Dual-use research and the H5N1 bird flu: Is restricting publication the solution to biosecurity issues?
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
Science and Public Policy Advance Access published September 27, 2013
Science and Public Policy (2013) pp. 1–14 doi:10.1093/scipol/sct064
Dual-use research and the H5N1
bird flu: Is restricting publication the
solution to biosecurity issues?
Sabrina Engel-Glatter*
Institute for Biomedical Ethics, University Basel, Bernoullistrasse 28, 4055 Basel, Switzerland.
*Corresponding author. Email: sabrina.engel@unibas.ch.
Recent studies altering the host range of the H5N1 bird flu virus have refueled intense debates
over the potential misuse of academic life science research. To curtail the bioterrorism threat,
it has been suggested that dissemination of the research results and methodology should be
restricted. However, doubts have been raised over the suitability and effectiveness of this
measure. Using the H5N1 studies as an example, this paper summarizes the main arguments
of the debate. Particular attention is paid to the issue of the tacit knowledge required to replicate
published life science research results, which has so far received limited attention. Taking into
account the importance of tacit knowledge for life science research, it is argued that preventing
publication of the methodology does not decrease the threat of bioterrorism.
Keywords: dual-use research of concern; life science research; biosecurity; tacit knowledge;
influenza.
1. Introduction to include non-state actors in late 2001, raised doubts
over its effectiveness to address the current challenges
Since the anthrax attacks in 2001, the dual-use potential of and initiated this controversial debate (McLeish and
life science research has been intensely discussed. Fears of Nightingale 2007; Tucker 2002).
published research findings serving as a blueprint for The concerns have been publicly discussed after selected
building biological weapons have especially troubled US scientific papers on pathogenic viruses were published,
policy-makers (Malakoff 2001). Initially, the term dual-use including a paper on an engineered mousepox virus that
applied to engineering and information technologies and overcomes vaccine protection (Jackson et al. 2001), a pub-
related knowledge that may be used for both peaceful and lication describing the first de novo synthesis of a pathogen
military purposes with little or no modification (Hall 1965; (the poliovirus) (Cello et al. 2002), a study identifying
Panel on Scientific Communication and National Security virulence factors that distinguish the causative agent of
1982). The rapid advances of life science research in recent smallpox, variola virus, from its close relative, vaccina
years have pushed its experiments, results, and related virus (Rosengard et al. 2002), a paper characterizing the
technologies into the center of the dual-use debate. reconstructed 1918 Spanish influenza pandemic strain
Long before the misuse potential of life science research (Tumpey et al. 2005), and recent studies on mutated
received so much public attention, the development, pro- versions of the H5N1 bird flu (Herfst et al. 2012b; Imai
duction and stockpiling by state actors of bacteriological et al. 2012). To prevent deliberate misuse of such published
and toxic weapons had been banned internationally by the knowledge, there have been calls for tighter regulations
1972 Bioweapons and Toxin Convention (BTWC). and restrictions on academic research and the dissemin-
However, the apparent ease with which life science know- ation of research results (Tucker 2002; Wallerstein 2002,
ledge develops and diffuses is perceived to have widened Resnik and Shamoo 2004; Gaudioso and Salerno 2004;
the range of actors with the potential to develop bio- van Aken 2006). In addition, a variety of international
weapons from state actors to non-state actors. The and local policy measures have been implemented. For
anthrax attacks, and the failure to reinforce the BTWC example, the US government has defined a list of ‘select
ß The Author 2013. Published by Oxford University Press. All rights reserved. For Permissions, please email: journals.permissions@oup.com2 of 14 . S. E. Glatter
agents’ with the potential to pose a severe threat to both any attempts to determine the pandemic potential of the
human and animal health, to plant health, or to animal virus remain speculative. It is this unknown potential, or
and plant products.1 US government regulations control in other words, our ignorance about the implications of
research with, and transfer of, these ‘select agents’.2 releasing the virus into the environment, that leads to the
Further measures include: publication guidelines for disagreement amongst virology experts on whether the
journals to pre-screen potentially sensitive publications mutated H5N1 virus and the related experiments can
(Atlas et al. 2003), US government oversight of research truly be termed ‘dangerous’ (Palese and Wang 2012;
projects and possible classification of sensitive research,3 Bouvier 2012; Inglesby et al. 2011; Morens et al. 2012).
the implementation of screening guidelines for synthetic With this in mind, a fixed list of ‘dangerous technologies
DNA orders from official providers for sequences and pathogens’ and ‘experiments of concern’, although
encoding select agents and toxins,4 and the establishment helpful, appears to be of limited use to define which type
of codes of conduct for scientists.5 of research may pose a threat to public health and safety.
In an attempt to address concerns over life science A more flexible perspective on dual-use technology and
research and to find a formal definition of dual-use ‘dangerous science’ is needed to ensure that policy
research, the US National Academies published the measures keep up with the rapid developments in life
highly influential report ‘Biotechnology Research in an science research and secure the benefits of academic
Age of Terrorism’ (National Research Council 2004). research. This different perspective on the development
The so-called Fink report6 defines dual-use research of of life science research and its impact on the dual-use
concern (DURC) as: debate has been discussed amongst sociologists (McLeish
and Nightingale 2007; Vogel 2008; Tucker 2011).
. . . research that, based on current understanding, can be
However, it has not yet found its way into the public
reasonably anticipated to provide knowledge, products, or
debate over dual-use research. By using the controversially
technologies that could be directly misapplied to pose a
threat to public health and safety, agricultural crops and debated H5N1 studies as a case study, this paper will em-
other plants, animals, the environment, or material. phasize the impact of an alternative perspective on the
dominant biosecurity discourse and highlight how an
In addition, the report defined seven categories of ‘experi- alternative view of ‘dangerous’ life science research may
ments of concern’ to be scrutinized when designing research alter the prioritization of policy responses.
programs and publishing sensitive results. These include: Furthermore, in the light of an alternative perspective,
(1) to render a vaccine ineffective this paper will evaluate the effectiveness of one key policy
(2) to create resistance to therapeutically useful antibi- measure: restricting dissemination of the research method-
otics or antiviral agents ology. It has been argued that the potential dangers and
(3) to enhance the virulence of a pathogen or render a disastrous consequences of academic research demand the
non-pathogen virulent implementation of precautionary measures such as restrict-
(4) to increase the transmissibility of a pathogen ing publication (Osterholm and Relman 2012; Kuhlau
(5) to alter the host range of a pathogen et al. 2011; Resnik 2013). Thus, besides outlining a differ-
(6) to enable the evasion of diagnostic/detection ent perspective on the potential dangers of dual-use life
modalities science research, this paper will summarize the main argu-
(7) to enable the weaponization of a biological agent or ments of the dual-use debate around the H5N1 case and
toxin make use of the precautionary principle (PCP) to evaluate
whether preventing publication of the methodology of
This paper will outline that categorizing and focusing on DURC is an effective tool to decrease the bioterrorism
‘dangerous technologies and pathogens’ reflects a limited threat.
perspective on the dual-use dilemma. As McLeish and
Nightingale (2007) pointed out, an unambiguous definition
of ‘dangerous’ cannot be agreed upon as this: 2. Research on highly virulent, possibly
. . . is not a descriptive term that denotes the property of some- human transmissible, H5N1 influenza
thing but an expressive term that refers to how we think about virus strains
the possible implications of the properties of something.
