The low template DNA profiling controversy: Biolegality and boundary work among forensic scientists

Exploring LT-DNA

LT-DNA itself represents a series of relatively minor, incremental adaptations to ‘conventional’ DNA profiling technology. It also deploys a probabilistic Bayesian system that is used in other areas of forensic practice. However, the international community of forensic genetic experts has a divided opinion of LT-DNA. Certain prominent critics of these ultra-sensitive techniques have upheld the power and certainty of conventional DNA profiling (Budowle et al., 2000). Subsequent scientific claims made in the scientific literature by the developers of LT-DNA methods have perpetuated further debate about the status of these particular techniques.

In the course of my research, I came to regard the following empirical questions as having priority:

Forensic science occupies an awkward position at the law–science nexus. Previous studies have demonstrated the complications with separating the ‘scientific’ character of forensic evidence from the originating context of criminal investigation. This kind of research has shown how the boundaries between various distinctions (‘scientific/non-scientific’) or specific categories (‘expert’, ‘technician’, etc.) are often contingently (and controversially) shaped in courtroom settings (Jasanoff, 1998; Lynch and McNally, 2003). Given the complexities faced by actors who draw these distinctions, many studies of forensic science have reflected a typical STS ‘sensibility to the messy practices of relationality and materiality of the world’ (Law, 2009: 142). The question of how biolegality functions to stabilize this potentially ‘messy’ clash of epistemic cultures became a key guiding theoretical orientation for such studies.

Following previous studies, my research adopted a methodological perspective that focuses on the interactional processes through which actors ‘build up and break down’ the credibility of LT-DNA (Lynch et al., 2008: 2–3). The construction of certain distinctions featured prominently in those processes. Hence, this study, to an extent, follows previous emphases on the kinds of ‘boundary work’ (Gieryn, 1983) that actors employ to make sense of the place of science in legal environments. What became clear, however, is that, in this case, the nature of the specific ‘boundaries’ often were more complex than previous studies on the topic have observed.

The initial stage of my research involved familiarization with LT-DNA methods and their uses in forensic investigation. This preliminary research included an examination of relevant documentary sources, ranging from peer-reviewed articles published in scientific journals, to court reports, Government commissioned reports and statements, news media, open correspondence and websites. The journal Forensic Science International: Genetics (FSI:G) was a particularly fruitful source. I also consulted a number of FSI:G articles, which included several exchanges between developers of LT-DNA methods and their scientific critics. I coded these as particular instances of boundary work.

I supplemented this documentary analysis with a small number of in-depth qualitative interviews (15 in total) with forensic scientists and other stakeholders based in the United Kingdom and United States. I also attended and participated in professional and academic seminars, which provided a further opportunity to engage with the arguments and discursive strategies surrounding the contested status of LT-DNA. This fieldwork enabled me to update and clarify knowledge of LT-DNA and to corroborate the nature of the deliberations that surround this technology.

The following sections discuss a number of key legal and regulatory interventions that have played significant roles in shaping understandings of ultra-sensitive DNA profiling techniques. I start with a brief history of LT-DNA as used in criminal investigations and scrutinized in courtroom proceedings and by regulators. This allows me to introduce some of the technical issues associated with LT-DNA, in order to outline the legal and scientific contours of the controversy surrounding this technology. I go on to discuss a series of unstable technological boundaries in the controversy, occasioned by persistent questions in the forensic community about the scientific efficacy of LT-DNA and the associated methods. A later section describes a candid series of exchanges between developers of LT-DNA technology and sceptics, published in FSI:G. This debate reveals the particularly complex forms of boundary work involved in shaping the precise nature of ‘expertise’ considered appropriate to assess the efficacy of LT-DNA.

LT-DNA: key moments in its forensic history ¹

The modern history of the forensic use of DNA is generally accepted to have begun in the early 1980s, following discoveries by Sir Alec Jeffreys at the University of Leicester (Gill and Buckleton, 2004: 2). Interest in extremely minute quantities of DNA developed as scientific methods and instrumentation improved and as police and policymakers became more attuned to the investigative potential of DNA. Aided by academic collaborators, forensic scientists developed techniques that they claimed could produce profiles from material containing DNA from a single human cell (Findlay et al., 1997), possibly deposited by touch alone (Lowe et al., 2002).

The UK FSS was instrumental in introducing this approach into criminal investigative work. The method that the FSS originally termed ‘low copy number’ (LCN) DNA analysis was first introduced into casework in 1999 and was presented as evidence over 40 times in UK criminal cases in the first 8 years of its use (Interview, US scientist, 2007). LCN, along with other ultra-sensitive methodological variants developed by forensic science providers, is now more commonly included under the term LT-DNA profiling. LT-DNA is used to collectively refer to the repertoire of methods used to produce profiles from under 200 pg of starting material (Caddy et al., 2008: 5). The recent use of the word ‘template’ reflects an emphasis on the low starting quantities of biological material in general, rather than the assumption of low amounts of DNA.

This innovation quickly became associated with the successful resolution of so-called cold case reviews: serious cases that had remained unsolved for several years (FSS, 2005). Ultra-sensitive DNA profiling techniques played an instrumental role in the pursuit of a number of systematic cold case review projects conducted by the English police forces between 2002 and 2006. Operation Phoenix, and Operations Advance 1 and 2 involved the use of LT-DNA methods to attempt to recover DNA profiles from fragments of evidence retained from unsolved cases. In cases where LT-DNA yielded profiles, these were subsequently searched on the NDNAD, with a small proportion returning successful matches (FSS, 2005; Home Office, 2005).

