University of Southern California Dissertations and Theses

Lesser evils? WMD pursuit beyond nuclear weapons

(USC Thesis Other)

Lesser evils? WMD pursuit beyond nuclear weapons

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content Lesser Evils?:
WMD Pursuit Beyond Nuclear Weapons
by
Miriam Barnum
A Dissertation Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(POLITICAL SCIENCE AND INTERNATIONAL RELATIONS)
August 2022
Copyright 2022 Miriam Barnum
Acknowledgements
This project would not have been possible without the help and support of numerous people
and institutions.
First, I am grateful to my dissertation committee: Pat James, Jacques Hymans, and
James Lo have all been incredibly supportive and engaged deeply with my work throughout
my graduate career. Their comments have shaped this dissertation for the better, and will
continue to have an immeasurable impact on my work moving forward. Beyond simply their
impactonthisdissertation,Ihavegrownasascholarthroughopportunitiestoworkwitheach
of them, and have benefited greatly from their mentorship and career advice. Thanks also to
Najmedin Meshkati for generously agreeing to serve as my outside committee member, and
for providing valuable suggestions for connecting my work to policy and practical scientific
audiences.
My experience at USC would have been unimaginably di↵ erent were it not for the
opportunity to work with the Security and Political Economy (SPEC) Lab. Thanks to Ben
Graham, Jonathan Markowitz, and Megan Becker for creating and sustaining such a vibrant
intellectual community. I am particularly grateful to Ben for bringing me into the Lab
during my second year at USC. Not only have I benefitted tremendously from working with
Ben, Jonathan, and Chris Fariss as coauthors, all three of them went above and beyond in
helping me communicate and hone my arguments as I applied for academic jobs. Finally,
I am particularly grateful to SPEC research assistants Lole¨ ı Brenot, Zoe Peach-Riley, and
Micaela Mengen for their invaluable assistance with data collection for this project.
ii
Another community that has been integral to the completion of this project is the US-
Asia Grand Strategy predoctoral fellows program at USC’s Korean Studies Institute. I am
grateful to David Kang for creating this community, and for welcoming me into it. Thanks
to Caleb Pomeroy, Noelle Troutman, Manseok Lee, Pongkwan Sawasdipakdi, Nayoung Lee,
and Jackie Wong for being incredible colleagues and friends this past year.
I have been lucky to pursue my PhD at USC. Beyond the faculty already mentioned,
Brett Carter, Andrew Coe, and Wayne Sandholtz provided crucial advice and support. I
am grateful to many of my fellow students not only for their helpful feedback on my work,
but also for their friendship. In particular: Kelly Zvobgo, Kyuri Park, Gabi Cheung, Kyle
Rapp, Nola Haynes, and Taylor Dalton.
Finally, thank you to my parents, who have always had much more confidence in
my potential to succeed than I have, and to my grandparents, who provided significant
motivation in the form of queries about my graduation timeline.
iii
Table of Contents
Acknowledgements ii
List of Tables iv
List of Figures v
Abstract viii
Chapter 1: Introduction: Weapons of Less Destruction 1
Chapter 2: National Security, Regime Security, and Proliferation Choices 12
Chapter 3: Measuring CBW Proliferation and Use 30
Chapter 4: Drivers of CBW Pursuit: Evidence from New Panel Data 44
Chapter 5: Libya: Substitution in a “Coup-proofed” State 64
Chapter 6: Rhodesia: Counterinsurgent CBW 89
Chapter 7: Conclusion: CBW Research Moving Forward 110
Bibliography 117
Appendix A: Regression Tables 135
iv
List of Tables
3.1 Chemical weapons exploration, pursuit, and possession: comparison of datasets. 40
3.2 Biologicalweaponsexploration,pursuit,andpossession: comparisonofdatasets. 41
3.3 Chemical weapons use by state actors, 1946-present. . . . . . . . . . . . . . . 42
A.1 Security threats and CW pursuit: fixed e↵ ects models . . . . . . . . . . . . . 136
A.2 Security threats and CW exploration: fixed e↵ ects models . . . . . . . . . . 137
A.3 Security threats and CW pursuit . . . . . . . . . . . . . . . . . . . . . . . . 138
A.4 Security threats and CW exploration . . . . . . . . . . . . . . . . . . . . . . 139
A.5 Security Threats and BW pursuit: fixed e↵ ects models . . . . . . . . . . . . 140
A.6 Security threats and BW exploration: fixed e↵ ects models . . . . . . . . . . 141
A.7 Security Threats and BW pursuit . . . . . . . . . . . . . . . . . . . . . . . . 142
A.8 Security threats and BW exploration . . . . . . . . . . . . . . . . . . . . . . 143
v
List of Figures
2.1 Weapons characteristics and theorized proliferation drivers . . . . . . . . . . 19
4.1 E↵ ects of internal security threat indicators on risk of weapons pursuit. Haz-
ard Ratios displayed here are from Cox proportional hazard models including
both internal and external security threat indicators. Models are stratified by
country. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
4.2 E↵ ects of ethnic fractionalization on risk of weapons pursuit. Hazard Ratios
displayed here are from Cox proportional hazard models including both inter-
nal and external security threat indicators. Models are stratified by country. 52
4.3 E↵ ects of external security threat indicators on risk of weapons pursuit. Haz-
ard Ratios displayed here are from Cox proportional hazard models including
both internal and external security threat indicators. Models are stratified by
country. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
4.4 E↵ ects of nuclear infrastructure index on risk of weapons pursuit. Hazard
Ratios displayed here are from Cox proportional hazard models including
both internal and external security threat indicators. Models are stratified by
country. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
4.5 E↵ ects of internal security threat indicators on risk of weapons pursuit. Haz-
ard Ratios displayed here are from Cox proportional hazard models including
both internal and external security threat indicators. . . . . . . . . . . . . . 58
vi
4.6 E↵ ects of ethnic fractionalization on risk of weapons pursuit. Hazard Ra-
tios displayed here are from Cox proportional hazard models including both
internal and external security threat indicators. . . . . . . . . . . . . . . . . 59
4.7 E↵ ects of external security threat indicators on risk of weapons pursuit. Haz-
ard Ratios displayed here are from Cox proportional hazard models including
both internal and external security threat indicators. . . . . . . . . . . . . . 60
4.8 E↵ ects of nuclear infrastructure index on risk of weapons pursuit. Hazard
Ratios displayed here are from Cox proportional hazard models including
both internal and external security threat indicators. . . . . . . . . . . . . . 61
vii
Abstract
This dissertation identifies a set of pathway through which security threats influence chemi-
cal and biological weapons proliferation. In doing so, it moves beyond analytical frameworks
that merely ask how chemical and biological weapons pursuit are related to nuclear weapons
pursuit. Rather, it identifies a key factor — internal security threats faced by a state’s gov-
erning regime — that can both motivate chemical or biological weapons pursuit in states
with little or no interest in nuclear weapons pursuit, and incentivize substitution or diversi-
fication of weapons programs in states that are interested in nuclear weapons. Such internal
securitythreatsarenotwidelyincorporatedintotheoriesofarming, andhighlighttheunique
role that chemical and biological weapons in can play in regimes’ security strategies.