Two recent studies on the H5N1 bird flu have intensified
As a result, different people, even experts, will evaluate the the international debate over DURC. They describe a set
dangerousness of technologies and pathogens differently. of mutations necessary for the highly virulent H5N1 bird
This is reflected by the current discourse around two flu virus to transmit amongst mammals, possibly humans
studies on the highly virulent H5N1 bird flu virus. (Herfst et al. 2012b; Imai et al. 2012). The H5N1 case
Researchers have mutated the virus so that it can spread raised concerns for several reasons: the bird flu virus is
amongst mammals via aerosol (Herfst et al. 2012b; Imai highly virulent, it is listed on the select agent and toxin
et al. 2012). However, as no human studies are desirable, list, and the performed experiments can be allocated toDual-use research and the H5N1 bird flu . 3 of 14
category 5 (to alter the host range of a pathogen) of the aggressive control programs, and to adapt current
Fink report. pandemic preparedness plans (Kawaoka 2012; Fouchier
When submitted to the journals Nature and Science for et al. 2012a, 2013).
publication, the editors referred the two studies to the However, the NSABB experts had initially decided that
National Advisory Board for Biosecurity (NSABB). the benefits of the study would not outweigh the risks of
Established by the US government as a result of the publication (Berns et al. 2012). The main risks in publish-
Fink report, the NSABB provides advice, guidance and ing both papers in full were described as twofold. First, the
leadership regarding biosecurity oversight of DURC in results and methodology of the H5N1 studies, if published,
the USA.7 Its decisions are, however, not legally binding. could be directly misapplied by others to pose a threat to
For the first time, the NSABB had voted against the pub- public health and safety (biosecurity risk). In particular,
lication of the methodological details in order to prevent the possibility of synthesizing the flu mutant strains based
‘those who would seek to do harm’ from replicating the on published sequences and the identified mutations raised
experiment.8 The board had concluded that the harms of concern. The NSABB members Osterholm and Relman
fully publishing these results exceeded the benefits of pub- argued that:
lication (Berns et al. 2012). Two crucial aspect may have
. . . reverse engineering and synthesizing a mutant A/H5N1
influenced the board’s so far unique decision: the official
virus strain will be easier for many scientists if they have
World Health Organization estimate for the case fatality
access to the complete methods and results of the two
rate of H5N1 avian influenza of around 60% (Beigel et al. research efforts and do not need to surmise the steps of the
2005)9, and misunderstandings of technical details investigators. (Osterholm and Relman 2012)16
(Morens et al. 2012).10 Hence, fears of potentially disas-
trous consequences for public health resulting from an Second, research with the virus could result in an uninten-
unintended or intended release of the engineered H5N1 tional release into the environment (biosafety risk).17 In
strains were particularly strong (Berns et al. 2012). order to mitigate the biosecurity and biosafety risks
To allow time for evaluating the risks and benefits of the associated with publication, the board had suggested
research, the influenza research community proposed a redacting the papers and excluding any description of the
three-month voluntary moratorium11 on research methodology (Berns et al. 2012). However, the effective-
involving highly pathogenic avian influenza H5N1. ness of partially restricting or even preventing publication
During that time, the pros and cons of publishing the was repeatedly questioned.
manuscripts were intensely discussed by scientific and The main argument used in favor of publishing the
security experts, the NSABB, the WHO and the US gov- research in full was the necessity to share the research
ernment. Finally, a majority vote of 6 to 1212, strongly results in the wider scientific community to ensure free
opposed by the NSABB member Michael and open science. Withholding the methodology has
T. Osterholm,13 led to the revision of the NSABB’s been criticized as unethical, and as an assault on the
decision resulting in full publication of the two papers openness, accessibility and quality of scientific research
(Herfst et al. 2012b; Imai et al. 2012). The research mora- (Bouvier 2012; Racaniello 2012). It was argued that
torium proposed by the influenza community was omitting the methods from the manuscripts presents a
extended beyond the originally proposed three months to barrier for researchers to engage in further research
provide time for further evaluation of biosafety14 and thereby limiting the scientific output essential to combat
biosecurity15 concerns (Fouchier et al. 2012a). The the natural pandemic threat of influenza viruses (Palese
research finally commenced one year later in January and Wang 2012; Morens et al. 2012; Kawaoka 2012;
2013 (Fouchier et al. 2013). Perez 2012; Faden and Karron 2012; Casadevall and
Shenk 2012). In contrast, it was claimed that, although
the results are valuable, publishing the results would not
3. To publish or not to publish? bring about any additional benefit for pandemic prepared-
Fouchier and Kawaoka, the two corresponding authors ness as resources to detect the virus and support programs
of the studies, defended conducting, publishing and for a number of the H5N1 endemic countries are often
continuing the work on mammalian transmissible H5N1 lacking (Osterholm and Relman 2012).
virus strains by emphasizing the benefits of their work Furthermore, it was argued that publishing the papers
and by stressing the importance of their research for in full may result in more laboratories working with the
pandemic preparedness. Their proof-of-principle experi- virus, possibly some with insufficient safety standards,
ments show that few mutations could enable the virus thereby increasing the chance of an unintentional release
to be transmitted via aerosols amongst mammals, (Inglesby et al. 2011; Osterholm and Henderson 2012).