LT-DNA was also used with success in jurisdictions outside the United Kingdom. In 2003, Swedish authorities requested the FSS to carry out LT-DNA profiling on a knife used in the murder of the Swedish foreign minister Anna Lindh. The subsequent profile contained a mixture of Lindh’s DNA and that of suspect Mijailo Mijailovic, which helped to bring the latter to justice. LT-DNA evidence also assisted in securing the conviction of Bradley Murdoch for the murder of British tourist Peter Falconio in the Australian outback in 2006.

Through these and other investigative successes, low-template methods became regarded as a cutting-edge technological development in the fight against crime. This image received a dramatic challenge, however, following the verdict in the case R v Hoey (2007: NICC 49), announced in December 2007. Sean Hoey was charged with 56 criminal charges in relation to his alleged role in the bomb attack, which took place in Omagh, Northern Ireland, in August 1998. Twenty-nine of the charges were counts of murder corresponding to the number of victims of the attack. Although the prosecution alleged that LT-DNA evidence indicated a match between Hoey and genetic material recovered from alleged bomb components, Hoey was found not guilty on all charges. ²

The failure to convict Hoey in the Omagh case brought LT-DNA evidence under the full glare of the media spotlight. In his judicial ruling, Judge Weir focused on two sets of concerns about the reliability of the technology. ³ The first concerned allegations of serious shortcomings in the way the DNA evidence had been recovered, packaged, stored and transported. The second concerned more fundamental questions of scientific opinion on the validity of the method. Judge Weir concluded that LT-DNA had not been appropriately validated by the scientific community (R v Hoey, 2007: para. 64). In his view, two articles published by the developers of the FSS method were insufficient to validate the technique. Weir accepted the Defence argument, which included references to the relative lack of uptake of LT-DNA in certain jurisdictions, and the lack of international agreement on the validation procedures. He also noted that two prosecution experts gave variable testimony about the technique. Weir contrasted the wide variance in standards for LT-DNA to the established validation guidelines for conventional DNA tests (R v Hoey, 2007: para. 62).

Defence witnesses also argued that LT-DNA data risked a far greater level of subjective interpretation than conventional forensic DNA analysis. The Court heard further criticisms related to the reproducibility of the results. In an experiment performed at an FSS laboratory, three LT-DNA tests had been used on the same sample. The consensus results obtained from the first two tests were contradicted by the third test. ‘Thus the normal approach used in the United Kingdom had unintentionally been demonstrated by its own proponents to be potentially (and in that particular instance actually) misleading’ (R v Hoey, 2007: para. 62). R v Hoey also dramatically exposed considerable differences of opinion about the reliability and validity of LT-DNA among the senior forensic scientists who testified in the case. In ruling that LT-DNA evidence was inadmissible in this case, Judge Weir used it to exemplify his more general concern about current procedures for validating scientific techniques for producing legal evidence.

The case verdict had notable repercussions for the criminal justice system. Following the collapse of the case, the Association of Chief Police Officers (ACPO) of England, Wales and Northern Ireland ordered a temporary suspension of the use of LT-DNA in criminal investigation, with police forces in Scotland following suit. The Crown Prosecution Service also announced a review of all cases involving LT-DNA evidence. The judgement resonated beyond the United Kingdom: in Australia, the verdict led to calls for a review of the Falconio murder case (Murdoch, 2007).

Concerns about LT-DNA had been voiced prior to the Hoey verdict (Buchanan, 2007). A review by England’s Forensic Science Regulator had been commissioned prior to the case, although the findings were published a few months following the Hoey judgement (Caddy et al., 2008). The report team, headed by Professor Brian Caddy of Strathclyde University, addressed a number of technical issues relating to sample recovery and extraction, quantification and interpretation methods, validation procedures and the place of LT-DNA in the criminal justice system as a whole. The report concluded that LT-DNA techniques were scientifically ‘robust’ and that they were ‘fit for purpose’ for forensic use (Caddy et al., 2008: 1). The report did, however, recommend further harmonization of standards for the production and interpretation of LT-DNA data.

The reaction to the conclusions of the Caddy report varied within the forensic scientific community. In some quarters (e.g. by presenters at a practitioner conference I attended in 2008), the report was hailed as providing a decisive endorsement of the technique, but it drew vehement criticism elsewhere (Gilder et al., 2009; Jamieson and Bader, 2008). Critics argued that the report did not fully address many of the issues arising from the R v Hoey verdict, such as the lack of international agreement, and that it did not acknowledge dissention within the UK scientific community itself over LT-DNA. They also claimed that the report paid insufficient attention to scientific issues concerning the way in which LT-DNA profiles were interpreted for possible matches. ⁴ Significantly, even though the Caddy report expressed satisfaction with the method through which LT-DNA had been validated, critics of the report noted that it failed to reproduce any actual validation data.

Consequently, the Caddy report represented only a partial resolution. The report’s methods for assessing LT-DNA, as well as its conclusions, remained a source of considerable dispute. In attempting to settle the reliability and validity questions surrounding LT-DNA, the report actually fomented debate among forensic scientists. Critics maintained that ‘the review [raised] important issues about what it [meant] for a forensic science technique to be validated’, amid concerns over the way LT-DNA profiles had been interpreted (Gilder et al., 2009: 535).

Issues relating to LT-DNA were again raised during the appeal of the murder case R. v. Reed (2009). Brothers David and Terence Reed had been convicted in 2007 of stabbing their associate Peter Hoe to death in his house in Eston in North East England. The Crown alleged that two pieces of plastic found at the crime scene were parts of knife handles, and that LT-DNA analysis of bodily material on them yielded the DNA profiles of both of the accused brothers. The Reeds appealed on the basis that the evidence reported by forensic scientist Valerie Tomlinson was inadmissible. Tomlinson had asserted that the DNA was deposited onto the knife handles through direct contact (primary transfer), and she argued against the possibility that it had been transferred through indirect contact. ‘Indirect’ contact would occur when the DNA was deposited through the actions of a third party (secondary transfer) or through an even more distant party (tertiary transfer) (R. v. Reed, 2009: para. 16). The defendants argued that Tomlinson was unable to make reliable assertions about the mode of transfer. Their appeal sought to challenge ‘the admissibility of evaluative evidence of the possible ways in which the DNA was transferred … on the basis that the scientific basis is insufficiently reliable’ (para. 114 (iii)).