Aquantitativeeventhistoryanalysis,basedonnewpaneldataonCBWpursuit,acqui-
sition, and use, suggests that internal security threats, while often disregarded in theories of
arming,playasignificantroleinmotivatingstatestopursuechemicalandbiologicalweapons
proliferation. In particular, governing regimes facing increases in the risk of a coup may be
more likely to initiate chemical and biological weapons programs, and regimes experiencing
violent domestic unrest may be more likely to pursue chemical weapons. Additionally, this
analysis o↵ ers tentative evidence for two pathways through which external security threats
can motivate CBW proliferation. The first is “reactive proliferation”: states to pursue their
own CBW capabilities in response to a rival’s acquisition of strategic weapons (in particular
nuclear, and sometimes biological). The second is “substitution”: in the face of negative
shocks to industrial and economic resources, states may begin pursue CBW as a hedge
against the failure of a costlier, more technically challenging nuclear weapons program. Two
case studies illustrate how the two di↵ erent types of regime security threats shape chem-
ical and biological weapons pursuit choices. In Libya, Qaddafi’s fear of being overthrown
by opposing elements within the governing elite led him to dismantle much of the existing
state bureaucracy. This “coup-proofed” state struggled to e↵ ectively monitor and manage
alarge,complexnuclearweaponsprogram,makingasmaller-scalechemicalweaponspro-
gram a desirable option to counter a range of security threats. In Rhodesia, an increasingly
desperate struggle against a numerically asymmetrical insurgency prompted chemical and
biologicalweaponspursuit, inthehopesincitingfearamongsttheinsurgentsandthecivilian
population and generating casualties covertly and with relatively little manpower.
Understanding these mechanisms provides a necessary foundation for further inquiry
into the causes and consequences of chemical and biological weapons proliferation. Interna-
tional relations scholars have built up a substantial body of knowledge regarding the causes
and consequences of nuclear weapons pursuit. In comparison, the research on chemical and
biological weapons proliferation has been less systematic and cohesive. Not only does this
dissertation advance our understanding of CBW proliferation, it also provides potential in-
sight into other tradeo↵ smadeinstates’armingdecisions,andintothesubstitutabilityof
foreign policy tools more generally.
viii
Chapter 1
Introduction: Weapons of Less
Destruction
On August 21, 2013, in the midst of the Syrian civil war, rockets carrying the potent nerve
agent sarin struck two locations in Ghouta, a rebel held area near Damascus. The deadliest
use of chemical weapons since the Saddam Hussein’s 1988 attack against Iraq’s Kurdish
population in Halabja, the attack was estimated to have killed at least several hundred, and
possibly more than 1,700 people — and injured thousands more. While this was not the
first use of chemical weapons by Bashar al-Assad’s regime in its fight against the rebels, its
scale prompted an investigation by the United Nations. The UN and established a joint
mission with the Organisation for the Prohibition of Chemical Weapons (OPCW) to oversee
the elimination of Syria’s chemical weapons program. While the mission shipped the last of
Syria’s declared chemical weapons out of the country for destruction in June 2014, apparent
chemical weapons attacks continued through 2018. How and why did the Assad regime
come to possess these weapons, and why were they willing to risk international intervention
not only by using chemical agents against rebels and civilians, but also by maintaining an
undeclared chemical weapons stockpile even after investigations by the UN and the OPCW?
In constructing nonproliferation and national security policy, decision makers often
1
invoke the category of “weapons of mass destruction,” grouping chemical, biological, radio-
logical, and nuclear weapons together — perhaps arbitrarily — and pursuing similar policy
strategies to stem the spread and use of these di↵ erent types of weapons. Following the end
of the Cold War, there has been much speculation about the increasing relevance of chemical
and biological weapons, and the potentially declining relevance of nuclear. However, little
empirical research has been done to examine whether this categorization makes sense as an
analytical construct, and whether nonproliferation lessons learned in the nuclear domain can
betransferredtothedomainsofchemicalandbiologicalweapons(CBW)withoutunintended
consequences.
In contrast to the literature on nuclear proliferation, which initially focused heavily
on reasons why states might want the bomb, and only later turned substantial attention
to factors that might explain the relative rarity of nuclear proliferation, the literature on
chemical and biological weapons has thus far been heavily focused on factors that might
limit their spread, such as arms control regimes, stigmatization, and the di culties of using
such weapons for traditional deterrence. The puzzle, then, is why, despite these factors,
some states nevertheless attempt to acquire chemical and biological weapons. If choosing
to violate international norms and invite external intervention, why not pursue the most
powerful possible weapons options?
One answer is that chemical and biological weapons are often regarded as a potential
“poorman’satomicbomb,” alesse↵ ectivesubstitutefornuclearweapons, pursuedbystates
that desire, but are unable to acquire, a nuclear deterrent. Yet, empirically, we observe a
variety of patterns of WMD pursuit, including chemical and/or biological weapons pursuit
by states that show little or no interest in nuclear pursuit — suggesting that motivations for
CBW pursuit are more complex than mere substitution. In this project, I ask what might
motivate chemical or biological weapons pursuit in states with little or no interest in nuclear
weapons pursuit?
Iarguethatregimeinsecurity—theriskposedtoaregime’ssurvivalbyinternal,rather
2
than external security threats — serves as a powerful motivator for chemical and biological
weaponsacquisition, astheincreased“useability”ofchemicalandbiologicalweaponsrelative
to nuclear weapons and the relatively smaller organizational scale at which chemical and
biological weapons can be acquired make them attractive as a tool to respond to domestic
threats. Furthermore, the internal security environment can mediate the e↵ ect of external
security threats on weapons proliferation choices. I test this argument using new data on
chemicalandbiologicalweaponspursuit, acquisition, anduse, andfindthatregimesaremore
likely to begin pursing chemical or biological weapons when the risk of a coup increases, and
thatregimesfacingviolentunrestandperceivedriskofcivilconflictaremorelikelytopursue
chemical weapons.
Additionally, I theorize and examine evidence for potential alternative drivers of CBW
pursuit: pathwayslinkingstates’externalthreatenvironmentstoCBWproliferationchoices.
These include CBW pursuit as a balancing reaction against rivals’ WMD acquisition, and
CBW pursuit as a hedge against the failure of other, costlier weapons programs.
Examining the evidence for these drivers of proliferation can provide a necessary
foundation for further inquiry into the causes and consequences of chemical and biologi-
cal weapons proliferation, give us insight into decision makers’ choices between military and
foreignpolicytoolsmoregenerally,andaidintheformationofmoree↵ ectivenonproliferation
policy for all types of WMD.
In the remainder of this chapter, I first provide a summary of the theoretical argument
and my empirical approach. Next, I discuss the contributions of the dissertation. Finally, I
provide a brief overview of the remaining chapters.
The Argument in Brief
While arming is typically understood as a response to external security threats, I argue that
leaders choosing to pursue chemical and biological weapons pursuit are likely motivated by
3
internal security threats as well. Building on work by Koblentz (2013), who in turn builds
on David (1991a), Bueno De Mesquita (1999), Schweller (2010), Hagan (1987), and Jackson
(2007), I divide security threats into two broad categories: national security threats, the
external threats to state survival with which realists have traditionally been concerned—
and which have been the primary focus of the proliferation literature—and regime security
threats, internal threats to the continued survival of a particular governing regime within
the state—which have been less emphasized in the proliferation literature to date.
Understandwhychemicalandbiologicalweaponsmightbemoredesirableoptionsthan
nuclearweaponstocountercertaintypesofsecuritythreatsrequiresfirstacknowledgingthat
the concept of “weapons of mass destruction” is not necessarily well suited to capture all
relevant aspects of chemical and biological weapons.