possibly humans, while maintaining virulence (Herfst Other opponents of publishing claimed that sharing the
et al. 2012b; Imai et al. 2012). According to the authors, results and methodology with a small group of selected
the study highlighted the need to monitor H5N1 outbreaks experts would be sufficient to secure beneficial outcomes
in poultry with more urgency, to implement more (Osterholm and Relman 2012; Fouchier et al. 2012b).4 of 14 . S. E. Glatter
Table 1. H5N1 debate: Summary of arguments for and against publication of research methodology
Pro publishing Reference Contra publishing Reference
Benefits for pandemic Fouchier et al. 2012b; Herfst et al. 2012a; Knowing the results (mutations) Osterholm and
preparedness: public health Perez 2012; Palese and Wang 2012; is sufficient (no enhancement Relman 2012
threat of H5N1 strains Kawaoka 2012; Morens et al. 2012 of disease surveillance or
(mammalian transmission, countermeasure availability
maintenance of virulence), expected in the near future)
enables monitoring of current
H5N1 strains for critical
mutations, need for
enhancement of current
preparedness plans ‘Nature is
more inventive than men’
Accessibility of the methodology Herfst et al. 2012a; Perez 2012; Palese and Access to the complete methods Osterholm and
through textbooks and other Wang 2012; Morens et al. 2012 and results makes it easier for Relman 2012
open sources such as published scientists to synthesize a mutant
papers A/H5N1 virus strain (increase in
bioterrorism threat through
easier access)
Need for trained experts and Palese and Wang 2012; Racaniello 2012; Replication not difficult in Zimmer 2012
high-tech research facilities to Fouchier 2012 technological terms
replicate results
Sharing the results in the wider Racaniello 2012; Bouvier 2012 Sufficiency of sharing the results Osterholm and
scientific is a basic requirement with a small group of selected Relman 2012;
for scientific endeavor (value experts, even benefit of that Fouchier et al.
of free science) is questionable 2012c; Osterholm
and Henderson
2012
Prerequisite for ensuring quality Bouvier 2012
of scientific research by enabling
post-publication peer-review and
replication (value of high quality
in science)
Barrier for researchers to continue Perez 2012; Palese and Wang 2012; Encouragement for more labs Osterholm and
the research, need for Kawaoka 2012; Morens et al. 2012; to initiate dual-use research Henderson 2012;
publication to ensure Bouvier 2012; Faden and Karron 2012; on H5N1 (increase of the Inglesby et al.
further research and Casadevall and Shenk 2012 biosecurity and biosafety threat) 2011
prospective benefits
This has been rejected for the same reasons as restricting when the possible outcomes and their respective
publication (Morens et al. 2012). It was also criticized on probabilities are known.
the grounds that, while providing a main barrier to With regards to the H5N1 case, especially at the time of
research, the measure would be futile in the long term, publication, the likelihood with which the risks and
given that the methodology can be found in virology benefits of the research will come to pass is unknown.
textbooks (Palese and Wang 2012; Morens et al. 2012; For example, to estimate the likelihood of misuse one
Perez 2012; Herfst et al. 2012a) (for a summary of the needs access to information on the degree of training,
arguments see Table 1). intentions, financial, and personnel resources of potential
bioweaponeers. Such knowledge is, however, not publicly
available. It will be held by intelligence and security
4. DURC and the precautionary principle
experts, if anyone has access to it at all (Selgelid 2009).
Several groups, including the WHO and the NSABB, have Furthermore, as Rappert (2008) pointed out, risk and
proposed adopting risk–benefit analysis when addressing benefits will be open to a considerable amount of interpret-
DURC issues.18 This recommendation has been followed ation. This is reflected in the discourse on the H5N1 case.
throughout the discourse around the H5N1 studies. For example, experts do not agree on how ‘dangerous’ the
However, as pointed out by Andrew Stirling, a risk mutated virus would be to human health, because data
expert, under conditions of uncertainty and ignorance from the ferret model cannot be directly extrapolated to
risk–benefit analysis fails (Klinke et al. 2006; Stirling and the human context (Bouvier 2012; Racaniello 2012; Fauci
Scoones 2009). It is traditionally used for routine risks, and Collins 2012).Dual-use research and the H5N1 bird flu . 5 of 14
In order to address the challenges associated with Along these lines, Resnik (2003) argues that:
DURC, the bioethicists Kuhlau et al. (2011) have
. . . the principle can be scientific provided that (1) the threats
proposed to make use of the PCP as a ‘rule of choice’ addressed by the principle are plausible threats, and (2) the
with regards to DURC: precautionary measures adopted are reasonable.
When and where serious and credible concern exists that He proposes to use epistemic criteria to determine whether
legitimately intended biological material, technology or know-
a threat is plausible and to use practical considerations to
ledge in the life sciences pose threats of harm to human health
and security, the scientific community is obliged to develop,
determine whether a response to a threat is reasonable.
implement and adhere to precautious measures to meet the Considering the ignorance of risks and benefits
concern. associated with DURC studies such as the H5N1
research, the PCP could be a useful tool to analyze the
Correspondingly, the NSABB members Osterholm and reasonableness of preventing or banning the publication
Relman have, in the light of the potentially disastrous of results and methods as a precautionary measure.
consequences associated with the release of a human trans- When applying the PCP using Resnik’s and Sandin’s
missible H5N1 virus, suggested to err in favor of ‘do no framework, there are two main questions to answer.
harm’. They proposed to make: First, how plausible are the threats associated with pub-
lishing the H5N1 studies? Second, is censorship a reason-
. . . use of the precautionary principle when approaching the able mean to address the threat and prevent possible
issue of DURC. (Osterholm and Relman 2012) negative consequences?
In order to determine whether publishing the H5N1
The PCP is designed for making practical decisions under papers in full presents a reasonable threat, one needs to
uncertainty when no reliable quantitative data is available. discuss whether the engineered H5N1 strain could, in prin-
It was originally created as a way to respond to environ- ciple, cause a disastrous pandemic and whether the H5N1
mental risks (Goklany 2001). For example, the PCP has publications could easily be misused by those who seek to
been used to address issues such as: global warming, in- do harm. In this context, the PCP does not specify the
tensive dichlorodiphenyltrichloroethane (DDT)19 use, and degree of scientific evidence required to determine what
the commercialization of genetically modified food. The a reasonable threat might be. To solve this issue, Sandin
principle has often been criticized for being too vague, et al. (2002) proposed to use a degree of evidence in
overly risk-averse and an unscientific approach to qualitative terms such as ‘scientifically supported strong
decision-making due to a lack of a scientific basis suspicions’ to evaluate the threat.