Bruce Budowle, a leading US forensic scientist and staunch defender of conventional DNA profiling methods, was called as a witness for the Defence (this was an unusual role for Budowle, who previously had worked for decades with the Federal Bureau of Investigation (FBI)). He testified that the science of DNA transfer had not developed sufficiently to reliably determine the source of DNA material, and whether it resulted from primary, secondary or tertiary transfer. Budowle opined that it was impossible for forensic scientists to draw conclusions on the method and timing of DNA transfer. Professor Allan Jamieson, who had criticized LT-DNA for the defence in R v Hoey, was also called by the Reeds’ counsel and reiterated his strong misgivings regarding ‘the present state of the validation of the science and methodology associated with the [LT-DNA] process and in consequence its reliability as an evidential tool’ (R. v. Reed, 2009: para. 6).

The Reed Appeal Court dismissed the claims of both these expert witnesses. The Court rejected Budowle’s assertion that forensic scientists were not qualified to make inferences about possibilities of transfer and ruled instead that it was acceptable for someone such as Tomlinson to seek ‘reference to her experience’ (R v Reed, 2009: para. 121). The judgement against Jamieson’s testimony constituted a significant reversal. Although the Appeal Court justices accepted that Jamieson had ‘written some peer reviewed papers and carried out academic work in areas of forensic science’ (para. 105), they inferred that he had very little practical background in the interpretation of low-template profiles. In support of this determination, the Court cited evidence that Jamieson had conducted no laboratory research of his own on LT-DNA, instead basing his knowledge of DNA analysis on ‘papers and discussion with other scientists’ (para. 106). The Court concluded that despite acquiring a ‘degree of experience’ of LT-DNA from testifying in cases involving these techniques and from discussions with scientists, his expertise did not match that of the casework experience acquired by Tomlinson (para. 110).

Defence witnesses essentially argued that the assertions made by Tomlinson were based on subjective past experience rather than formal scientific method and that she overstepped a scientist’s remit for assisting the Court. However, the Court regarded Tomlinson’s testimony, despite its less formalized epistemological basis, as relevant and therefore admissible. Tomlinson’s practical know-how was considered superior to Jamieson’s abstract, theoretical claims regarding how scientific methods should be validated.

In subsequent discussions to which I had access, lawyers voiced concerns about some of the precedents established by R v Reed. These related to a perception that expert witnesses had gained more influence at the expense of legal counsel. One practicing barrister claimed that when conducting a defence, advocates were now becoming mere mouthpieces for the expert opinions that the defendants saw as more helpful to their case, which precluded proper legal scrutiny (seminar discussion, Criminal Justice Barrister 1, 2011). Following Reed, experts are now able to evaluate and enumerate a number of different possibilities concerning the probativity of evidence in court proceedings. They regard this situation as impinging on the preserve of counsel, with the risk of diminishing their traditional roles and powers. One barrister regarded the adversarial system as coming under a profound existential threat, as he opined that there was a ‘feeling amongst lawyers that we are moving toward an inquisitorial system’ (seminar discussion, Criminal Justice Barrister 2, 2011).

This brief overview suggests that further questions accompany debates over the validity of LT-DNA. These questions concern the allocation of specific roles at the law– science nexus, namely, which parties are best suited to making judgements about this technology, and with what kind of epistemological warrant. The legal system has exhibited a fluid posture toward scientific debates, with some courts accepting critical testimony about LT-DNA in some cases while others reject it. In the case of Reed, the Appeal Court made a significant intervention when it favoured Tomlinson’s personal experience over Jamieson’s theoretical, abstracted claims. By doing so, the Appeal Court shaped its own ordering of scientific epistemology, favouring personal knowledge over a more academic account. On the other hand, this decision was seen by some commentators as advancing the authority of scientific ‘experts’ over legal professionals. Here, the dispute over LT-DNA becomes an instance of ‘legal metascience’ (Lynch et al., 2008: 47) in action, with questions about technology, and the resultant outcomes, holding implications not only for the way in which law scrutinizes science, but also how the latter interacts with the former. Further dimensions characterize this controversy, as discussed in the next section, which provides a more in-depth exploration of the technical aspects of LT-DNA in order to highlight disputed forms of technological ‘closure’.

The technical ‘distinguishing work’ of LT-DNA interpretation methods

The ultra-sensitive technique originally developed by the FSS differed from the conventional DNA profiling, in that it employed an increased number of cycles of the polymerase chain reaction (PCR). PCR is a means of repeatedly copying a specific DNA sequence in a sample; normally, the PCR cycle is carried out 28 times in the course of conventional DNA profiling. In contrast, the FSS method involved 34 duplication cycles, due to the low quantity of biological material in the initial sample. The extra cycles were used in order to allow analysts to identify target sequences of DNA that would have otherwise been insufficient to develop a profile. A number of commercial firms have subsequently sought to develop different forms of ultra-sensitive profiling. For example, the company LGC introduced improved methods for removing contaminating substances from crime scene samples, which it claims also enables DNA profiling to be carried out under ultra-low volume conditions (Forster et al., 2008).