1
Accordingly, my theory takes as its
starting point two key aspects in which chemical and biological weapons di↵ er from nuclear
weapons: CBW are both less destructive and easier to acquire than nuclear weapons. These
di↵ erences, which tend to make weapons relatively more attractive tools for responding to
internalsecuritythreats,aremostpronouncedforchemicalweapons,withbiologicalweapons
falling somewhere in between chemical weapons and nuclear weapons in terms of both de-
structiveness and di culty of acquisition. Thus, while we would expect internal security
threats to play more of a role in motivating both chemical and biological weapons pursuit
than they do in motivating nuclear weapons pursuit, we should expect biological weapons
pursuit to be relatively less motivated by internal security threats (and more motivated by
external threats) when compared to chemical weapons.
1
On conceptual stretching, see Sartori (1970). While “WMD” as a category in the classical
sense struggles to encompass CBW, it might be possible to conceive of “WMD” as a radial cate-
gory, centered on nuclear weapons, with biological and chemical weapons in secondary or tertiary
positions (Collier and Mahon, 1993).
4
Regime Security Threats
Domesticthreatstoastate’sleaderorgoverningregimecaneithercomefromeitherfromthe
general population (through rebellion and potentially civil war), or from political or military
elites (through a coup d’´ etat). Each of these threats can motivate chemical and/or biological
weapons pursuit for di↵ erent reasons.
Chemical and biological weapons’ relative lack of destructiveness—the very feature
that makes them poor strategic weapons—make them an attractive option for responding
to domestic threats, particularly violent unrest, insurgency, and civil war. Because they are
primarily useful for casualty generation and provoking terror, and are not well suited to the
destruction of infrastructure or hardened military targets, CBW may play useful roles in
civil conflict or internal repression, where the goal is to eliminate, incapacitate, or demor-
alize adversaries without destroying one’s own territory and infrastructure. Particularly in
domestic contexts, threats to use CBW may be more credible than threats to use other,
more destructive weapons.
For leaders concerned about the risk of a coup, chemical and biological weapons o↵ er
a military option that can be acquired and used under relatively centralized control. “Coup-
proofing” strategies — institutional changes undertaken to consolidate a leader’s power and
reduce the risk of challenges from within the military or state bureaucracy — tend to have
detrimental consequences for both military e↵ ectiveness and state capacity. Because they
lendthemselvestocentralizedcontrol, chemical, biological, andnuclearweaponscanprovide
ameansofrecoupingmilitarye↵ ectiveness. However, a coup-proofed state may struggle
to managing an e↵ ective nuclear weapons program. In this environment, chemical and
biological weapons o↵ er a means of regaining some centrally controlled military capability
with relatively lower technical and organizational hurdles.
5
National Security Threats
There are several ways in which external threats could motivate chemical and/or biological
weapons pursuit. First, leaders may believe that CBW can fill a specific need for balancing
againstadversaries(distinctfromneedsfilledbypotentialnuclearcapabilities). Ontheother
hand, potential proliferators may view chemical and biological weapons as fulfilling (albeit
less e↵ ectively) similar external security needs as nuclear weapons.
When facing adversaries that possess CBW capabilities, states may hope to deter
the use of those weapons by acquiring the capability to retaliate in kind. Chemical or
biological retaliation is perceived as more severe than limited conventional retaliation, while
remaining more credible than nuclear retaliation. Where a rival has not only possesses, but
has demonstrated a willingness to use CBW in previous conflicts, incentives to acquire CBW
are heightened.
Additionally, states may view chemical and biological as imperfect substitutes for nu-
clear weapons. While chemical and biological weapons may be poorer providers of strategic
deterrence than nuclear weapons, they can likely provide a minimal “last resort” deterrent.
Even if a state would ideally prefer to acquire nuclear weapons, it may face constraints on
its ability to actually acquire those weapons. Decision makers in such states may seek to
“diversify”—pursuing multiple weapons options, in the hope that at least one program will
be successful.
Empirical Approach
In order to evaluate this argument, I employ a mixed methods approach. Large-n time-
series cross sectional analysis leverages original data collection to reveal systematic patterns
in states’ choices to pursue CBW, while in-depth case studies of several countries’ CBW
programs shed light on the mechanisms underlying CBW pursuit choices.
For the quantitative analysis, I developed two new datasets: one covering characteris-
tics of the specific CBW agents and delivery systems states pursue, and another on states’
6
histories of WMD use. These data reveal that internal threats may play a significant role
in arming choices. In particular, governing regimes facing increases in the risk of a coup
are more likely to initiate chemical and biological weapons programs. Furthermore, regimes
experiencingviolentdomesticunrest, andconcernedabouttheriskofcivilwar, maybemore
likely to pursue chemical weapons. This analysis also provides evidence for CBW pursuit as
asubstitutefornuclearweaponspursuit. Statesaremorelikelytoinitiatechemicaland/or
biological weapons pursuit after a rival acquires nuclear weapons. Additionally, holding
threat levels constant, states facing negative shocks to industrial and economic resources
may be more likely to pursue CBW — perceived as “cheaper” and easier to acquire — as a
substitute for other proliferation options.
Notably, while they are still relevant to internal security, biological weapons may be
perceived by potential proliferators as more similar to traditional “strategic weapons” than
chemical weapons. States appear to balance other state rivals’ bioweapons acquisition by
pursuing their own chemical and bioweapons capabilities, and domestic threats from the
general public do not appear to motivate bioweapons pursuit.
While this quantitative analysis provides evidence of broad patterns in CBW pursuit
choices, it cannot confirm that leaders choosing to pursue these weapons actually perceive
particulartypesofsecuritythreats, orthatabeliefthatchemicalorbiologicalweaponscould
be useful for countering such threats is actually driving pursuit choices. Thus, for evidence
of these mechanisms, I turn to in-depth case studies of changes in chemical and biological
weaponsproliferationbehaviorwithinindividualcountriesovertime. Becausethegoalofthe
case study analysis is to shed light on the pathways through which security threats influence
proliferationbehavior, Iselect“typical”casesthatfeaturetheexpectedrelationshipbetween
internalsecuritythreatsandCBWpursuit(SeawrightandGerring,2008;WellerandBarnes,
2016). I focus here on Libya and Rhodesia, two countries that faced unusually high levels
of both internal and external threat, in order to illuminate which factors play a larger role,
and the potential interactions between internal and external security factors.
7
Libya pursued both nuclear weapons (unsuccessfully) and chemical weapons (success-
fully), and at least considered pursing biological weapons, while Rhodesia pursued and ac-
quired both chemical and biological weapons, making these cases an ideal opportunity to
assess how decision-makers integrate di↵ erent types of perceived security threats and eval-
uate trade-o↵ s between di↵ erent weapons proliferation options. Furthermore, the two cases
together provide the opportunity to examine the two theorized internal security motivations
for CBW pursuit: coup-proofing in Libya and counterinsurgency in Rhodesia. Finally, both
represent instances of CBW programs that were ultimately dismantled, and the governing
regimesthatranthoseprogramsarenolongerinpower. Thus, informationontheseweapons
programs is relatively accessible, compared to information on ongoing programs, or program
that were halted but never declared, which are typically shrouded in intense secrecy.
Contributions
This dissertation contributes to our understanding of chemical and biological weapons pro-
liferation in several areas.
First, it provides a systematic test of regime security theory across all countries in
the post-World War II time period. While Koblentz (2013) finds evidence that domestic
threats motivated CBW pursuit and use in South Africa and Iraqi, there has been no more
systematic investigation of the theory across a larger sample of countries, or to asses the
significance of the role played by regime security concerns relative to other factors. While
some recent large-n work has considered the e↵ ects of regime insecurity on nuclear weapons
proliferation, these studies tend to approach regime insecurity through the lens of fears of
externally imposed regime change. Furthermore, they do not asses how regime security
shapes the relative desirability of di↵ erent weapons options.