(Brombacher 1999; Gray and Bewers 1996). Contrasting Thus, what is the scientific evidence supporting the sus-
this critique, risk experts argue that under conditions of picion that engineered H5N1 strains could pose a reason-
uncertainty,20 ignorance21 and ambiguity22 sound scientific able threat to human health? To begin with, it is true that
methods of risk assessment have in any case little justifica- the H5N1 virus infects humans, although transmission
tion (Stirling 2007; Hoffman-Riem and Wynne 2002). It is between humans remains limited (Yang et al. 2007).
under these conditions that the PCP can be viewed as a Another source of concern is the official mortality rate of
general normative guide for policy-makers to acknowledge about 60% (Beigel et al. 2005).23 Virology experts, such
the benefit of doubt (Stirling 2007). as Palese, Bouvier, and Taubenberger believe the rate to be
The critique of the PCP may also be related to the fact overestimated (Bouvier 2012; Palese and Wang 2012).24
that the principle does not define which types of hazards it The true case fatality rate of the unmodified H5N1 virus
applies to, what level of (scientific) evidence is required, is estimated to be between 1% and 33% (Li et al. 2008;
which preventive measures should be taken and with Morens et al. 2012). In comparison, the fatality rate of the
what force they are recommended (Sandin et al. 2002). 1918 pandemic virus was estimated to be 2.5% (Li et al.
However, Sandin, a bioethicist, argues that the PCP does 2008). The worst human influenza pandemic in history so
not contradict science and has recommended establishing far, raged from 1918 to 1919 and claimed an estimated
the following criteria when deciding whether to use the 50 million lives (Taubenberger and Morens 2006).
PCP (Sandin et al. 2002; Sandin 2009): However, even if the true mortality rate of the mutant
virus strains was much below the rate of the 1918 virus,
An action A is precautionary with respect to something pandemic spread of a deadly H5N1 variant would be a
undesirable U, if and only if 1) A is performed with the
cause for significant public health concern.
intention of preventing U, 2) the agent does not believe it to
Further data comes from the animal model in which
be very likely that U will occur if A is performed, and 3) the
agent has externally good epistemic reasons a) for believing human influenza viruses cause influenza-like illness
that U might occur, b) for believing that A will in fact at (Maher and DeStefano 2004). Ferrets are susceptible to
least contribute to the prevention of U, and c) for not believing infection with the mutated H5N1 virus. According to the
that it to be certain or very likely that U will occur if A is two studies, the virus can be transmitted amongst ferrets
performed. via aerosol while maintaining virulence (Herfst et al.6 of 14 . S. E. Glatter
2012b; Imai et al. 2012). Hence, the mutant virus has the reports covering the H5N1 case, for example by the New
potential to infect and spread amongst humans. However, York Times (Zimmer 2012).
it should be noted that the results of these infection studies Indeed, as has been pointed out, life science research has
cannot be used directly to predict the transmissibility and evolved rapidly over the past decades. The emergence of
pathogenicity of the virus in humans (Bouvier 2012; gene technology, the development of high-throughput
Morens et al. 2012; Racaniello 2012; Fauci and Collins DNA synthesis and DNA sequencing machines, the com-
2012). pletion of the human and other genome projects have sig-
In conclusion, the threat is plausible: it is possible that nificantly enhanced research efficiency and opened new
the virus could cause a pandemic that would be devastat- possibilities. However, by taking this view, two important
ing to the world’s population, although there is no (and aspects are overlooked. First, it ignores the significant
there will not be) concrete scientific evidence available. financial means and trained specialists required to operate
Next, the second question needs to be answered: is pre- these new technologies. For example, to ensure constant
venting or restricting publication a reasonable response access to this vital scientific and technological expertise
to the threat of misuse? Would it significantly contribute and to help to rationalize institutional resources, more
to the prevention of the feared consequences? and more so-called ‘core service facilities’ are being estab-
lished at major universities with a life science focus (Sa
2008).25 Those facilities are operated by staff scientists
5. The role of tacit knowledge
with various backgrounds and expert training ranging
Many scientists have argued that preventing publication from bioinformatics to molecular biology, chemistry and
of the results and methodology does not prevent misuse. engineering.
They argued that even if the methodology was omitted Second, the non-technological component of research is
from the paper, an individual with the relevant training overlooked. In addition to sophisticated machines and
would be able to repeat the experiments. For example, technologies, there are many more steps involved when
Fouchier and colleagues, stated that their methodology performing life science research. These often require very
can be found in many virology textbooks (Herfst et al. specific expertise and technical skills of the individual re-
2012a): searcher that can only be learned through constant
Individuals with bad intentions do not need to read the details practice and observation of peers. Hence, even if ‘labora-
in our manuscript because the methods for creating similar tory processes have become more automated’ as Chyba
viruses have already been published widely. (2006) pointed out, it does not follow that ‘less and less
tacit knowledge is needed’ to conduct life science research.
A similar argument was used by Wimmer (2006): Tacit knowledge was, and still is, an important factor not
. . . all methods used for the synthesis of poliovirus were pub- only for the success of life science research, but for science
lished long before the experiment was conceived. Thus, we and technology in general.
neither described new technologies to synthesize DNA nor The term tacit knowledge was first coined by Polanyi
invented novel methods to convert cDNA into infectious who defined it as:
viral RNA.
. . . things that we know but cannot tell. (Polanyi 1962)
This argument contrasts the current discourse about
DURC, which assumes a direct correlation between Since the 1960s much research has been done on the
technological advances in the life sciences and an increased concept of tacit knowledge. Different forms have been
threat of bioterrorism. By focusing on the ease of diffusion defined including somatic-limit and collective tacit know-
of knowledge and technologies, socio-technical aspects of ledge,26 which are widely accepted, for example in the field
biotechnology such as tacit knowledge, complexity and of economics, psychology, and education.
contingency are overlooked and possible challenges are Scholars have attempted to describe and measure the
disregarded, thus leading to the perception of an increased impact of tacit knowledge on the success of academic
threat (Vogel 2008). Moreover, relying on this linear model endeavors and the translation of academic knowledge into
of technical change may lead to a false prioritization of industrial applications (Leonard and Insch 2005; Somech
policy measures and an inappropriate focus on high-tech and Bogler 1999; Zucker et al. 2002). For example, Harry
research (McLeish and Nightingale 2007; Vogel 2008; M. Collins, a social scientist, has presented an insightful
Leitenberg 1999). analysis of the importance of tacit knowledge for scientific
Chyba (2006) has compared the rapid development of research in physics (Collins 2001). He described the attempt
life science technologies to the evolution of computer of Western researchers to repeat experimental results pub-
power. He argued that the pace of development will lished by Russian scientists during the Cold War. Only in
make DNA synthesis and other biotechnologies increas- the 1990s, after closely observing how the experiments were
ingly available to individuals with only basic scientific performed, did the Western researchers succeed in
skills. This view was repeatedly presented in media replicating the results. Collins drew the followingDual-use research and the H5N1 bird flu . 7 of 14
conclusions from his field observations: In order to success- rare? Could the lack of capabilities be a main reason?
