causes as yet unknown, and whose fallacious influence can only be avoided, or placed in extreme doubt, by having recourse to the observation of many phenomena produced by the action of several different tests on the substance we are examining—tests which, from their nature, are not subject to the same fallacies as those previously adopted; and whose concurrent testimony is therefore the more valuable’.47 Rees’s resolution of the disruptive effects of underdetermination was thus twofold. His first strategy was to advocate that the toxicologist have recourse to proof generated not through a single process, but to the corroborative effects of numerous different tests. This was a position that echoed many of the leading toxicological treatises of the day. Robert Christison, in his review of the relative merits of contemporary tests for arsenic, downplayed the dangers attributable to individual shortcomings of any one process by reference to their collective effect: ‘there is no single test on which thorough reliance can be placed; but [the] fallacies to which they are liable apply each to one test only. Hence, if each of the three re-agents, when applied with due care, gives a precipitate of the characteristic tint, the proof of the presence of arsenic is decisive’. This was the common-sense answer, in his view, to the
45
Watson (1842), pp. 595–596 and 597. Watson’s identification of a slippage between arsenical and antimonial traces in the Marsh test was particularly disruptive, from a medico-legal point of view, because antimony was a widely employed medicinal remedy for natural complaints such as cholera whose clinical symptoms mirrored those of arsenical poisoning. The framework of meaning established within the chemical lab, then, not only confronted difficulties within its own terms of reference, but found it difficult to retain its integrity when translated into other systems of classification—in this case a language of symptoms. Faced with this difficulty, chemists searched for and published dozens of ‘improvements’ to the basic Marsh test, but Watson, writing in 1841, remained unconvinced that the fundamental problem had been addressed by any of these ‘ingenious’ additions. Indeed, Watson cautioned, these additions, while leaving the practical problem of underdetermination largely in place, had introduced new dangers of technical complexity, complications which made the revised tests even less compatible with medico-legal requirements (ibid., pp. 598–599). 47 Rees (1841), pp. 166–167; emphasis added. 46
306
I. A. Burney / Stud. Hist. Phil. Sci. 33 (2002) 289–314
unnecessary controversies thrown up by identification of slippages within individual processes: ‘The antagonists of the tests have been content with proving how so many fallacies lie in the way of each, that no dependence can be put in any of them individually: They have not considered that the fallacies attached to one are obviated by the conjunct indications of the others’.48 In seeking refuge from slippages of toxicological demonstration through corroborative testimony, Rees, Christison and other toxicologists could find support amongst the leading legal theorists of the day. Best, in his treatise on circumstantial evidence, declared that ‘where several tests, based on principles totally distinct, are applied to different portions of a suspected substance, and give each the characteristic results of a known poison, the chances of error are indefinitely removed, and the proof of the existence of that poison in that substance comes short only of positive demonstration’.49 Such support from legal quarters might well be anticipated, since it accorded with a form of proof that contemporaries would have readily associated with the courtroom: the corroborative effect of disparate circumstances each of which operated independently from the others. What is worth noting, however, is the way in which this recourse to what Rees himself describes as ‘concurrent testimony’ threatened to undermine any claim to distinctiveness that might be made of expert testimony in poisoning trials. In other words, rather than being called as a special witness who had access to a form of evidence that was different in kind to ordinary testimony and that could be superimposed on the trial sui generis, the toxicologist was now in the position of constructing a proof within a logic strictly analogous to his counterparts in the law. Just as the advocate built his case by assembling a web of testimony that indicated but did not demonstrate the truth of his account, so too did the toxicologist call on disparate witnesses (reagents) to construct a coherent, but manipulated, discourse of proof. Rees’s second strategy, that of engaging in an on-going research program into possible sources of fallacy lurking unknown in the natural world, was in another sense damaging to a vision of expertise standing above the legal fray, and several commentators were quick to point out this threat. If toxicologists were to follow Rees’s stricture to ‘look to the chances of error from probable causes as yet unknown’, these critics warned, toxicologists would be doing nothing less than providing the advocate with ammunition in his professional quest to obscure. An anonymous reviewer of Taylor’s first edition of his classic On poisons captures this sense of peril:
48
Christison (1829), pp. 194–195. The Marsh test, Taylor agreed, should not to be used in isolation, but in corroboration with others: ‘The great object of chemical evidence is not to show a court of law what may be done by the use of one test only, by peculiar manipulations on imponderable traces, but to render the proof of the presence of poison in the substance examined most clear and convincing’ (Taylor, 1859, p. 356). 49 Best (1845), pp. 166–167. Wills similarly validated corroborative results: ‘The concurrence, moreover, of a plurality of characteristic tests, separately fallacious, but fallacious from different causes, may, in connection with strong moral facts, yield a result of so high a degree of probability as to be perfectly convincing, though the poison has not been reproduced’ (Wills, 1862, p. 233).