Additionally, the theoretical framework presented here adds nuance to existing cross-
nationalstudiesofchemicalandbiologicalweaponsproliferationbyidentifyingspecifictypes
8
of regime security and national security threats that the unique features of CBW are suited
to address, and examines the evidence that security threats are motivating CBW pursuit via
these distinct pathways. In doing so, it illuminates potential interactions between regime
securityandnationalsecuritymotivationsforweaponsproliferation. Inparticular,itsuggests
thatobserveddynamicsof“substitution”betweennuclearweaponsandchemicalorbiological
weapons may be partly driven by the e↵ ects of domestic security concerns on the costs and
feasibility of nuclear weapons pursuit.
Finally, this dissertation presents new data on CBW programs, capabilities, and use.
Existing datasets covering chemical and biological weapons proliferation tend to cast a wide
net, employing a relatively broad concept of weapons pursuit, resulting in the inclusion of
some observations of CBW pursuit or acquisition that are not conceptually appropriate to
test a theory of leaders’ motivations for initiating chemical weapons pursuit. I re-evaluate
these codings to produce a more limited dataset that is more directly focused on the pursuit
and acquisition of deliverable chemical and biological weapons, suitable to use in combat,
and operationally controlled by the government of the state in question. I then use this data
in combination with existing data to distinguish between correlates of CBW exploration and
correlatesCBWpursuit—adistinctionwhichhasbeenprevalentintheliteratureonnuclear
proliferation, but, to date, has not been applied in quantitative studies of CBW pursuit.
In order to measure one of the independent variables in this study, rivals’ history of
chemical weapons use, I have constructed a dataset of incidents of chemical weapons use in
inter- and intra-state conflict, as well as assassinations, including details on agents, delivery
methods, and targets. To my knowledge, this is the first comprehensive dataset on chemical
weapons use that includes use in intrastate conflict as well as interstate conflict.
9
Structure of the Dissertation
In the body of the dissertation, I develop the theory of proliferation choice outlined above
in further depth, and examine the empirical evidence for potential chemical and biological
weapons proliferation motivations through a large-n panel data analysis, and qualitative case
studies of two countries’ CBW proliferation choices.
Chapter 2 reviews the existing literature on drivers of chemical and biological weapons
pursuit, before laying out a theory of proliferation choice that links the unique character-
istics of chemical and biological weapons to decision-makers’ perceptions of these weapons
feasibility and utility for countering particular types of threats and proposing a set of hy-
potheses concerning the internal and external security threats that might motivate chemical
and biological weapons proliferation.
InChapter3andChapter4, Iconductaquantitativetestofthetheoreticalpredictions
outlined in Chapter 2, using event history analysis and panel data, including newly collected
dataonCBWprogramsandcapabilities. Chapter3discussestheconceptualissuesthatarise
when measuring weapons proliferation, and my approach to data collection, and introduces
newlycollecteddataonCBWpursuit,possession,anduse. Chapter4discussesthemodeling
approach and presents the results of the quantitative analysis.
Chapter 5 and Chapter 6 employ in-depth case studies to illustrate how two kinds of
regime security threats — coup risk and violent unrest — can shape chemical and biological
weapons pursuit choices. Chapter 5 examines the trajectory of Libya’s chemical weapons
program, illustrating how, in a state experiencing both significant regime security threats
and significant national security threats, regime security factors can contribute to substi-
tution dynamics. A “coup-proofed” state lacked the institutional capacity to e↵ ectively
monitor and manage a large, complex nuclear weapons program, making a smaller-scale
chemical weapons program a desirable option to counter a range of security threats. Chap-
ter 6 examines Rhodesia’s chemical and biological weapons programs, illustrating how, in a
state experiencing violent unrest and civil war, chemical and biological weapons’ perceived
10
counterinsurgency utility can motivate pursuit. The Rhodesian case also shows how biolog-
ical weapons’ greater perceived destructiveness, relative to chemical weapons, complicates
bioweapons decision-making in internal security contexts.
Chapter 7 concludes with a discussion of the implications of my findings for theory
and policy, as well as potential avenues for future research.
11
Chapter 2
National Security, Regime Security,
and Proliferation Choices
Thischapterprovidesanoverviewofthecurrentstateoftheliteratureondriversofchemical
andbiologicalweapons(CBW)pursuit,andlaysoutatheoryofproliferationchoicethatlinks
theuniquecharacteristicsofchemicalandbiologicalweaponstodecision-makers’perceptions
of these weapons feasibility and utility for countering particular types of threats.
In contrast to the literature on nuclear proliferation, which initially focused heavily
on reasons why states might want the bomb, and only later turned substantial attention to
factors that might explain the relative rarity of nuclear proliferation, the literature on chem-
ical and biological weapons has thus far been heavily focused on factors that might limit
their spread, such as arms control regimes, stigmatization, and the di culties of using such
weapons for traditional deterrence. The puzzle, then, is why, despite these factors, some
states nevertheless attempt to acquire chemical and biological weapons. Taking the security
environmentasastartingpointforunderstandingchemicalandbiologicalweaponsprolifera-
tiondecisionmaking,thischaptertheorizesthemechanismsthatlinkchemicalandbiological
weapons proliferation decisions to two broad categories of security threats: threats to na-
tional security from external actors, and domestic threats to the security of the governing
12
regime.
RegimesecuritythreatscanmotivateCBWpursuitbecausetheseweaponsrelativelack
of destructiveness—the very feature that makes them poor strategic weapons—make them
an attractive option for responding to domestic threats. Furthermore, internal security con-
cerns can make managing an e↵ ective nuclear weapons program more di cult — chemical
and biological weapons o↵ er options for weapons that lend themselves to highly centralized
control, while still being (relatively) easy to acquire. However, external national security
threats may still play a role in proliferation decision-making, through two pathways: First,
states may seek to deter CBW-armed adversaries by matching those capabilities. Threats of
in-kind retaliation to potential CBW attacks are likely more credible than other retaliatory
threats. Second, given the prevalence of supply-side constraints on nuclear weapons pursuits
and uncertainty surrounding future levels of such constraints, some states may pursue chem-
ical and/or biological weapons as a risk reduction strategy—a means of hedging against the
potential failure of a more ambitious nuclear program.
Theremainderofthechapterproceedsasfollows: First, Ireviewtheexistingliterature
on CBW pursuit. Next, I lay out the theoretical framework. I begin by discussing the char-
acteristics that distinguish chemical and biological weapons from other proliferation options,
such as nuclear weapons. I then identify pathways through which these features might make
CBW a preferable option to counter regime security threats, and pathways through which
these features might make CBW proliferation an appealing response to national security
threats.
Disentangling Motivations for WMD Pursuit
Existing political science literature on the pursuit of chemical and biological weapons is
relatively sparse, particularly in comparison to the literature on nuclear weapons. Some
document the tendency among policymakers and analysts to treat chemical, biological, and
13
nuclear weapons as a single unified category, and argue that such grouping can produce
misleading inferences, and therefore ine↵ ective policy choices (Bentley, 2013; Hymans, 2004;
Narang,2015).