fully replicate an experimental result, researchers first have Previous studies on failed and successful attempts to
to understand and acknowledge the importance of tacit develop bioweapons provide possible answers. Two
knowledge. Second, they have to trust the validity of the studies illustrating the impact of socio-technical aspects
results, and third they need to know how difficult a proced- on bioweapons have been presented by Ouagrham-
ure is, meaning how much patience with trial and error is Gormley and Vogel (2010) and by Leitenberg (1999).
required (Collins 2001). Ouagrham-Gormley and Vogel (2010) show that a
The concept of tacit knowledge has not yet gained much substantial part of the Soviet smallpox bioweapon devel-
attention in the biosecurity debate within life science opment program was unable to create a workable biolo-
research. Tucker (2011) and Vogel (2008) are among the gical weapon even though they had substantial expertise,
few who have highlighted the importance of tacit know- resources, time, infrastructure and even detailed protocols
ledge used by life scientists during experimentation. In the at their disposal. They highlight:
context of life science research, Tucker (2011) defines ‘tacit
. . . the local and personal character of bioweapons know-
knowledge’ as knowledge:
ledge, specialized skills, and scientific know-how, which
. . . that cannot be transmitted in writing but must be gained cannot be transferred easily from one person to another and
through years of hands-on experience in the laboratory. from one location to another. (Ouagrham-Gormley and Vogel
2010)
Tacit knowledge plays, as Tucker (2011) recognizes, a
central role when aiming to determine the biosecurity They also concluded that depending on the social context,
risks associated with life science research. tacit knowledge plays an important role not only on the
One of the few studies on tacit knowledge in the life individual, but also on the communal level (collective tacit
sciences has been conducted by Kathleen Vogel. It knowledge). As the most effective policy recommendation
involved members of the Wimmer group who published to prevent former Soviet weaponeers working in the USA
the first de novo synthesis of a virus (Cello et al. 2002; from selling their knowledge, they suggested separating
Vogel 2008). The group revealed that the most difficult them from one another in order ‘to allow the BW-related
part of the poliovirus synthesis was not the partially auto- skills to decay over time’.
mated DNA synthesis. Instead, it was the construction of Leitenberg, a security expert, reached a similar conclu-
the virus: the production of the virus components, the sion on the importance of tacit knowledge for biological
assembly of those components into virus particles, and weapons construction. Despite extraordinary financial
finally the incorporation of the virus’ synthetic genetic resources, access to modern equipment, despite sufficient
material. According to the group, these steps, although time and the fact that there were individuals with graduate
around for more than 20 years ‘. . . require specialized la- and postgraduate training amongst their members, the
boratory know-how and practices only obtained by religious Japanese Aum Shinrikyo group failed to
training, care, and attention to detail [and are difficult produce and disperse botulinum toxin or Bacillus anthracis
even for experienced researchers] (Vogel 2008). That the (Leitenberg 1999; Ballard et al. 2001). Leitenberg
reconstruction or recovery of a pathogen would currently concluded that:
require someone ‘skilled in the art’ has been supported by . . . the experience of the Aum is therefore in marked contrast
other experts including the NSABB Working Group on to the legion of statements by senior US government officials
Synthetic Genomics.27 In the context of the H5N1 and other spokesmen claiming that the preparation of biolo-
studies, Palese and Wang argue that the generation of en- gical agents and weapons could be carried out in ‘kitchens’,
gineered H5N1 based on the published raw sequence ‘bathrooms’, ‘garages’, ‘home breweries’, and is a matter of
would require: relative ease and simplicity.
. . . (i) access to a sophisticated laboratory setting, (ii) profi- In a related study from 1995, the sociologists McKenzie
ciency in relevant concepts of molecular biology, and (iii) ex- and Spinardi (1995) highlighted the importance of socio-
perience with laboratory methods related to viruses. (Palese technological aspects with regards to nuclear weapons
and Wang 2012) development. In their analysis based on interviews with
nuclear weapons designers and computing experts, they
The author of one of the criticized H5N1 studies has used a
emphasized that:
similar argument:
. . . tacit knowledge is also crucial to nuclear weapons
It takes a lot of knowledge, together with perhaps ten years of
development.
training and a well-equipped high containment lab, even to
reproduce our work. (Fouchier 2012) Thus, the work of Ouagrham-Gormley and Vogel,
Leitenberg, and McKenzie and Spinardi indicates that:
In his simplistic portrait of biotechnology research,
Chyba (2006) nevertheless raised an important question: . . . Because tacit knowledge is transmitted person to person,
Why is it that modern bioterrorist attacks28 have been so there are greater barriers to the spread of competence than8 of 14 . S. E. Glatter
the traditional view might lead us to expect. (McKenzie and knowledge needed to reproduce the experimental results,
Spinardi 1995) the task may pose significant hurdles even for specially
Another aspect to be considered in this context is the trained individuals.
accessibility of bioweapons material. Rather than buying Two possible conclusions can be drawn from this
state-of-the-art laboratory equipment and investing great analysis. Unskilled individuals or groups do not possess
the explicit and tacit knowledge currently needed to make
financial resources and time to synthesize a virus from
use of the published methodology of the H5N1 studies and
scratch, pathogens more suited for a bioweapon could,
reproduce the results. On the other hand, omitting the
for example, be stolen or isolated from the wild and sub-
methodology would significantly delay, but not prevent, a
sequently modified to turn it into a category A bioterror-
skilled team of specifically trained scientists with the will,
ism agent as defined by the US Center for Disease
enough time, the right equipment and sufficient financial
Control29 (Herfst et al. 2012a; Tucker 2003). Moreover,
resources from replicating the H5N1 study results. With
a study by Suk et al. (2011) suggests that instead of
this in mind, prohibiting dissemination of the methodology
‘high-tech’ activities, ‘low tech’ activities may be especially
could be regarded as ineffective in reducing the bioterror-
attractive to bioterrorists:
ism threat and as a result may be considered an unreason-
Contamination of food and water, and direct injection/appli- able response to the biosecurity threat of DURC.
cation of a pathogen, all have much lower technical hurdles
and might be expected to be rather more successfully deployed.