In addition to sample preparation, the issue of interpreting profile data produced by these kinds of ultra-sensitive methods has attracted considerable scientific discussion. Many of the methodological problems associated with conventional DNA profiling are intensified in the case of LT-DNA. The small starting quantities involved in LT-DNA analysis can exacerbate problems with the interpretation of evidence and the determination of matching profiles, as amplifying the amount of DNA in the sample also amplifies the significance of possible artefacts and contaminants (Buckleton and Gill, 2004). Forensic analysts have developed a specific nomenclature for describing the interpretation issues associated with LT-DNA analysis (see Figure 1(a) and (b)). For example, drop-out describes the phenomenon whereby the lack of sample, or adverse environmental conditions, means that specific genetic components of DNA profiles (known as alleles) are not represented in the profile at all. Conversely, drop-in describes the possibility that contaminating alleles will appear in a profile. Further interpretative issues may arise when DNA from more than one individual creates a mixed profile.

OPEN IN VIEWER

Figure 1.

(a) ‘Drop-in’. The peaks in this diagram represent allele peaks in LT-DNA electropherograms, which are repeated analyses of the same sample. Although three peaks match, a fourth allele appears in the lower electropherogram, possibly the result of ‘drop-in’. (b) Drop-out. In this repeated LT-DNA analysis, the left and right alleles appear to match. The centre allele is absent in the lower electropherogram data, possibly as a result of ‘drop-out’.

Such interpretative issues led developers of LT-DNA to formulate probabilistic algorithms for evaluating the significance of evidence in a particular case. These forensic scientists attempted to model the probability of spurious peaks in profiles that otherwise matched. They utilized the Bayesian form of probability theory, which is also used to express the probative value of conventional DNA profiles and other forms of evidence, to produce algorithms that, they claimed, took into account the probability of allele drop-in and drop-out under certain conditions. In the event of a partial match, these algorithms were designed to determine the extent to which spurious peaks affected the probative value of LT-DNA evidence. In some cases, these algorithms were converted into computer software for in-house use in forensic science organizations (Gill et al., 2007).

As the Reed case demonstrated, the transfer of DNA-containing material raises a further set of interpretative issues. Some forensic analysts claim that individuals are liable to shed DNA in a variety of ways, for example, by exhaling or by simply touching surfaces (Lowe et al., 2002). Even though extremely low quantities of DNA would be deposited onto surfaces by touch, and then transferred onto the hands of other individuals, who themselves may then deposit this DNA elsewhere, LT-DNA may be sufficiently sensitive to identify such transferred DNA. The potential phenomena of secondary or tertiary transfer pose considerable problems when trying to match a profile found at a crime scene with an individual.

The interpretative issues surrounding LT-DNA profiles were a source of disquiet in forensic scientific circles. For example, one interviewee was highly critical of the use of terms such as ‘drop-out’ to explain LT-DNA profiles, arguing that it biased investigators toward incriminating rather than exculpating suspects:

You can’t exclude people when you invoke dropout [and say]: ‘the reason why there’s the alleles are not there is because they’ve dropped out’. Or, it may be that the alleles are never there in the first place! (Forensic Science Consultant, 2008)

When elaborating upon an example of a typical LT-DNA profile, this interviewee saw the process of assigning an LT-DNA profile match as fraught with ambiguity and vulnerable to confirmation bias:

[LT-DNA] technique[s] [are] so unreliable … that’s why the DNA analysts work back, they say ‘well if this is the story, the evidence is consistent with it’, and that’s a terrible way of looking at evidence. Because the question really is ‘how many other stories is it consistent with?’ (Forensic Science Consultant, 2008, emphasis added)

This forensic analyst criticized the interpretation of LT-DNA profiles for being influenced all too often by the kind of investigative narrative already pursued by police. Hence, this analyst regarded the use of drop-in and drop-out to justify decisions about ambiguous data as a way of merely reinforcing the path an investigation already might have taken.

Another interviewee used equally frank terms to criticize the use of ‘drop-in’: ‘Contamination is contamination. There’s no degree of contamination … contamination ruins evidence’ (UK Policing Researcher, 2008, emphasis added).

Constructs such as drop-in and drop-out originally were developed as integral components of the process of operationalizing LT-DNA. Their continued use demonstrates how the production of evidence, through which minute traces of biological material are rendered into instrumental data, is inextricably bound up with practices of interpreting those data. Terms such as ‘drop-in’ and ‘drop-out’ shape the narratives that make sense of mute electropherogram traces. However, the above respondent’s criticisms about the reliance on ‘stories’ for interpreting LT-DNA data highlight the vulnerability of these processes to dissenting views. This, together with the ambiguities of transfer, echoes the observations made by Lynch et al. (2008) that ‘DNA evidence is meaningful only when it is embedded in stories that mention other evidence, possible suspects, and how the evidence itself was handled and interpreted’ (p. 191). ⁵

The incorporation of drop-in and drop-out into Bayesian algorithms can be regarded as a possible instance of ‘mechanical objectivity’ (Daston and Galison, 2007). The production of ‘objectivity’ through the development of these algorithms facilitates the synthesis of technological comprehension with casework narrative. By becoming part of a platform of automated Bayesian reasoning, drop-in and drop-out are coated with a sheen of sophisticated probabilistic inscription. While LT-DNA data gain meaning from the stories that surround them, the conversion of drop-in and drop-out into probabilistic functions represents a significant reinforcement of the feedback process. The technologizing of investigative narratives is a particularly powerful objectifying strategy. Here, it is not technology alone that is ‘black boxed’ but also the accompanying ‘stories’ that are required to add meaning to LT-DNA profiles. The technology and the social fabric in which it resides are mutually constituted. ⁶