1
Atthesametime,suchdiscourseshavebeenappropriatedbysomedevelop-
ing countries to justify their continued possession of chemical weapons (Price, 1995). Others
have focused on the history of arms control with respect to chemical weapons, identifying
early implicit and explicit arms control agreements (Legro, 1995), or examining the emer-
gence of norms against the use and possession of chemical weapons (and linkage to norms
governing nuclear use) (Price, 1995, 1997). Additionally, there is a sizable policy literature,
primarily focused on the importance of chemical and biological weapons in the post-Cold
War security environment (Carus, 1991; Einhorn and Flournoy, 2003; Smith, 2000).
Primarily driven by historians and natural scientists, this literature tends to be more
oriented toward description than theory development and testing. However, much of this
literature contains implicit or explicit theories of CBW proliferation. These explanations
generally correspond to the main theoretical ‘models’ of nuclear proliferation (Sagan, 1997).
National security models of proliferation posit that proliferation is a tool to ensure a state’s
survival. States proliferate in response to external security threats posed by rivals that
possess the target weapons capabilities, or are otherwise militarily superior (Spiers, 1989;
Geissler and van Courtland Moon, 1999; Wheelis and R´ ozsa, 2009; Chevrier, 1993; Burck
and Flowerree, 1991). Domestic politics models attribute proliferation to a narrower set
of interests: the bureaucratic or parochial interests of a relatively small set of domestic
actors, such as scientists, military leaders, and the ruling coalition (Guillemin, 2005; Brown,
2017; Robinson, 1989). Norms models attribute proliferation to the symbolic value of a
particular type of weapons (and conversely, attribute restraint to stigmatization, or loss in
the symbolic value of a weapon). While nuclear weapons have often been seen as sources
of international prestige, and symbols of modernity and military power, the literature tends
1
More generally, this is an example of “conceptual stretching” (Sartori, 1970).
14
to treat normative issues as primarily pushing states away from CBW proliferation (Tucker,
2000; Harris, 1990).
One implicit theory that has been pervasive in (Western) policy and public discourse
surrounding chemical and biological weapons proliferation is what is often called the “poor
man’s atomic bomb” thesis: the idea that countries that desire nuclear weapons, but lack
the resources to realize their nuclear ambitions, turn to chemical or biological weapons as
cheaper, easier to acquire substitutes. This framing of chemical and biological weapons
became dominant starting in the early 1980s, with a shift away from focusing on the East-
West balance of CBW capabilities, as many Western powers wound down their chemical and
biological weapons programs, and chemical weapons uses by Egypt and Iraq drew attention
to the spread of those weapons in the “Third World” (Toulabi, 2021). For instance, Deputy
AssistantSecretaryofDefenseThomasWelch’s1985testimonybeforetheSenateCommittee
on Armed Services encapsulates this view quite well:
It is also interesting to note when we stopped in the late sixties or early seven-
ties, stopped producing chemicals, stopped testing and training, a lot of nations,
mostly Third World, picked up this weapons system. It has been called the
Poor Man’s Atomic Bomb and indeed it does fit into this idea of a cheap mass
destruction weapon. (United States Congress, 1985)
This understanding was the CBW analogue to Kennedy’s infamous 1960 prediction ‘that 10,
15, or 20 nations will have a nuclear [weapons] capacity[...] by the end of the Presidential
o ce in 1964,’ (Kennedy, October 13, 1960) — immense concern about rapid spread of
weapons to new actors (and to new types of actors) that did not, in fact, materialize.
More recently, there has been an increase in scholarly attention paid to the idea of
interrelation between weapons types, as empirical studies of the relationship between nu-
clear weapons and the pursuit of other types of strategic weapons have begun to emerge.
For instance, Early and Way (2012) investigate a pair of technologies that clearly function
as complements, finding that pursuit of missile technology makes later pursuit of nuclear
weapons more likely. Frankenstein et al. (2015) look at states’ nuclear, cyber, and biological
15
(but not chemical) weapons motivations and latent capabilities, and find a mix of country
profiles—some pursue biological weapons separately form nuclear and cyber, while others
pursue them together. Horowitz and Narang (2014) provide the only direct test of the rela-
tionshipbetweennuclear, chemical, andbiologicalweaponsproliferation, findingthatsimilar
factors drive nuclear, biological, and chemical proliferation, that countries pursuing one type
of weapons are more likely to pursue the others, and that states are more willing to give up
or stop pursuit of chemical and biological weapons once they acquire nuclear weapons.
Additionally, Horowitz and Narang (2014) outline some theoretical reasons that nu-
clear and CBW might be treated as either complements or substitutes. Proliferators would
treat nuclear and other WMD as complements if they view chemical and biological weapons
as serving di↵ erent roles that both contribute to overall security. On the other hand, pro-
liferators would treat nuclear and other WMD as substitutes if they were viewed as serving
the same purposes and could therefore be used as replacements for each other. (Toulabi,
2021) also challenges the “poor man’s atomic bomb” argument, but focuses more on the
types of states—in terms of level of development—that are likely to pursue chemical and
biological weapons, and the types of programs these states are likely to pursue—in terms of
scale—rather than engaging directly with the question of substitution dynamics.
In looking for proliferation motivations that separate chemical and biological weapons
from nuclear weapons, Koblentz (2013) argues thats domestic threats to regime survival
(rather than just state survival) can serve as a powerful motivator for CBW acquisition and
use. States that experience high levels of regime insecurity — the threat of losing power
through a coup or domestic uprising — will tend to perceive these threats as equally, if
not more important, than external security threats. Chemical and biological weapons are
particularly useful for the suppression of domestic threats, both as counterinsurgency tools,
and as tools of assassination. This work pushes the literature on chemical and biological
weapons forward in several important ways: first, it emphasizes the development and testing
ofanexplicittheoreticalframework,incontrasttothefocusondescriptionwithonlyimplicit
16
theorizing prevalent in the preceding CBW proliferation literature; second it points out
that, while theorists of proliferation devote much attention to the international security
and international political benefits provided by particular weapons, as well as the domestic
political benefits, little attention has been paid to these weapons potential domestic security
benefits, and begins to fill this gap. Koblentz’s work serves as a jumping-o↵ point for the
theoreticalframework,whichattemptstointegrateinternalandexternalsecuritymotivations
for chemical and biological weapons pursuit.
A Theory of Proliferation Choice
The theory presented here takes security concerns as the primary motivation for leaders’
choice to purse particular weapons options. While there are a variety of non-security reasons
that might motivate decision makers to pursue or eschew particular weapons or military
capabilities,
2
security is typically the stated motivation for weapons acquisition, and thus a
natural starting point for analysis. Specifically, states’ choices to pursue particular weapons
are likely to be tied to decision-makers’ beliefs about the tactical and/or strategic utility of
theseweaponsforcounteringanticipatedsecuritythreats(Chapman, ElbahtimyandMartin,
2018). For weapons such as nuclear, chemical, and biological weapons, which are both high
salienceandrestrictedunderinternationallaw,itisparticularlylikelythatsuchdecisionswill
involve the perceptions of political leaders themselves, rather than simply decision-making
within military organizations.
Arming is typically understood as a response to external security threats (Fearon,
2018; Waltz, 1979; Nordhaus, Oneal and Russett, 2012; Jervis, 1978). However, I argue
that leaders choosing to pursue chemical and biological weapons pursuit are likely motivated
by internal security threats as well. I focus on two broad categories of security threats:
2
See, e.g., Eyre, Suchman et al. (1996); Hymans (2006)
17
national security threats, the threats to state survival with which realists have traditionally
been concerned—and which have been the primary focus of the proliferation literature—and
regime security threats, internal threats to the continued survival of a particular governing
regime within the state—which have been less emphasized in the proliferation literature to
date. States that experience high levels of regime insecurity — the threat of losing power
through a coup or domestic uprising — will tend to perceive these threats as equally, if not
more important than external security threats.