6. Discussion
Indeed, in 1984, followers of Bhagwan Shree Rajneeshee,
an Indian guru, deliberately contaminated salad bars at ten This paper outlines that prohibiting publication of the
restaurants in the USA with salmonella resulting in 754 methodology and results is not a reasonable response to
individuals with food poisoning (Wheelis et al. 2006). the uncertain biosecurity threat. It contrast a recent
Furthermore, many scientists have pointed out that analysis published by Resnik (2013) who argues that:
although ferrets are a good model for human influenza
. . . re-dacted publication would have been a reasonable
infection, the results may not be directly applicable to response to the threats posed by the controversial H5N1
humans as the potential for the transmission and lethality papers if not for practical and legal problems.
of engineered H5N1 in humans is unknown (Bouvier 2012;
Morens et al. 2012; Racaniello 2012; Fauci and Collins Resnik seems to support the common assumption that the
2012). Consequently, bioterrorists run the risk of discover- papers:
ing that the mortality and transmission rates of the engin- . . . provide terrorists with a recipe for making a bioweapon.
eered strains are comparable to a seasonal influenza after
having invested much time and resources in replicating the However, I would question that assumption and argue that
H5N1 experiments. replicating the H5N1 experiments is not very easy for in-
Thus, it seems that attempting to replicate published dividuals or groups with minimal training, or even for
dual-use results may pose more challenges than the trained specialists. Neither is replicating the H5N1
common discourse reflects. Considering the importance studies the most suitable and easiest way to obtain a bio-
of tacit knowledge, an individual with only minimal weapon. However, for individuals with specialty know-
training should not be able to repeat the experiments ledge in influenza virology, this will not provide a real
even if the methodology was published. The more barrier. With enough time and resources at hand, they
pressing question is how easily could a professional virolo- do not need to rely on the short methods section in the
gist replicate the results and produce sufficient amounts of H5N1 papers, but will be able to combine their expert
the virus? First, withholding the methodology would likely knowledge with textbook methodology in order to repli-
result in a need for more time for ‘trial and error’. Another cate the experiments.
factor that comes into play is the form and length of the Although replicating the results of the H5N1 studies is
methodology sections in papers. The explanations in such doable for trained individuals even without access to the
sections are usually short and of limited depth due to space methodology, omitting this information will provide a
limitations, besides only providing information on explicit barrier for well-intentioned scientists. As the current
knowledge. Commonly, researchers exchange more system of academic research is built on openness and the
detailed protocols or even visit each other in their exchange of study results, classification of results and
laboratories to be able to replicate published results. potential funding cuts may divert scientists to other
Furthermore, tacit knowledge may be essential for research fields. Prohibiting publication may even impose
translating or adapting technologies to a new environment, new risks. As Sandin et al. (2002) point out, the PCP
not only on an individual level, but also on a communal should also:
level (Ouagrham-Gormley and Vogel 2010). In summary, . . . be applied to precautionary measures prescribed by the
depending on the degree of training and the tacit principle itself.Dual-use research and the H5N1 bird flu . 9 of 14
Less scientific research may mean a lack of crucial know- This analysis does not conclude that we should ignore
ledge for pandemic preparedness necessary to address the DURC problem, as it would not be a reasonable
threats imposed by future natural human H5N1 pandemics. response to the threat. Preventing terrorists and states
For example, the development of counter-measures (e.g. from using bioweapons is important. The DURC
vaccination) may be prevented, thereby precluding society problem calls for a more detailed investigation of the
from obtaining the benefit of the research. This is especially biosafety and biosecurity risks associated with DURC
worrisome as the prospective of benefits for public health that includes a fact- and case-based assessment of the po-
are the motivation of such research and justification for tential for misuse. In this respect, the H5N1 papers could
public funding. In their analysis of the case, the bioethicists be regarded as a wake-up call for the scientific community.
Faden and Karron (2012) refer to a: Although the dual-use issues associated with life science
research have been discussed for the last decade, the
. . . moral obligation to ensure that the results of that research
H5N1 example illustrates how the restriction of research
are used to help reduce risks to global health.
funding and publication could become a reality in the near
There are further risks from limiting scientific openness. For future. In response to the H5N1 case, the US government
example, sharing the results is a prerequisite for responsible has issued a new policy that allows research funding pro-
conduct in science as peer review significantly contributes to posals sent to government agencies to be screened for
ensuring the quality of the research and the research teams. DURC. It also allows for the classification of research
Preventing this mechanism might have a negative impact results and the restriction of funding of research that
on the quality of the research and thus may increase the raises concerns.31 However, how research proposals will
likelihood of viral escape from the laboratory. be evaluated, that is how risks and benefits will be
Regarding the biosafety concerns associated with a assessed and who will evaluate the proposals, remains
higher number of researchers working with the engineered unclear.32 Major differences in assessment of the term
strains—this issue may better be solved through enhanced ‘dangerous’ in association with DURC might either
biosafety regulations than through censoring the research, result in the underestimation of risks and threats or may
given the potential negative implication of censorship. lead to the restriction of legitimate academic research.
Although it was discussed that enhanced Biosafety Level Thus, it has become more important than ever for the
3 (BSL 3) conditions are sufficient (Casadevall aned Shenk scientific community to address public concerns over
2012; Herfst et al. 2012a), Imperiale and Hanna (2012) DURC by establishing a transparent governance sys-
proposed using BSL4 conditions until more evidence is tem (see Miller and Selgelid (2008) for an overview of
available on the danger of the virus. A further biosafety possible options).
measure would be to plan ahead and use a less virulent Governance options that restrict and control access to
strain of the H5N1 virus. In fact, one of the two labs, in research results, as for example proposed by experts from
which the controversial H5N1 studies have been per- the Center for International and Security Studies at
formed, did decide to use less virulent strains for their ex- Maryland (Steinbruner et al. 2007), may not be useful in
periments (Imai et al. 2012). the light of the analysis presented above. For ensuring
In contrast to censorship, education and personnel oversight, the WHO has been suggested as an ideal gov-
screening to identify individuals with malicious intentions ernance body (Fouchier et al. 2012b). However, based on
may be more reasonable and effective measures to mitigate decades of experience with regulating smallpox research,
biosecurity risks. For example in the USA, the Public Health the WHO may not be the ideal choice to prevent deliber-
Security and Bioterrorism Preparedness and Response Act ate misuse of DURC due to financial and political
(PL107-188) from 2002 requires that persons seeking to dependencies and a strong influence of the pro-research
possess, use, or transfer agents and toxins listed on the fraction (Tucker 2006).