The development of a nomenclature for LT-DNA profile interpretation is significant for other reasons, in that these terms, and their embedding into Bayesian frameworks, work to distinguish LT-DNA from ‘conventional’ forms of forensic DNA profiling. They reinforce LT-DNA as a particularly sophisticated variant, a source of new forensic innovations and a driver for more objectifying technologies of criminal investigation. Such distinguishing work is perhaps necessary in the light of a variety of indeterminacies that were highlighted in other discussions that highlighted problems with defining LT-DNA as distinct from ‘conventional’ DNA profiling:

If you asked anybody in the field at the moment who’s had any education in forensic genetics, you’re going to get a variety of responses. … There is no agreed upon answer at the moment, and this is one of the problems … are any of these definitions helpful, and has anyone actually agreed upon them? (seminar discussion at English University, 2011) ⁷

Different parameters have been used to distinguish LT-DNA from conventional DNA profile evidence. These include the volume of DNA analysed in a sample and the degree of stochastic deviations from the expected observations of DNA data, as well as methodological categories such as the number of PCR cycles – a number that has been found to vary depending on which manufacturer’s equipment is used (seminar discussion, 2011).

Defining LT-DNA in terms of the volume analysed was regarded by one discussant as particularly problematic:

Is it useful to have a threshold cut-off point of the quantity of DNA for a technique threshold? I don’t particularly think it is, one of the reasons is that its very difficult to accurately quantify DNA at this level … if you can’t be sure what you started with how can you be sure what you’ve diluted down? This is another issue with Low Template DNA … every time you pipette, how do you know how much you’ve actually picked up? … If we don’t know what we’ve got to start with, we don’t actually know how much we’re putting in … how do we know how much is actually in the pot? It is a problem this DNA quantity as a standard, as a threshold, as a definition for what we’re looking at … even today, quantification systems are not good enough to tell us how much we’ve got. (seminar discussion, 2011)

This highlights the uncertainty at the heart of LT-DNA methods. The minute volumes associated with LT-DNA, typically in the picogram (10⁻¹² g) range, mean that accurate quantification of samples is very difficult. At times, analysts may not know exactly what kind of evidence they are preparing. Definitions as to what constitutes the volumetric threshold at which ‘conventional’ DNA profiling becomes ‘low-template’ conditions were found to vary and ‘haven’t been agreed upon’ (seminar discussion, 2011).

LT-DNA interpretation algorithms represent a notable aspect of biolegality in action, rendering the narrative epistemologies upon which forensic DNA relies for meaning into an ever more immutable form and shaping an even more intimate relationship with scientific ‘objectivity’. Yet the distinguishing work of constructing Bayesian functions to objectify ‘drop-in’ and ‘drop-out’ is challenged by other practical and technical factors. These render the threshold for separating ‘low-template’ conditions from ‘conventional’ DNA profiling highly indeterminate.

As the next section demonstrates, the instability surrounding LT-DNA has led to complex forms of boundary work. A series of exchanges in FSI:G is presented to show how disputes over the status of LT-DNA led scientists to (re)shape the character of science vis-à-vis the ‘non-scientific’ domain in different ways. They continue to deliberate whether LT-DNA is a bona fide ‘scientific’ technology or an example of dubious ‘pathological science’ (Langmuir, 1953) that has only survived due to law’s mistaken interference based on incorrect assumptions. ⁸ This has led contending groups to project different portrayals of the law–science relationship and of the character of ‘novel science’.

Forensic science international: genetics – a site of attrition in the low-template wars

One of the most eagerly debated issues in FSI:G concerned the authority over decisions about the validity of LT-DNA. In addition to the Reed decision, defenders and developers of the technique cited a case heard in New York, People v Megnath (2010). 8 February 2010), in which LT-DNA was subject to an admissibility hearing in terms of the State’s use of the Frye standard of ‘general acceptance’ in the relevant scientific community. ⁹ The prosecution argued that LT-DNA had been ‘generally accepted’ in the United Kingdom and New Zealand, and the judge agreed that results from the technique were admissible evidence. The ruling was hailed by some commentators as a significant vindication of LT-DNA in the Federal Court, which had previously demonstrated notable scepticism towards the method.

During a series of comments and replies published in FSI:G, Bruce Budowle and his co-authors, who have often been publicly sceptical of LT-DNA (despite previous stated support for conventional DNA profiling), claimed that a ‘legal threshold of general acceptance’ did not ‘equate to scientific reliability’, and that ‘the proper forum for reliability discussions is the scientific arena’ (Budowle and Van Daal, 2011a: 13). Budowle and Van Daal (2011a) strongly criticized developers of LT-DNA, including noted forensic scientists like Peter Gill and John Buckleton, for seeming to rely on the courts to decide what constituted ‘validity, reliability and quality’ (p. 13). Budowle and his colleagues viewed this kind of epistemic dependency as having previously led forensic science to fail to meet the standards expected of the ‘pure’ sciences. ¹⁰ They claimed that the Megnath hearing relied on erroneous statistical interpretations, which undermined the guilty verdict. They also criticized the use of the Frye ‘general acceptance’ standard to justify adoption of the technique and claimed that this standard was wholly legal and based on previous hearings, rather than on actual ‘scientific conclusions’ (Budowle et al., 2011: 5). They also claimed that no data had been released in either the United Kingdom or New Zealand to provide substantive scientific backing for the validity of LT-DNA.