I. A. Burney / Stud. Hist. Phil. Sci. 33 (2002) 289–314
307
It seems to be the fashion to ransack the archives of chemistry, and when a substance can be found presenting a reaction common to it with the poison, to set it down as a fallacy, without considering the probability, or, in many cases, the possibility, of its really being in the way. When this is found laid down in a work of medical jurisprudence, it is at once pounced upon by lawyers, who, in profound ignorance of the real value of the so-called objections, bring it forth in crossexamination—sometimes with the effect of puzzling a medical witness … and sometimes unsettling the minds of an ignorant jury, by a pompous display of sources of fallacy which have no real existence, and thus indirectly encouraging the cold-blooded poisoner.50 In admitting an open-ended search for fallacy as a normal part of the toxicologist’s medico-legal responsibilities, in other words, expert testimony would be contaminated by the very contingency that lay at the heart of legal proceedings. The effect, like the recourse to corroborative testimony, would be that toxicology, rather than bringing stable knowledge to bear upon an otherwise contentious form of inquiry, would simply join the agonistic process of fact finding. Yet ultimately the toxicologist had to accept a degree of contingency, and the consequences that went with it. Here is how the London Medical Gazette summed up the precarious position of the contemporary toxicologist: The certain test of 1820 is no longer the certain test of 1840; and who can answer what this will be in 1860? Until chemistry becomes a fixed science, and the action of every possible combination of substances has been tried, how can we be sure of our facts, and confidently prove a negative? Every story, says the vulgar proverb, is good, till another is told; and every test is valid, till a fallacy is discovered in it.51 Pragmatic though this advice might be, in tying the processes of toxicological testing so closely to the realm of fallacy and story-telling, the London Medical Gazette simply highlighted the tension underlying the toxicologist’s appearance in the court of law: his promise of materiality and ‘matter of fact-ness’ on the one hand, and his need to construct a narrative out of signs which were themselves subjective, ephemeral, and open to dispute, on the other.
3. Testing testimony: concluding thoughts on the ‘legal laboratory’ In the above analysis I have argued that there are sound methodological and epistemological grounds for treating toxicological and legal modes of proof symmetrically. We have tacked between legal and chemical theory, and between the manipu-
50 51
Anon. (1848), p. 610. Anon. (1840–1841), pp. 410–411.