‘select agents and toxins’ list must undergo registration and Nevertheless, the scientific community has a duty to
must have a legitimate need to handle or use such agents and consider the negative implications of their research and
toxins.30 Finally, secrecy is an ineffective tool of prevention. thus must find a way to address and possibly regulate
Before publication, the information on the H5N1 experi- DURC issues (Kuhlau et al. 2008). As has been repeatedly
ments had been available to many scientists and the editors suggested, scientists should receive education in bioethics
of the two journals. Even if the US-based journal, Science, and familiarize themselves with ‘experiments of concern’ as
had been forced to follow the original NSABB recommen- defined by the Fink report and the BTWC (Dando et al.
dation to not publish the methodology, the UK-based 2008; Pearson and Mahaffy 2006; Rappert et al. 2006). If
Nature could have (and did) proceed with publication. possible, plans to mitigate potential risks, for example
There are many other journals and editors who do not fall intensified security measures and personnel training and
under US legislation and may be willing to publish such screening, must be prepared before the results of DURC
manuscripts. Redaction of the papers would thus not be ef- are communicated to the public.
fective in reducting the threat as it would not prevent dissem- However, scientists will not be able to ensure that their
ination of the sensitive knowledge. research will never be misused. Thus, it is important that10 of 14 . S. E. Glatter
the research is conducted with the intention of benefitting 7. National Science Advisory Board for Biosecurity
the public. The peer review system of science, which can (2007), About NSABB accessed 11
stricted, needs to ensure that only such research is sup- September 2013.
ported. Furthermore, the potential benefits of the 8. National Institute of Health. Press statement on the
research need to be explained to the public as best as NSABB review of H5N1 research, 2011 accessed 11
facts. Scientists will need to explain in their papers, and September 2013.
possibly with accompanying media campaigns, the rele- 9. World Health Organization. FAQs: H5N1 influenza,
vance and potential benefits of their DURC to ensure Q1: What is H5N1? accessed 11 September 2013.
10. The Fouchier lab reported fatal infections of ferrets
Acknowledgments following ‘intratracheal inoculation’ with the
The author wishes to thank Anna Genske, Bernice Elger mutated H5N1 virus (Herfst et al. 2012b). However,
and the anonymous reviewers for their valuable input. intratracheal inoculation is a method that:
. . . is not directly relevant to viral transmissibility or
Funding natural pathogenesis. [. . .] Influenza viruses that are
otherwise considered to be of low pathogenicity
This work was supported by the Forschungsfonds often induce severe and even fatal disease in
University Basel and the Käthe-Zingg-Schwichtenberg animals when administered by this route. (Morens
Fonds of the Swiss Academy of Medical Sciences. et al. 2012: 336)
11. A moratorium can be defined as a period of time
Notes during which a certain activity is not allowed.
1. The list includes a selection of viral, bacterial and 12. 14 April 2012, statement on NSABB’s recommenda-
fungal species as well as toxins. An updated list of tions to the National Institute of Health on H5N1
select agents is accessible at accessed 11 September 2013. September 2013.
2. Uniting and Strengthening America by providing 13. See accessed 11 September 2013.
Obstruct Terrorism (USA Patriot) Act, 2001 Public 14. Biosafety includes:
Law 107–56, and the Public Health Security and . . . containment principles, facility design, practices
Bioterrorism Preparedness and Response Act, 2002. and procedures to prevent occupational infections
3. National Institute of Health (2012) United States in the biomedical environment or release of the
Government Policy for Oversight of Life Sciences organisms to the environment. (Nordmann 2010)
Dual Use Research of Concern accessed 11 September 2013. . . . the sum of risk management practices in defense
4. US Government, Screening Framework Guidance for against biological threats’, which includes aversion
Synthetic Double-Stranded DNA Providers, of biological terrorism and other disease breakouts.
Department of Health and Human Services, Office (Meyerson and Reaser 2002)
of the Secretary Federal Register Vol. 74, p. 224 16. The press has also discussed how easily the results
accessed 2012).
11 September 2013 17. See accessed 11 September 2013.
Statement on biosecurity, Trieste, Italy accessed 11
September 2013. Just as scientists should weigh benefits against
6. The report committee was headed by Professor Gerald the risks when deciding which projects to pursue,
R. Fink, Professor of Genetics, Whitehead Institute, governments should be encouraged to do the same
MIT, Boston, MA. when deciding which projects to fund.Dual-use research and the H5N1 bird flu . 11 of 14
The NSABB (2007) stated that: Synthetic_Genomics.pdf> accessed 11 September
2013.
The NSABB will [. . .] develop guidelines for the
28. According to the US Center for Disease Control:
oversight of dual use research, including guidelines
for the risk-benefit analysis of dual use biological . . . a bioterrorism attack is the deliberate release of
research and research results. viruses, bacteria, or other germs (agents) used to
cause illness or death in people, animals, or plants.
19. DDT was extensively used as an insecticide in agricul- ( accessed 11 September 2013)
20th century. Today, due to concerns over its potential
toxic side effects, intensive use of DDT is banned 29. The US Center for Disease Control separates bioter-
worldwide. rorism agents into three categories, depending on how
20. Under the condition of uncertainty, possible outcomes easily they can be spread and the severity of illness or
can be characterized, but the available information or death they cause. Category A agents are considered
analytical models do not present a definitive basis for the highest risk, because they can be easily spread or
assigning probabilities (Stirling 2007). transmitted from person to person, they result in high
21. Under ignorance neither probabilities nor outcomes death rates and have the potential for major public
can be fully characterized (Stirling 2007; Collinridge health impact, they might cause public panic and
1980). social disruption, and they require special action for
22. Under the condition of ambiguity, outcomes are prone public health preparedness accessed 11
2007). September 2013. With a few exceptions, there are
23. World Health Organization. FAQs: H5N1 influenza, more steps required beyond the synthesis and replica-
Q1: What is H5N1? (2012) accessed 11 September 2013. ization and dispersal of the pathogenic material
24. The mortality rate is based on diagnostically confirmed (Tucker 2003). For example, the pathogen Bacillus
cases of human H5N1 infections in hospitalized anthraces could be isolated from the wild as it is still
patients in Southeast Asia. Due to a lack of epidemio- endemic amongst cattle, e.g. in Equatorial Africa.
logic and virologic studies and limited access to health Infections are also sporadically detected in the USA
care in rural areas, mild forms of infection may likely (see accessed 11 September 2013). However, viru-
have retrospectively identified previously undetected lence varies amongst naturally occurring Bacillus
asymptomatic infections (Beigel et al. 2005). anthraces strains, and environmental conditions deter-
25. For example, the Functional Genomics Center Zurich, mine the stability and hence the infectiousness of
a joint state-of-the-art research and training facility of anthrax spores. Bioweaponeers would need to isolate
the ETH Zurich and the University of Zurich, many strains to find a suitable one, and would need to
Switzerland, provides the latest technologies and stabilize the spores (Dragon and Rennie 1995). The
expert support in genomics, transcriptomics, prote- same would be true for naturally occurring H5N1
omics, metabolomics, and bioinformatics accessed 11 September 2013. modified to adapt to humans as a host.