The issue of how scientific data could be regarded as ‘generally accepted’ was conceded by supporters of LT-DNA as being problematic from a scientific standpoint. Concerning the issue of validation, they admitted that profound differences remained between scientists and lawyers over the extent to which deviations from reproducibility were considered acceptable:

The two replicates at low template level work will typically be broadly similar but, for example, some alleles present in one replicate may be smaller or not visible in another replicate. This is common knowledge to forensic scientists but was met with deep concern when presented to a group of prosecutors at Auckland, New Zealand, in 2008, thereby exposing the gap between scientific and legal perceptions. (Buckleton, 2009: 258)

While scientists working with LT-DNA always anticipated a degree of variability in results, the occurrence of phenomena such as allele drop-out surprised and worried prosecutors. Buckleton (2009) noted further that the ‘very considerable gap in view and expectation between forensic scientists and lawyers’ (p. 258) was likely to endure. This gap persisted despite proclamations by some of the same authors who hailed decisions like Megnath: ‘It is welcome to hear from the judiciary that evidence interpretation does not have to be perfect to be admissible’ (Balding and Buckleton, 2009: 2).

In a further round of the exchange, Buckleton and Gill (2010) argued that it was ‘misleading to describe reproducibility to be either a Daubert requirement ¹¹ or a Frye requirement’ (p. 222). ‘Variability’, they argued, ‘and indeed uncertainty, is a part of most, if not all, scientific endeavours’ (p. 222). Their argument portrays variability and uncertainty as inevitable and unavoidable in scientific practice and thus not necessarily a detriment to the validity of particular evidence items.

Another key issue was the extent to which a discrepancy existed between the theoretical basis of LT-DNA interpretation, and the actual practice of interpretation. During the course of the FSI:G exchanges, critics of LT-DNA argued that, contrary to the representations published in the literature, the exact manner in which LT-DNA profiling was used in casework was unclear at best: ‘The forensic science community does not know what the practices of [LT-DNA] laboratories are and whether they are valid and reliable’ (Budowle and Van Daal, 2011b: 15). More seriously still, these critics alleged that recent cases had exposed apparent methodological double standards in the FSS: ‘a difference between what they recommended and what is practiced’ (Budowle and Van Daal, 2011a: 12). In a further comment in the exchange, they added,

We now know that the Forensic Science Service (FSS), for example, does not follow the often cited Gill et al. article (written by FSS employees) for statistical assessment of [LT-DNA] evidence even a decade after its publication. (Budowle and Van Daal, 2011b: 15)

Budowle and Van Daal (2011a) also accused forensic scientists in other jurisdictions of engaging in practices that are ‘very different than what is recommended in the scientific literature’ (p. 12). They alleged that some of the key developers of these recommended practices had not promulgated their own methods to colleagues engaged in casework. They alleged that the New Zealand lab where John Buckleton had been involved in developing Bayesian LT-DNA interpretation methods did not actually use the recommended methods, and that these inconsistencies extended to casework conducted in Australia.

Gill and Buckleton did not directly respond to these allegations, but in an earlier part of the exchange, they conceded that adoption of their theoretical solutions had been forestalled. They admitted that there was a need for ‘specialist software to enable the probabilistic solutions to be fully implemented’ (Gill and Buckleton, 2010: 226). ¹² They explained that the delay stemmed from lack of business interest in developing the technique, rather than from strictly technological limitations: ‘It is of course disappointing that nearly a decade later, vendors still have not developed commercial solutions based on our statistical thinking’ (Gill and Buckleton, 2010: 223, emphasis added).

The Bayesian philosophy underpinning the statistical methods previously advocated (but not necessarily practiced) by these developers also forms the basis for other techniques developed by forensic scientists (Lawless and Williams, 2010). Perceived ‘mistakes’, such as were brought out in the Hoey verdict, were explained away by proponents of LT-DNA in terms of supposed misconceptions about the ‘relevance’ of LT-DNA evidence to a particular case. Specifically, they argued that such mistaken presumptions concerned how DNA evidence related to prior ‘activity’. In this connection, Buckleton and Gill mention the ‘hierarchy of propositions’, a commonly understood concept in UK forensic science, which is strongly associated with Bayesian evidence interpretation. They claim that erroneous interpretation would be reduced if Bayesian techniques were more widely adopted: they argue that ‘nothing else is required, other than to educate scientists, judges and lawyers’, in the ‘uses and practicalities’ of such techniques (Gill and Buckleton, 2010: 226, emphasis added).

Their argument implies that ignorance and a lack of interest in the kind of techniques they advocate is responsible for the apparent lack of consistency between theory and practice. This account of the implementation gap is framed in terms of a certain kind of deficit model, ¹³ implying that the recommended interpretative techniques could be used in practice, but that a series of commercial, communicative and institutional barriers stand in their way.

In an earlier contribution to in the exchange, Buckleton (2009) cited other reasons for delays in implementation, such as the lack of availability of validation protocols in the technical literature due to general tendency not to publish replications:

Journal editors are understandably reluctant to publish validation papers. If the technique has been published once and subsequently other laboratories repeat the work and obtain the same or similar results the subsequent papers will be, rightly, branded not novel and hence unworthy of publication. (p. 257)

Buckleton (2009) added that, while this tendency was widely known in scientific circles, it was ‘possibly not widely understood within the legal community’ (p. 257). Buckleton (2009) cited other institutional factors that preclude open publication of validation protocols:

Access can be further constrained by intellectual property considerations. In my own organisation the placement of a validation report in the Institute website would require permissions and many steps and has never been attempted. (p. 257)

He also cited organizational practicalities related to the problem of gaining consensus in professional scientific bodies:

One alternative is to have professional bodies or especially commissioned committees endorse certain procedures … The obvious, but incorrect, implication is that anything not currently endorsed does not make the standard. Obtaining agreement by large committees is difficult and time consuming. Progress can often become stalled on matters of mind numbing detail or on what level of different but parallel methodology is acceptable. Any endorsement tends to set a technique in stone thereby inhibiting progress. (p. 257)