308
I. A. Burney / Stud. Hist. Phil. Sci. 33 (2002) 289–314
lation of words for things and the manipulation of things as words. By way of conclusion, I want to extend the homologies thus developed in a couple of ways, in order to refocus my initial observations about the commonplace opposition assumed between expert testimony and legal adversarialism. Firstly, consider the place of language in these two ostensibly distinct arenas. The assumption—or perhaps rhetorical position—on the part of experts is that it is only when they enter the courtroom that they have to contend with the vicissitudes of language. That is, in the laboratory, the expert works with, even manipulates, things, ‘plain matters of fact’, to which he has direct access, unmediated by the play of signs. I have tried to show how in the case of toxicological evidence at least, this is at best an oversimplification. But this point can be taken further: not only was the toxicologist by necessity constrained to act through a complex sign system, but he—and chemists more generally—explicitly cast this activity within a linguistic model. Chemistry, in other words, was routinely compared to a specific form of speech. Examples of this comparison can be found scattered throughout specialist texts and in popular representations of toxicology, but in my readings none is more striking than that developed by the internationally celebrated German chemist Justus Liebig. In 1844 Liebig delivered a set of lectures at the University of Giessen introducing students to the principles of analytical chemistry. Liebig’s views were guaranteed a substantial English readership through the auspices of the Lancet, which published the complete cycle of lectures, whilst on its editorial pages actively promoting them as exemplary chemical method. Liebig commenced his course with the following reflections: All our observations, taken collectively, form a language. Every property, every alteration which we perceive in bodies, is a word in that language. Certain definite relations are manifested in the deportment of bodies toward each other, a similarity in form or analogy in properties, or diversities in both respects. Such diversities are as numerous and various as the words of the most copious language, and they are no less varied in their signification and in the relations which they bear to our senses. The verbal meaning conveyed by the properties of bodies,—to pursue the illustration,—changes according to the mode in which these elements are arranged. As in all other languages, we have in that language whereby material bodies hold converse with us, articles, substantives, and verbs, with their variations of cases, declensions, and conjugations. We have also many synonyms. The same quantities of the same elements produce a poison, a remedy, or an aliment, a volatile or a fixed body, according to their manner of arrangement. ‘This’, Liebig concluded, ‘is CHEMICAL ANALYSIS’.52
52
Liebig (1844), p. 5.
I. A. Burney / Stud. Hist. Phil. Sci. 33 (2002) 289–314
309
It would take far too much space to do full justice to this extraordinary statement. For now, I will content myself with extracting several of Liebig’s observations that pertain to the workings—or, perhaps better, concerns about possible mis-workings— of toxicological demonstration. Chemical language is a visual one, Liebig notes at the outset, one based on perceptible changes in chemical forms. These forms, or words, exist in structural, grammatical relation to one another, and it is a knowledge of this grammar that is the principal tool of the chemical analyst. A full appreciation of differences between ostensibly similar units of meaning, moreover, is of critical importance, an observation that Liebig illustrates through the familiar toxicologically based slippages between the “synonymous” phenomena of poison, medicine and disease. The importance of Liebig’s formulation for this essay, then, is as an index of the self-conscious linguistic sensibilities at work in the practice of chemical analysis. Through it we can, instead of seeing the toxicological witness as entering into the perilous use of language for the first time when testifying in court, reframe his intervention as an attempt to translate between different, rule-bound, and highly stylized modes of signification. This brings us to the second concluding observation: the linguistic sensibilities underlying chemical analysis not only allow us to reposition the toxicologist as an expert necessarily interested in the manipulation of signs, but to place this expert in comparative relation to what can be regarded as another, competing expert who, like the toxicologist, conducts tests on raw matter in a conventionalized space in order to establish some underlying reality that can be brought forward as evidence. This alternative expert is the trial advocate, his means of testing is cross-examination, his space is the open court, and his analytical object is verbal testimony. In order to appreciate the depth of this analogy, we must return briefly to developments in the conceptualization of evidence in English felony trials during the period we have been considering. Modern legal scholars agree that what we now recognize as the adversarial framing of these proceedings emerged over the course of the eighteenth and early nineteenth centuries. What J. M. Beattie describes as the ‘old’ form of felony trial, elements of which survived up to the beginning of the nineteenth century, was distinct in several key and interrelated respects. It was short—even a complex case involving a charge of murder would rarely be expected to last more than a few hours. Counsel was typically absent, parties to the trial presenting their cases without legal assistance. It involved little in the way of formally introduced testimonial evidence by the parties concerned, beginning not with a synthetic and persuasive overview of the arguments to be adduced, but with a simple statement from the injured party. It was the judge (and to a much lesser extent the jury) who was responsible for developing testimony, acting as examiner and cross-examiner, and determining for himself when sufficient evidence had been adduced.53 Moreover, as Stephen Landsman argues, the type of evidence most actively sought by this