Thistheorytakesasitsstartingpointtwokeyaspectsinwhichchemicalandbiological
weapons di↵ er from nuclear weapons: CBW are both less destructive and easier to acquire
than nuclear weapons. Each of these features makes chemical and biological weapons a
desirable option to counter di↵ erent types of security threats — both internal and external.
The subsequent sections in this chapter discuss these two distinguishing characteristics in
more depth, and then theorize how these characteristics contribute to perceptions of CBWs
utility for countering di↵ erent kinds of security threats.
Distinguishing features of Chemical and Biological Weapons
First, CBW are generally perceived as less destructive than nuclear weapons, and therefore
more credibly “useable” in lower stakes situations. This makes them attractive tools for
responding to domestic threats and for use on the battlefield against adversaries also armed
with CBW, while nuclear weapons are more useful for large-scale deterrence of existential
threats posed by other states. In particular, chemical weapons programs may include a
variety of incapacitating but non-lethal agents, and even lethal agents must be delivered
and concentrated in large quantities to generate high levels of casualties (Cordesman and
Fillinger, 2001; Zelico↵ ,2001). Evenbiologicalweapons,whicharepotentiallysignificantly
more destructive, are poorly suited for traditional deterrence purposes, due to uncertainty
about the e↵ ects of a planned bioweapons attack, the relatively long timeframe over which
those e↵ ects would take place, and need to keep weapons capabilities secret so that adver-
18
Figure 2.1: Weapons characteristics and theorized proliferation drivers
saries are not able to develop the necessary defensive measures (Koblentz, 2010, 2003). In
general, the ability of chemical and biological weapons to produce high casualty and fatality
rates is very sensitive to environmental and contextual factors such as weather conditions,
concentration of the target population, and employment of passive defensive measures by
the adversary. Furthermore, they have limited strategic counterforce utility, as they produce
no kinetic destruction of physical infrastructure.
Second, CBW are generally easier to acquire, requiring less scientific/technical knowl-
edge than nuclear weapons, and relying more on readily available dual-use components.
Furthermore, it is possible to successfully acquire chemical (and to a lesser extent, biological
weapons) even with a relatively small program, while successful nuclear weapons programs
are necessarily much larger scale. Program timelines to successful acquisition are generally
shorter for chemicaland biologicalweaponsthannuclear weapons, anddetermined prolifera-
tors appear to be able to acquire chemical weapons capabilities given su cient e↵ ort, which
is certainly not always the case for nuclear weapons Horowitz and Narang (2014).
19
These two features — which, as discussed in the following section, tend to make
weapons desirable for addressing internal security threats — are most pronounced for chem-
ical weapons, with biological weapons being somewhere in between chemical weapons and
nuclear weapons in terms of both destructiveness and the technical di culty of weaponiza-
tion and necessary program scale. Thus, while we would expect internal security threats to
play more of a role in motivating both chemical and biological weapons pursuit than they
do in motivating nuclear weapons pursuit, we should expect biological weapons pursuit to
be relatively less motivated by internal security threats (and more motivated by external
threats) when compared to chemical weapons. Figure 2.1 illustrates where di↵ erent weapons
typesfallonthetwodimensions—destructivenessandeaseofacquisition—andtheimplied
expectations for which types of threat motivate pursuit.
Internal Threats and Regime Security
One area in which CBW might be complements for nuclear weapons — that is, where there
might be motivations for CBW pursuit that go beyond those for nuclear pursuit — is in
their application to internal security. Some scholars have noted tendencies for CBW to be
used as tools of counterinsurgency or assassination (Tucker, 2000; Spiers, 1989), and to be
pursued by regimes with revolutionary histories (Zanders, 1995). While existing theories of
WMD proliferation deal with the international security benefits, as well as the international
and domestic political benefits that possession of such weapons can confer, their potential
domestic security benefits are largely ignored.
The one exception is Koblentz (2013), who highlights the relative ease of using CBW
covertly and the di culty of attribution as attractive features for domestic use, and provides
compellingevidencethatdomesticthreatstoregime survival(ratherthanjuststatesurvival)
served as a motivator for CBW acquisition in South Africa and Iraq. However, there has
been no more systematic investigation of the theory across a larger sample of countries, or
attempt to asses the significance of the role played by regime security concerns relative to
20
other factors. Recently, some large-n work has considered the e↵ ects of regime insecurity on
nuclear weapons proliferation, but these studies tend to approach regime insecurity through
the lens of external threats. Way and Weeks (2014) find that personalist dictatorships
are more likely to pursue nuclear weapons, as they are more likely to fear regime change
(or interference in internal a↵ airs) from outside actors, and face fewer domestic political
constraintstonuclearpursuit. FuhrmannandHorowitz(2014)findthatleaderswithprevious
experience in rebel movements, are likely to have a heightened awareness of the risk of losing
power (again, driven by outside actors), and thus are more likely to pursue nuclear weapons
as the ultimate guarantor of regime survival. Brown, Fariss and McMahon (2016) find
that states pursue nuclear (and to a lesser extent, chemical and biological) weapons as an
alternative means to protect against external security threats, following losses in military
e↵ ectiveness due to “coup-proofing”. Where these studies do note internal regime security
concerns as drivers of nuclear pursuit, the primary role of these concerns is to make nuclear
weapons a more desirable tool for the provision of external security (relative to conventional
military power). Chemical and biological weapons, on the other hand, o↵ er direct internal
security benefits because they can be used as tools to suppress internal security threats.
In this project, I propose a broader test of regime security theory, using cross-national
data. Additionally, I theorize specific pathways linking di↵ erent types of regime security
threats to chemical and biological weapons proliferation motivations. Domestic threats to a
state’sleaderorgoverningregimecancomefromeitherfromthegeneralpopulation(through
rebellionandpotentiallycivilwar),orfrompoliticalormilitaryelites(throughacoupd’´ etat).
CBW can contribute to the suppression of both types of threats, through distinct pathways.
Threats from elites
For leaders at risk of overthrow via coup d’´ etat, chemical and biological weapons o↵ er a mil-
itary option that can be acquired and used under relatively centralized control. In order to
mitigate the risk of a coup, leaders may engage in ”coup-proofing” strategies, such as coun-
21
terbalancing (creating redundant parallel military institutions, and limiting communications
between these branches), and appointment or promotion of o cers based on loyalty and
ties to the regime, rather than merit (McMahon and Slantchev, 2015; Powell, 2012), with
detrimentalconsequencesformilitarye↵ ectiveness(Brooks,2013;Talmadge,2015). Because
they lend themselves to centralized control, chemical, biological, and nuclear weapons can
provide a means of recouping military e↵ ectiveness (Brown, Fariss and McMahon, 2016).
However, large, complex scientific and technical endeavors, such as a nuclear weapons
program, can be di cult to manage and organize in insecure regimes. Other state orga-
nizations often exhibit pathologies similar to those in ”coup-proofed” militaries, as leaders
either fragment or dismantle state institutions in order to secure their hold on power (Braut-
Hegghammer, 2016). These pathologies can limit both the capacity to motivate and manage
scientists (Hymans, 2012), and the capacity to e↵ ectively utilize technical assistance from
outside sources (Montgomery, 2013; Kemp, 2014). Chemical weapons, in particular, o↵ er
ameansofregainingsomecentrallycontrolledmilitarycapabilitythroughsmaller,more
manageable programs with relatively lower technical hurdles.