Elsewhere, however, Gill and Buckleton made different claims about the scientific status of LT-DNA. In 2009 and 2010, they published articles that highlighted an ‘underlying confusion’ (Gill and Buckleton, 2010: 221) in a low-template analysis. They claimed that ‘no satisfactory definition’ could ‘be applied to delineate’ between LT-DNA and ‘conventional’ DNA profiling conditions (p. 221). Specifically, they argued that the interpretation problems associated with LT-DNA actually applied to all forms of DNA profiling and that no absolute distinction could be drawn between the techniques, even though ultra-sensitive profiling had previously been offered as a distinct product by commercial forensic science providers. They attacked what they now characterized as ‘an arbitrary definition of LT-DNA vs. conventional DNA profiling’ (p. 221) and stated that they ‘now reject this definition because the stochastic effects associated with the analysis of LT-DNA are undeniably observed with all DNA profiling technologies’ (p. 221, emphasis added). Elsewhere, they stated that they had ‘abandoned the concept in favour of development of a universal strategy that can be used for all DNA profiles regardless of technique used’ (Gill and Buckleton, 2009: 555).

These assertions were dismissed by critics (Budowle and Van Daal, 2011b):

Their response has not ended our concerns, and we disagree with much of what they attempt to argue to justify avoiding the vagaries of [LT-DNA] typing: [LT-DNA] typing is not a way of thinking; it is an analytical tool with reproducibility issues that must be considered and properly addressed. (Budowle and Van Daal, 2011b: 15, emphasis added)

In a seminar I attended, a discussant (whose name is withheld for confidentiality) also questioned these assertions:

… it seems the people who invented the technique are now trying to say its not really a technique at all … it does look like they are backtracking in many ways to avoid some of the issues … (seminar discussion, 2011)

The debate in the pages of FSI:G came to an abrupt end when the editors of the journal stated that they would publish no more correspondence on the matter (Schneider et al., 2011). However, this did not halt the dispute. One discussant, a scientist working in a UK academic institution who has researched LT-DNA, reported, during a seminar I attended, that a recent meeting of the International Society for Forensic Genetics had seen the debate continue as vehemently as before. The revised claims of key developers of LT-DNA have not as yet silenced criticisms of the technique. Furthermore, their move arguably complicated efforts by regulators to better comprehend LT-DNA through exercises like the Caddy inquiry. However, as yet, the ‘denial of threshold’ argument does not appear to have affected the understandings of LT-DNA articulated by legal practitioners.

My analysis of FSI:G shows how the novelty of LT-DNA has been treated as an opportunity to reconstruct the law–science boundary and to carve out distinct, but competing, shapes for science as a zone of authority. Sceptics such as Budowle and his colleagues sought to ‘ring-fence’ LT-DNA by arguing that only scientists, when free from interference of the courts and police, possess the epistemological warrant to assess its reliability. These sceptics portray legal actors as scientifically naïve and prone to premature and mistaken conclusions.

It is useful to consider the sceptics’ portrayals in the context of the notion of the ‘law-set’ (Edmond, 2001). The term ‘law-set’ has been used to describe collectives of scientific and non-scientific actors who negotiate the meaning, relevance and reliability of scientific evidence presented in court. Edmond (2001) argues that the knowledge produced as a result of these situated negotiations is neither reducible to simple categorizations of scientific knowledge nor ascribable to the ‘distorting’ effects of adversarial legal discourse. In this case, however, LT-DNA sceptics seemingly aim to narrow the ‘law-set’ to a smaller ‘core-set’ (Collins, 1985) of scientists – not necessarily laboratory scientists but duly qualified forensic analysts who have academic credentials.

Although Budowle et al. construct LT-DNA as an object for an exclusive ‘scientific’ debate, they do so by framing it as a potential instance of ‘pathological’ science if used in an unregulated way by unqualified personnel. Equally significant is their argument that LT-DNA is a distinct ‘analytical tool’, despite what Buckleton and Gill sometimes claimed about it being continuous with DNA profiling in general. Budowle et al.’s critical framing of LT-DNA can help to maintain its ‘tool-like’ existence. Nevertheless, Budowle and his colleagues still had to contend with the ongoing acceptance of LT-DNA evidence in court cases such as Reed and Megnath.

Buckleton and Gill, who helped develop LT-DNA methods, appeared to adopt a more convoluted position. At times, they also ring-fenced LT-DNA by portraying it as a novel innovation that only qualified ‘scientists’ can use. However, unlike critics of the technique, they argue in favour of its ‘valid’ status. They explain the seeming discontinuity between their theoretical prescriptions and the practical implementation of LT-DNA profiling by citing legal, commercial and organizational factors that perpetuate ignorance about their technology, and prevent it from reaching its full investigative potential. By rhetorically disavowing how their ‘scientific’ technology is misused in the legal context, these originators of LT-DNA assign validity to a space of possibility rather than immediate use (if we did not have these constraints our science could be shown to be fully efficacious, etc.).

At other times, however, proponents of LT-DNA reconstruct the law–science boundary quite differently to reinforce its legal status, for example, when arguing in support of the Megnath decision. The justification of uncertainty as part of everyday scientific practice contrasts notably, however, with their attribution of authority to the courts, given their expressed misgivings about the perceived misunderstandings of scientific realities on the part of lawyers.

As I noted earlier, Buckleton and Gill elsewhere attempted to dissolve the distinction between ‘low-template’ and ‘conventional’ DNA profiling. This argument apparently is an attempt to extend their Bayesian methods to all DNA evidence. By contesting the idea that it is a distinct technological approach, they dismiss as spurious the distinct epistemic status that critics assigned to LT-DNA. But, as I noted above, this argument has been received with considerable scepticism from many of their scientific peers, and leading forensic science providers, government regulators and courts have explicitly treated LT-DNA as a distinct method.