53 Beattie (1986), Ch. 7, esp. pp. 340–378. See also two classic studies: Langbein (1978) and Green (1985).
310
I. A. Burney / Stud. Hist. Phil. Sci. 33 (2002) 289–314
inquisitorial examination was not discursive or dialogic but static, trials predominantly being ‘a search for a determinative piece of documentary proof rather than the examination of evidence in open court’.54 This framework had begun to erode by the early to mid eighteenth century, gradually replaced by an emerging ‘adversarial’ system. For our limited purposes, two related features of this new procedure are noteworthy: the emphasis on carefully scrutinized oral testimony as the evidentiary focal point of trials, and the increasingly pivotal role played by advocates in developing, presenting and examining such evidence to a more passive judge and jury according to increasingly strict rules of admissibility. While the search for documentary evidence absorbed most of the attention of the first acknowledged systematizer of English evidentiary law, Geoffrey Gilbert, by the mid eighteenth century commentators such as William Blackstone (the writer commonly used by legal historians as a barometer of procedural change) had begun to emphasize the critical interrogation of evidence provided by witnesses physically present in court. It was this, in Blackstone’s view, that provided the best means to ‘sift out the truth’.55 Cross-examination was, as the nineteenth-century evidentiary theorist Thomas Starkie observed, ‘one of the principal tests which the law has devised for the ascertainment of truth’.56 Moreover, in theory and in practice, this adversarial examination increasingly came to be seen as the responsibility of a new courtroom figure—the advocate. Under the ‘old’ trial regime the judge was supposed to act as the principal advocate, charged not only with developing the evidence but also with making sure that it was fairly presented, especially as far as the defendant was concerned. But by the second half of the eighteenth century, and in the context of criticism directed against judicial failure to protect defendants in political trials, the notion of an organized ‘defence’, professionally conducted and designed to cast doubt on the validity of the factual case presented against the defendant, had emerged. This in turn put an increased burden on the presentation of a tight prosecution ‘case’, and from this dynamic the contest between two well developed and opposing presentations of evidence became the core and emblematic feature of the English criminal trial. Indeed, as Landsman argues, by the beginning of nineteenth century ‘cross-examination of witnesses by
54 Landsman (1990b), pp. 592–593. While it is true, as Barbara Shapiro points out, that witness testimony played an increasingly important role in jury trials from at least the sixteenth century, this raised numerous thorny questions (concerning, for example, witness credibility, and the relationship of written to oral evidence) that remained contentious well into the eighteenth century. Gilbert was certainly suspicious of oral testimony, as it depended on memory and possible witness partiality, and his preference for documentary evidence might reflect a preference for the Romano-canonical style of legal reasoning, whereby each piece of evidence had a fixed value within a hierarchical conception of proof. The following paragraphs on the growth of oral testimony and its examination in the first decades of nineteenth century are indebted to Landsman (1990a), esp. pp. 591–603, and Landsman (1990b), esp. pp. 1160–1186. 55 Blackstone (1765–1769), cited in Landsman (1990b), p. 591. 56 Starkie (1833), Vol. 1, p. 160. The object of cross examination, Edmund Powell agreed, was ‘to sift, detect, and expose’ (Powell, 1859, p. 442).
I. A. Burney / Stud. Hist. Phil. Sci. 33 (2002) 289–314
311
skilled counsel was of such importance that the process was rendered suspect without it’.57 The interactive, adversarial encounter between advocate and live witness performed within a rule-bound courtroom space, then, was the modern way to ‘test’ for legal truth. But if counsel was skilled in the performance of a crucial test, how did this test operate, and in what did this special skill inhere? On the one hand, cross-examination tested the coherence and consistency of testimony. In searching out suppressed presumptions in a seamless narrative, it exposed hidden gaps in testimony. But it was not merely manifest speech that was the object of legal ‘testing’. Witness testimony was a composite entity, composed of several essential elements that could be subject to analysis. Legal treatises, reflecting the growing importance attached to the skilful examination of such testimony, urged advocates to attend to signs of an underlying truth obscured by the crude, raw matter of narrative intent. Witness credibility, Starkie advised, could be judged according to the manner in which they related their testimony (were they overzealous, evasive and affecting indifference, or prompt and frank?), and in his view it was the ‘viva voce examination’ that best elicited this evidentiary indicator.58 Jeremy Bentham, the most committed exponent of trials as live, public and adversarial contests, was in full agreement: indeed, as far as Bentham was concerned, the indirect signs elicited from cross-examination were one of the few categories of evidence that could be classified as ‘real’. ‘How instantaneously the points of agreement and disagreement are brought to view!’, he enthused. ‘How instructive is the deportment exhibited on both sides on the occasion of such a conference!’ 59 To get at this level of truth existing beneath the surface, the advocate’s principal mode of testing was to subject witnesses to what was commonly described as ‘severe’ interrogation. Late eighteenth-century attorneys such as William Garrow and Thomas Erskine rose to celebrity status on the basis of their ability to subject witnesses to an intense, if carefully calibrated, verbal examination.60 In effect, skilled advocates