These weapons then o↵ er a tool that leaders can use against opposition. Given the
ease of covert use and di culty of tracing, CBW are popular options for the assassination
of regime opponents (Koblentz, 2013).
3
Relative to conventional means, CBW can produce
casualties with relatively little manpower (Spiers, 1994), meaning that a leader who only
retains control over a small portion of the military can still leverage them against opponents
(including disloyal segments of the military).
This suggests the following hypothesis:
Hypothesis 1. Regimes facing higher levels of coup risk will be more likely to pursue chem-
ical weapons and/or biological weapons.
3
Recent high-profile examples include the assassination Kim Jong-nam with VX nerve agent
(North Korea), and the attempted poisoning of Alexi Navalny with a Novichok agent (Russia).
22
Threats from the general population
For leaders fearing (or experiencing) domestic uprising or civil war, chemical and biologi-
cal weapons can be useful tools for repression, with applications ranging from riot control
to counterinsurgency warfare. When used as counterinsurgency tools, the risk of in-kind
retaliation is low, covert use is easy (reducing the risk of an international response), and
the weapons are cost-e↵ ective, generating high casualties and spreading terror, while requir-
ing relatively little expenditure of manpower (particularly in remote or inaccessible areas)
(Koblentz, 2013). Additionally, threats of CBW use may be more credible than other, more
destructive weapons, particularly when bargaining over lower-stakes issues, or if carrying
out the threat would require deploying the weapon’s on a state’s own territory. Given the
historical record, chemical weapons, which have been employed in multiple conflicts across
regions and decades — and including civil conflicts — may be seen as more “useable” than,
e.g., nuclear weapons.
4
Chemical and biological weapons relative lack of counterforce utility
actually makes them particularly attractive options for use against domestic populations.
CBW are primarily useful for casualty generation and are not well suited to the destruction
of infrastructure or hardened military targets (Horowitz and Narang, 2014). Thus, they
may be more likely to play roles in civil conflict or internal repression, where the goal is to
eliminate adversaries without destroying one’s own territory and infrastructure.
Whilechemicalandbiologicalweaponsvaryintheirphysicaldestructivenessandability
togeneratecasualties,theycanhavesignificantpsychologicale↵ ects(JamesandOroszi,2015;
Ritchie, Watson and Friedman, 2015; Romano Jr and King, 2001; Stokes and Banderet,
1997). Even the threat of chemical or biological weapons use can incite extreme fear and
have significant e↵ ects on insurgents’ (or even professional soldiers’) morale and willingness
to fight, and on citizens willingness to protest. These weapons may be particularly likely
4
For discussion of nuclear weapons’ lack of “useability,” see Tannenwald (1999), Paul (2009),
and Press, Sagan and Valentino (2013).
23
to induce terror and dampen morale because of their unfamiliarity, lack of initial kinetic
e↵ ect, their physiological e↵ ects on the human body, and their ability to render traditional
strategies of seeking safety counterproductive
5
(Chapman, Elbahtimy and Martin, 2018).
Thus, CBW can be particularly e↵ ective as a deterrent in against domestic threats.
This suggests the following hypothesis:
Hypothesis 2. Regimes facing higher levels of domestic unrest, particularly violent unrest,
will be more likely to pursue chemical weapons and/or biological weapons.
External Threats and National Security
There are several ways in which external threats could motivate chemical and/or biological
weapons pursuit. First, leaders may believe that CBW can fill a specific need for balancing
againstadversaries(distinctfromneedsfilledbypotentialnuclearcapabilities). Ontheother
hand, potential proliferators may view chemical and biological weapons as fulfilling (albeit
less e↵ ectively) similar external security needs as nuclear weapons.
Reactive Proliferation: Balancing Adversaries
If chemical and biological weapons are regarded as strategic weapons similar to nuclear
weapons, we would expect CBW pursuit and acquisition to be motivated by the strategic
logicofdeterrence. WhenfacingadversariesthatpossessCBWcapabilities, statesmayhope
to deter the use of these weapons by possessing the capability to retaliate in kind. Thus
states that perceive a risk of facing a chemical or biological attack — states with rivals that
possess CBW — are likely motivated to pursue CBW capabilities of their own. Incentives
to acquire CBW are heightened when a potential adversary not only possesses CBW, but
has demonstrated willingness to use them, by using them in previous conflicts.
5
For instance, heavier-than-air agents such as sarin sink into underground shelters that would
o↵ er protection from conventional bombing. Likewise, biological agents that can spread through
interpersonal contact may impede cooperation and produce mistrust.
24
This type of “reactive proliferation” dynamic has played an important role in driving
patterns of nuclear proliferation, and an understanding of these dynamics — and attendant
anticipation of potential proliferation — has played an important role in shaping successful
proactive US nuclear nonproliferation policy (Miller, 2014).
AcquisitionofindigenousCBWcapabilitiesisnottheonlyavailableoptionforinternal
or external balancing against a CBW armed adversary. However, it may serve as an impor-
tant component of a balancing strategy, as chemical or biological retaliation is perceived
as more severe than limited conventional retaliation, while remaining more credible than
nuclear retaliation. Because of CBW’s status as WMD, a conventional response, whether
by the target state or an ally, may be seen as insu cient. On the other hand, given CBW’s
relative lack of destructiveness, the threat of nuclear retaliation lacks credibility, particularly
if the threat comes from an ally, rather than the target itself.
This suggests the following hypotheses:
Hypothesis3a. States with rivals that possess chemical and/or biological weapons are more
likely to pursue their own chemical and/or biological weapons capabilities.
Hypothesis 3b. States with rivals that have previously used chemical and/or biological
weapons are more likely to pursue their own chemical and/or biological weapons capabilities.
While the lesser destructiveness of chemical and biological weapons likely means they
will be perceived as insu cient for the deterrence of a nuclear-armed adversary, chemical
or biological weapons proliferation may still be motivated by a rival’s nuclear acquisition
in a “reactive proliferation” pattern for several reasons. First, proliferators may anticipate
that CBW will be of use in bargaining, particularly over lower-stakes issues, with a nuclear
armed rival, as the threat of their use is seen as more credible, and an adversary may not be
able to credibly threaten nuclear use in return. Additionally, as discussed in the following
section, CBW may be viewed as imperfect substitutes for nuclear weapons. Thus, I propose
the following hypothesis:
25
Hypothesis 4. States with rivals that possess nuclear weapons are more likely to pursue
chemical and/or biological weapons capabilities.
Hedging: Substitution as Risk Reduction
Substitution dynamics may be far more prevalent than has been suggested by existing work.
There are several reasons that this might be the case despite the important di↵ erences
between chemical, biological, and nuclear weapons. The first is that, for deterrence, it is
the adversary’s perception or expectation of a weapons destructiveness that matters, rather
than its actual e↵ ects—and CBW are generally perceived as highly destructive, a perception
which is evident in the conventional wisdom that CBW are a “poor man’s atomic bomb.” It
is also true that chemical and biological weapons can be highly e↵ ective, particularly in the
context of asymmetric capabilities, where the adversary lacks an o↵ ensive or defensive CBW
capability of its own. Additionally, chemical and biological weapon are typically viewed as
easier to acquire than nuclear weapons, in terms of both material resources and technical
knowledge, so they may be viewed as a cheaper way to acquire a comparable, though less
e↵ ective deterrent than nuclear weapons.