Conclusion

While other recent advances in forensic DNA technology, such as phenotypic profiling and familial searching, are more readily identifiable as distinct developments, the exact status of LT-DNA as an innovation is less clear. I have nevertheless taken the opportunity to explore how this contested forensic technique has been reconstructed through the deliberations of developers and their scientific critics. It should be noted that in the FSI:G exchanges, both sets of interlocutors operated on a relatively even footing. Neither side could be regarded as subjugated or marginalized; instead, all enjoyed roughly equivalent degrees of professional standing and experience.

My study reveals how the technological boundaries that surround LT-DNA showed a high level of multivalency. In the debate discussed here, a series of distinctions were drawn: basic distinctions between ‘valid’ science versus potential ‘pathological science’, and ‘new’ versus ‘the same’ technology, as well as some more subtle distinctions (LT-DNA as a ‘tool’ vs LT-DNA as a set of ‘conditions’). Much of the concern revolved around exactly how, if at all, LT-DNA differs from ‘conventional’ profiling. Attempts by LT-DNA developers to further the debate, by questioning the notion of a discrete threshold between ‘conventional’ and ‘low-template’ conditions, have not yet fully resolved the issue.

My analysis of the FSI:G exchanges also shows how the relationship between the principles and practice of LT-DNA was differentially constructed. At times, ‘typical’ actualities of scientific work, such as the variability of results, were naturalized by developers who portrayed them as ‘inevitable’. Legal actors were portrayed as misunderstanding this inevitability, even though their contribution to maintaining the credibility of LT-DNA was recognized. At other times, developers defended their Bayesian LT-DNA interpretations by claiming that a series of external constraints impeded their wider uptake. Together these instances show how the practical realization of LT-DNA was accounted for in diverse ways, depending on who was upholding its credibility, in principle, and by what means.

In highlighting this, I have shown how some forensic scientists negotiated the nature and limits of epistemic responsibility; placing such responsibility in relief against the wider domain in which it operated. At times this was partially delegated to the realm of law, but at other times LT-DNA theory was defended by identifying a series of factors that impeded its wider application. Framings of LT-DNA were co-produced with the differential projection of technical expertise. In this way, deliberations about the precise nature of LT-DNA were intimately linked to claims about who was qualified to make decisions about it, and by what means. However, the drawing of these epistemic boundaries to accommodate LT-DNA has, at times, been complex and convoluted.

Competing arguments exposed some problematic interdependencies. For example, the portrayal of LT-DNA by sceptics like Budowle may ostensibly represent a more straightforward instance of boundary work. The depiction of LT-DNA as a potentially ‘pathological’ technology sought to reinstate scientific authority over other forms of decision-making, reducing the ‘law-set’ of relevant actors to a narrower ‘core-set’ of qualified experts. But in attempting to keep the controversy open in this way, the technology was portrayed in a more familiar light to prosecutors. In rejecting some of the assertions of Buckleton and Gill and continuing to portray LT-DNA as a distinct ‘tool’, they aligned with the comprehensions of lay users, entrenched through instances of court cases and regulatory interventions.

Law itself may not be entirely unaffected, and this study revealed other possible interdependencies. Some legal discussions of LT-DNA in court cases emphasized possible future consequences for the capacity of lawyers to marshal expert evidence. For example, the outcome of the Reed case upheld the English Court’s prerogative to determine which form of ‘expertise’ was most relevant to a specific case. I have demonstrated, however, that Reed raised concerns among some lawyers about adverse implications for the future relationship between them and expert witnesses. They expressed fears about how Reed could set precedents that risked further encroachment of scientific witnesses into ‘their’ domain, threatening their ability to make decisions about expert evidence. The response to the Reed decision, therefore, indicates how the situated production of law-set knowledge could potentially induce counteractive effects in terms of the future relationship between legal actors and scientific witnesses. This observation raises questions regarding the future capacities of legal actors to influence law-set construction in the face of contested scientific evidence and amid concerns over possible knowledge deficits.

The ‘distinguishing work’ of LT-DNA has struggled to fully close technical and practical indeterminacies. Scientists continue to debate its precise status and the extent to which it is a ‘distinct’ innovation. LT-DNA has, however, gone some way toward regaining its image as a reliable forensic approach following a high-profile setback in the form of the Hoey verdict. In focusing on this tension, my case study contributes to the understandings of biolegality by providing a more detailed view of the processes underpinning forensic innovation. I have shown how deliberations over a contested forensic technique have led scientific and legal actors into an ongoing (and sometimes inadvertent) process of reconstructing relations and delegations of responsibility. The example of LT-DNA shows how the twin authorities of science and law readjust to the challenge of comprehending emergent technologies. What I have also shown, however, is that the increasing intimacy that arises between these authorities presents complications for both scientific experts and lawyers. The boundary work involved in stabilizing LT-DNA is itself highly contingent. This is apparent not only in courtroom proceedings, but even in discussions published in specialist technical literature. Together, these instances expose ongoing pressures that both drive and challenge biolegal evolution.

LT-DNA is a clear product of biolegality in that it emerged as a result of the desire to create ‘novel possibilities for transmuting and extending the reach of police surveillance and investigation’, which has equally ‘challenged legal institutions’ (Lynch and McNally, 2009: 284). My findings, however, show that more detailed attention must be paid to the practices, processes and dynamics that constitute, and are constituted by, biolegality. In summary, my analysis demonstrates how the reconstruction of relationships between scientific expertise and the legal realm, and the nature of subsequent technological outcomes of that reconstruction, presents a constant, ongoing series of challenges to biolegal orderings. How this tension continues to play out invites further scrutiny.