26. According to the social scientist Harry M. Collins, 30. According to the Public Health Security and
somatic limit and collective tacit knowledge are Bioterrorism Preparedness and Response Act of
experienced and acquired by humans through immer- 2002, Public Law 107–188, June 12, 2002, Sec. 351A,
sion in society and guided practice: Somatic-limit tacit pp. 637ff.
knowledge is defined by the limited capacities and . . . the Secretary [. . .] shall include provisions to
particular nature of the human brain and body, and ensure that persons seeking to register under such
collective tacit knowledge regulations have a lawful purpose to possess, use,
has to be known tacitly, because it is located in or transfer such agents and toxins. [. . .] The
human collectivities and, therefore, can never be Secretary shall maintain a national database that
the property of the any one individual. (Collins includes the names and locations of registered
2007) persons, the listed agents and toxins such persons
are possessing, using, or transferring, and informa-
27. NSABB. Addressing biosecurity concerns related to tion regarding the characterization of such agents
the synthesis of select agents (2006)12 of 14 . S. E. Glatter
persons [. . .] deny access to such agents and toxins by Bouvier, N. M. (2012) ‘The science of security versus the
individuals whom the Attorney General has security of science’, Journal of Infectious Diseases, 205/11:
1632–5.
identified as ‘restricted persons.’ Restricted persons Brombacher, M. (1999) ‘The precautionary principle threatens
are defined as individuals reasonably suspected of to replace science’, Pollution Engineering, Summer: 32–4.
‘committing a crime [. . .]’, individuals with Casadevall, A. and Shenk, T. (2012) ‘Mammalian-transmissible
‘knowing involvement with an organization that H5N1 virus: Containment level and case fatality ratio’, MBio,
engages in domestic or international terrorism’ [. . .] 3/2: e00054–00012.
Cello, J., Paul, A. V. and Wimmer, E. (2002) ‘Chemical
or ‘being an agent of a foreign power.’ synthesis of poliovirus cDNA: Generation of infectious
virus in the absence of natural template’, Science, 297/5583:
31. The National Institute of Health (2012) US 1016–8.
Government Policy for Oversight of Life Sciences Chyba, C. F. (2006) ‘Biotechnology and the challenge to arms
Dual Use Research of Concern ( accessed 11 September 2013). States that: knowledge and somatic-limit tacit knowledge’, Organization
Studies, 28: 257–62.
If the risks posed by the research cannot be Collins, H. M. (2001) ‘Tacit knowledge, trust and the Q of
adequately mitigated with the measures above, sapphire’, Social Studies of Science, 31: 71–85.
Federal departments and agencies will determine Dragon, D. C. and Rennie, R. P. (1995) ‘The ecology of anthrax
whether it is appropriate to: [. . .] Classify the spores: Tough but not invincible’, Canadian Vetinary Journal,
research [. . .] Not provide or terminate research 36: 295–301.
Faden, R. R. and Karron, R. A. (2012) ‘Public health and
funding. biosecurity. The obligation to prevent the next dual-use
controversy’, Science, 335/6070: 802–4.
32. The policy demands the following action for research Fauci, A. S. and Collins, F. S. (2012) ‘Benefits and risks of influ-
that involves certain agents and toxins as listed in the enza research: Lessons learned’, Science, 336/6088: 1522–3.
document or/and if the research includes experiments Fouchier, R. A. (2012) ‘Interview with virologist Dr. Ron
of concerns as defined by the Fink report. In case of Fouchier’, BIF Futura, 27/3: 12.
such research it demands federal departments and ——, Garcia-Sastre, A., Kawaoka, Y., Barclay, W. S. et al.
(2012a) ‘Pause on avian flu transmission research’, Science,
agencies to: 335/6067: 400–1.
——, ——, ——, —— et al. (2013) ‘Transmission studies
Assess the risks and benefits of such projects,
resume for avian flu’, Science, 339/6119: 520–1.
including how research methodologies may ——, Herfst, S. and Osterhaus, A. D. (2012b) ‘Public health
generate risks and/or whether open access to the and biosecurity. Restricted data on influenza H5N1 virus
knowledge, information, products, or technologies transmission’, Science, 335/6069: 662–3.
generates risk. (National’ Institute of Health (2012) ——, Osterhaus, A. B., Steinbruner, J., Yuen, K. Y. et al.
(2012c) ‘Preventing pandemics: The fight over flu’, Nature,
United States Government Policy for Oversight of 481/7381: 257–9.
Life Sciences Dual Use Research of Concern Gaudioso, J. and Salerno, R. M. (2004) ‘Science and govern-
accessed 11 September Goklany, I. (2001) The Precautionary Principle: A Critical
Appraisal of Environmental Risk Assessment. Washington,
2013) DC: Cato Institute.
Gray, J. S. and Bewers, J. M. (1996) ‘Towards a scientific defin-
However, no further details are given. ition of the precautionary principle’, Marine Pollution Bulletin,
32: 768–71.
Hall, J. A. (1965) ‘Atoms for peace, or war’, Foreign Affairs, 43:
References 602–15.
Herfst, S., Osterhaus, A. D. and Fouchier, R. A. (2012a) ‘The
Atlas, R., Campbell, P., Cozzarelli, N. R., Curfman, G. et al. future of research and publication on altered H5N1 viruses’,
(2003) ‘Statement on scientific publication and security’, Journal of Infectious Diseases, 205/11: 1628–31.
Science, 299/5610: 1149. ——, Schrauwen, E. J., Linster, M., Chutinimitkul, S. et al.
Ballard, T., Pate, J., Ackerman, G., McCauley, D. et al. (2001) (2012b) ‘Airborne transmission of influenza A/H5N1 virus
‘Chronology of Aum Shinrikyo’s CBW activities’, pp. 1–17. between ferrets’, Science, 336/6088: 1534–41.
Monterey, CA: James Martin Center for Nonproliferation Hoffmann-Riem, H. and Wynne, B. (2002) ‘In risk assessment,
Studies. one has to admit ignorance: Explaining there are things we
Beigel, J. H., Farrar, J., Han, A. M., Hayden, F. G. et al. (2005) can’t know could improve public confidence in science’,
‘Avian influenza A (H5N1) infection in humans’, New Nature, 416/6877: 123.
England Journal of Medicine, 353/13: 1374–85. Imai, M., Watanabe, T., Hatta, M., Das, S. C. et al. (2012)
Berns, K. I., Casadevall, A., Cohen, M. L., Ehrlich, S. A. et al. ‘Experimental adaptation of an influenza H5 HA confers
(2012) ‘Policy: Adaptations of avian flu virus are a cause for respiratory droplet transmission to a reassortant H5 HA/
concern’, Nature, 482/7384: 153–4. H1N1 virus in ferrets’, Nature, 486/7403: 420–8.You can also read