57
Landsman (1990b), p. 599; emphasis original. Starkie (1833), Vol.1, p. 481. This emphasis on deportment, as Barbara Shapiro’s work demonstrates, was by no means new. The seventeenth-century jurist Matthew Hale valued oral testimony because ‘the very Manner of a Witness’s delivering his Testimony will give a probable Indication whether he speaks truly or falsely’, while in his Principles of the law of evidence, with elementary rules for conducting the examination and cross-examination of witnesses (1849), William Best cites the canonist Lancellottus to bolster his discussion on the value of this form of scrutiny afforded by adversarialism. See Shapiro (1983), pp. 195–197. What does seem new in Starkie and his contemporaries, as Landsman suggests, is the emphasis on skilled cross-examination as an integral feature of trial proceedings. 59 Bentham (1827), Vol. 2, p. 470; emphasis original. ‘The evidence, and the only evidence, which cannot lie’, he suggests, ‘is that which, without the intervention of any human testimony, presents itself directly to the senses of the judge. In this case it is real evidence; and such involuntary evidence as is exhibited by the deportment of a party or an extraneous witness while undergoing the process of interrogation’ (Bentham, 1827, Vol. 3, p. 250). 60 Landsman describes Garrow as the apotheosis of adversarialism, whose methods of cross-examination were at once ‘contentious and clever’, and ‘cruel’ (Landsman, 1990b, p. 564). Contemporary evidence supports such a characterization: indeed, as J. M. Beattie has found, their testing courtroom performances were immortalized in satirical sketches such as Rowlandson’s ‘Being nervous and cross-examined by Mr. Garrow’ (Beattie, 1991, p. 247). 58
312
I. A. Burney / Stud. Hist. Phil. Sci. 33 (2002) 289–314
succeeded in their testing operations by transforming the courtroom into an experimental space in which, by regulating key variables—notably the ‘temperature’ at which their objects of analytical scrutiny were constrained to react—pure forms might be set free. Garrow, Erskine and their peers were experts, equipped with their own battery of ‘reagents’, charged with the task of decomposing raw compounds into the elemental units of truth sought in trial. What is more, and what brings us back to the discomfited scientific witness we met at the very beginning of this article, the testimony most in need of this form of exacting legal test, in the view of evidence theorists, was that delivered by the scientific expert. Contemporary legal treatises voiced particular suspicion of expert evidence: ‘Perhaps the testimony which least deserves credit with a jury is that of skilled witnesses’, John Pitt Taylor warned in his 1848 Law of evidence, adding that ‘they do not, indeed, wilfully misrepresent what they think, but their judgments become so warped by regarding the subject in one point of view, that, even when conscientiously disposed, they are incapable of expressing a candid opinion’. It was for this reason that Thomas Starkie, after explaining the value of cross-examination as a means of allowing those present in the court to judge the reliability of testimony on the basis of more than words alone, issued this simple warning: ‘These observations apply with peculiar force to all questions of skill and science’.61 The newly constituted expert advocate, then, had his sights trained on his scientific counterpart. With this we have arrived full circle. According to the Pharmaceutical Journal, the scientific expert’s plain matter of fact was put at risk by the play of signs orchestrated by the advocate—the veritable antithesis of true expertise. We have seen, on the one hand, that the toxicological expert was himself necessarily adept in orchestration, that his testimony was the end product of a complex and contestable process through which conventionally derived signifying tools (reagents) were activated to indicate a hidden presence. Moreover, through an analysis of the tools employed by the advocate (presumption and cross-examination) we have seen how this figure, rather than embodying ‘anti-expertise’ as is conventionally assumed, might instead be cast as an alternative expert, one who operates not incommensurably but in parallel ways to his scientific foil. By treating the process of expert fact-making in the symmetrical terms that this essay has advocated, then, we might develop a framework for understanding the agonistics of the courtroom that allows us to cut through the analytically restrictive law–science opposition that has for so long held sway.
61 Taylor (1848), pp. 54–55; emphasis original; Starkie (1833), Vol. 1, p. 482. Nor, it should be noted, was this validation of testing expertise through cross-examination a position held exclusively by law scholars: the medico-legal writers Paris and Fonblanque observed that ‘there is a natural propensity in human nature, from which the most honourable minds are not free, to view all questions through the medium of some preconceived opinion; in law and in science it is too often apparent. Hence our law has wisely contrived its modes of viva voce examination, in which the judge, the jury, and the counsel, on both sides, are equally empowered to sift the truth, and thus counteract the leaning which any witness may be supposed to have towards the party producing him’ (Paris & Fonblanque, 1823, Vol. 1, p. 163).
I. A. Burney / Stud. Hist. Phil. Sci. 33 (2002) 289–314
313
Acknowledgements I would like to thank the following people for their comments on earlier drafts of this essay: Mario Biagioli, Roberta Bivins, Tom Green, Andy Hammond, Rebecca Herzig, Jeff Hughes, David Knight, Martin Kusch, Sandra Mols, John Pickstone and Joan Scott. I also thank the Wellcome Trust for its support of my research on the history of criminal poisoning.
References Accum, F. (1818). A practical treatise on the use and application of chemical tests (2nd ed.). London: Thomas Boys. Anon. (1840–1841). Some difficulties in forensic medicine. London Medical Gazette, 1, 410–411. Anon. (1848). Mr Taylor on poisons. Monthly Journal of Medical Science, 8, 609–614. Anon. (1853–1854). Chemical truth. Pharmaceutical Journal, 13, 150–151. Bartrip, P. (1992). A ‘pennurth of arsenic for rat poison’: The Arsenic Act, 1851 and the prevention of secret poisoning. Medical History, 36, 53–69. Beattie, J. M. (1986). Crime and the courts in England, 1660–1800. Princeton: Princeton University Press. Beattie, J. M. (1991). Scales of justice: Defense counsel and the English criminal trial in the eighteenth and nineteenth centuries. Law and History Review, 9(2), 221–267. Bentham, J. (1827). Rationale of judicial evidence. London: Hunt & Clarke. Best, W. M. (1845). A treatise on presumptions of law and fact, with the theory and rules of presumptive or circumstantial proof in criminal cases. Philadelphia: T. & J. W. Johnson. Brande, W. T. (1827–1828). Lectures on chemistry, delivered at the Royal Institution of Great Britain. Lancet, 2, 64–69. Burney, I. A. (1999). A poisoning of no substance: the trials of medico-legal proof in mid-Victorian England. Journal of British Studies, 38(1), 59–92. Christison, R. (1824). On the detection of minute quantities of arsenic in mixed fluid. Edinburgh Medical Journal, 22, 60–83. Christison, R. (1829). On poisons. Edinburgh: Adam Black. Danger, F., & Flandin, C. (1841). De l’Arsenic suivi d’une instruction propre a´ servir de guide aux experts dans les cas d’empoisonnement. Paris: Bachelier. Faraday, M. (1829). Chemical manipulation: Being instructions to students in chemistry on the methods of performing experiments of demonstration, or of research, with accuracy and success. London: John Murray. Golan, T., & Gissis, S. (Eds.). (1999). Science and law, special issue. Science in Context, 12, (1). Golinski, J. (1992). Science as public culture: Chemistry and Enlightenment in Britain, 1760–1820. Cambridge: Cambridge University Press. Green, T. A. (1985). Verdict according to conscience. Chicago: University of Chicago Press. Guy, W. (1860–1861). On the colour tests for strychnia. Pharmaceutical Journal, 2, 558–606. Hacking, I. (1975). The emergence of probability. Cambridge: Cambridge University Press. Howell, T.B. (Ed.). (1809–1826). Complete collection of state trials. London: R. Bagshaw, Longman. Jasanoff, S. (1992). What Judges should know about the sociology of science. Jurimetrics, 32, 345–359. Jasanoff, S. (1995). Science at the Bar. Cambridge, MA: Harvard University Press. Jasanoff, S., & Lynch, M. (Eds.). (1998). Contested identities: Science, law and forensic practice, special issue. Social Studies of Science, 28 (5–6). Knight, D. (1999). Seeing and believing in chemistry. In Interface 4: Video Ergo Sum (pp. 181-193). Hamburg: Hans-Bredow Institut. Landsman, S. (1990a). From Gilbert to Bentham: The reconceptualization of evidence theory. The Wayne Law Review, 36, 1149–1186.
314
I. A. Burney / Stud. Hist. Phil. Sci. 33 (2002) 289–314
Landsman, S. (1990b). The rise of the contentious spirit: Adversary procedure in eighteenth-century England. Cornell Law Review, 45, 497–609. Langbein, J. (1978). The criminal trial before the lawyers. University of Chicago Law Review, 45(2), 263–316. Liebig, J. (1844). Lectures in organic chemistry delivered during the winter session, 1884, in the University of Giessen. Lancet, 1, 3–7. Lofft, C. (1791). The law of evidence, by Lord Chief Baron Gilbert, considerably enlarged by Capel Lofft, barrister at law. London: A. Strahan & W. Woodfall. MacNally, L. (1802). The rules of evidence on pleas of the Crown. Dublin: J. Cooke. Male, G. E. (1818). Elements of juridical or forensic medicine, 2nd ed. London: E. Cox & Son. Paris, J. A., & Fonblanque, J. S. M. (1823). Medical jurisprudence. London: W. Phillips. Phillipps, S. M. (1815). The theory of presumptive proof; or, an inquiry into the nature of circumstantial evidence: including an examination of the evidence on the trial of Captain Donnellan. London: W. Clarke & Sons. Powell, E. (1859). The principles and practice of the law of evidence, 2nd ed. London: John Crockford. Rees, G. O. (1841). On the existence of arsenic as a natural constituent of human bones. Guy’s Hospital Reports, 6, 163–171. Roberts, L. (1995). The death of the sensuous chemist: The ‘new’ chemistry and the transformation of sensuous technology. Studies in History and Philosophy of Science, 26(4), 503–529. Roberts, P., & Willmore, C. (1993). The Royal Commission on Criminal Justice: The role of forensic science evidence in criminal proceedings. London: HMSO. Schaffer, S., & Shapin, S. (1985). Leviathan and the air-pump. Princeton: Princeton University Press. Shapiro, B. (1983). Probability and certainty in seventeenth-century England. Princeton: Princeton University Press. Shapiro, B. (2000). A culture of fact: England, 1550–1720. Ithaca: Cornell University Press. Smith, J. G. (1825). An analysis of medical evidence, comprising directions for practitioners, in the view of becoming witnesses in courts of justice; and an appendix of professional testimony. London: Thomas & George Underwood. Starkie, T. (1833). A practical treatise of the law of evidence and digest of proofs, in civil and criminal proceedings 2nd ed. London: J. & W. T. Clarke. Taylor, A. S. (1844). Manual of medical jurisprudence. London: J. Churchill. Taylor, A. S. (1859). On poisons in relation to medical jurisprudence and medicine 2nd ed. London: J. Churchill. Taylor, J. P. (1848). The law of evidence as administered in England and Ireland. London: A Maxwell & Son. Thompson, A. T. (1836–1837). Lectures on medical jurisprudence, at the University of London. Lancet, 2, 448–457. Watson, H. H. (1842). On detecting the presence of arsenic. Memoirs of the Literary and Philosophical Society of Manchester, 6, 590–615. Welch, A. (1992). Strong representations: Narrative and circumstantial evidence in England. Baltimore & London: Johns Hopkins University Press. Wills, W. (1862). An essay on the principles of circumstantial evidence by the late William Wills, 4th ed. (A. Wills, Ed.). London: Butterworths. First published 1838.