More specifically, I argue that potential proliferators tend to view chemical and bio-
logical as imperfect substitutes for nuclear weapons. I believe this is the case for several
reasons. First, the existing empirical finding that acquisition of nuclear weapons appears
to satisfy the demand for chemical and biological weapons (in that countries that acquire
nuclear weapons are willing to give up chemical and biological weapons pursuit) (Horowitz
and Narang, 2014). Second, characterizations of the di↵ erent types of WMDs as comple-
ments simply because countries pursuing one are more likely to pursue the others typically
treats pursuit as a binary choice, ignoring the intensity of pursuit and the resources allo-
cated to each type of weapons program relative to others. Given the uncertainty inherent to
the pursuit of WMD, states could begin one type of weapons program, but then reallocate
resources to another type of weapon (where the chances of success may be higher), without
26
fully abandoning the first program.
Whilechemicalandbiologicalweaponsmaybepoorerprovidersofstrategicdeterrence
(in the sense of mutually assured destruction) than nuclear weapons, they can likely provide
the sort of “existential” or “last resort” that many nuclear aspirants are seeking—the ability
to threaten devastating retaliation in response to serious threats to the state’s survival.
The majority of nuclear proliferators today are not seeking (nor can they reasonably hope
to achieve) a strategic arsenal, but rather more limited capabilities that o↵ er deterrence
against proximate threats. Patterns of use of chemical weapons (the most used of the three
in practice) suggest that chemical weapons can be e↵ ective responding to such threats,
particularly in the context of civil conflict.
Formally,X andY are substitutes if an increase in the price ofX leads to an increase
in the demand for Y. Specifically, I argue that as nuclear acquisition becomes more costly,
relative to a state’s available resources, demand for chemical and biological weapons will
increase:
Hypothesis 5. States with fewer resources and less technical infrastructure relevant to the
production of nuclear weapons, ceteris paribus, are more likely to pursue chemical and/or
biological weapons.
Even if a state would ideally prefer to acquire nuclear weapons, it may face both
internal (resource) and external (supply side controls, opposition from allies or adversaries)
constraints on its ability to actually acquire those weapons. Decision makers in such states
may seek to “diversify”—pursuing multiple weapons options, in the hope that at least one
program will be successful. Given the uncertainty inherent to the pursuit of WMD, states
could begin one type of weapons program, but then reallocate resources to another type of
weapon (where the chances of success may be higher), without fully abandoning the first
program. Thus, if nuclear acquisition becomes more costly, relative to a state’s available
27
resources, demand for chemical and biological weapons may increase.
6
While the focus of this project is on security threats, both the “hedging” and “reactive
proliferation” pathways need not be driven purely by security threats. Indeed, they can be
consistent with other externally-focused “models” of proliferation motivation evident in the
existing literature on nuclear proliferation; all that the substitution argument requires is
that chemical and biological weapons are able to (or even just perceived to) fulfill some of
the needs that drive proliferants’ demand for nuclear weapons. Taking some of the common
motivations suggested by the literature on nuclear pursuit, it appears that there are some
grounds to view this assumption as reasonable. As discussed above, CBW may be perceived
as su ciently destructive to provide a minimal or “last-resort” deterrent.
Much like nuclear pursuit, CBW pursuit can be used to garner international attention,
as the great powers have been invested in preventing the spread and use of CBW. Addition-
ally, while CBW possession may carry less “prestige” than nuclear weapons, an extensive
CBW program is still a notable technological and organizational achievement, and CBW
may o↵ er some of the “normative” benefits provided by nuclear weapons—specifically, a
CBW program may signal, either to nationalist publics or various groups of states (such as
the Nonaligned Movement), a willingness to resist the dominant global order.
Conclusion
In this chapter, I lay out a theoretical framework and propose a set of hypotheses concern-
ing the internal and external security threats that might motivate chemical and biological
weapons proliferation. The remainder of this manuscript is devoted to examining the em-
pirical evidence for these propositions. In the next chapter, I use time-series cross sectional
data — including newly collected data on CBW pursuit, acquisition, and use — to reveal
6
States may also pursue other options, such as seeking nuclear-armed allies, but those choices
are outside of the scope of this project.
28
systematic patterns in states’ choices to pursue CBW. In the following chapters, I use case
studies of several CBW programs to shed light on the mechanisms driving these patterns.
29
Chapter 3
Measuring CBW Proliferation and
Use
This chapter introduces the cross-country data on chemical and biological weapons prolif-
eration and use since 1945 that will be used in the following following chapter to test the
theoretical predictions outlined in Chapter 2. I begin by discussing the conceptual issues
that arise when measuring weapons proliferation, and my two-pronged approach to measur-
ing the dependent variable. Next, I provide an overview of my data and a comparison to
existing datasets. Finally, I discuss my collection of data on chemical weapons use, provide
an overview of the data, and note its implications for the theory.
Conceptualizing and measuring weapons pursuit: two
approaches
The literature on nuclear proliferation has long grappled with the question of what ‘pro-
liferation’ means. Proliferation can refer to a broad spectrum of behavior related to the
exploration, pursuit, and acquisition of weapons. Approaches to coding nuclear prolifera-
tion have included coding “decisions to pursue nuclear weapons acquisition,” (Meyer, 1986),
30
broadly coding the presence of “nuclear weapon activities” (Mueller and Schmidt, N.d.),
and breaking proliferation activities down into a variety of di↵ erent stages (Bleek, 2017; Jo
and Gartzke, 2007; Kegley, 1980; Singh and Way, 2004). While distinctions between pursuit
and acquisition are common, several studies, including the most up-to-date attempt to com-
prehensively code nuclear proliferation (Bleek, 2017), also distinguish between ‘pursuit’ and
mere ‘exploration’. According to Bleek, exploration “seeks to capture whether leaders are
openingthedoortopotentialnuclearweaponsprograms,”(2017,p.2), andcanincludeactiv-
ities such as studies to asses the feasibility or desirability of nuclear weapons acquisition, or
various low level activities that might increase a state’s capacity to produce nuclear weapons
but fall short of launching an actual weapons program. Pursuit, on the other hand, involves
an active decision to launch a program with the goal of acquiring weapons or a breakout
capability. Acquisition refers to the possession of a (potentially rudimentary) deliverable
weapon.
When it comes to chemical and biological weapons, determining what constitutes
weapons pursuit poses similar challenges. These issues are heightened for several reasons.
First, the generally smaller scale of chemical and biological weapons programs means that
it is easier for them to remain secret for longer — leading to more limited publicly available
information on weaponization activities and leaders’ decision-making surrounding weapons
programs. Second, there is more overlap between o↵ ensive and defensive CBW research,
1
sometimes making it di cult to conclusively determine whether research into weaponiz-
able chemical and biological agents actually indicates the presence of an o↵ ensive weapons
program. This study makes use of two di↵ erent approaches to measuring these outcomes,
detailed below. The first follows existing quantitative work in taking an expansive view of
what constitutes a chemical or biological weapons program, while the other employs original
data based on a more restrictive understanding of what constitutes a weapons program.
1
For instance, states may seek to manufacture small quantities of various agents in order to test
their own defensive measures.
31
Existing data: an expansive approach
The most comprehensive existing dataset covering chemical and biological weapons was
compiled by Horowitz and Narang (2014). I take Horowitz and Narang’s (2014) pursuit and
acquisition dates as a starting point. The pursuit dates provided by this dataset rely on
afairlyexpansiveconceptualizationofchemicalandbiologicalweaponspursuit,including
all states where intelligence estimates have indicated suspicion that a chemical and/or bi-
ological weapons program existed. In practice, some states coded as pursuing chemical or
biological weapons in this dataset are engaged in activities that are more consistent with the
concept of weapons “exploration”’, discussed above — research and development or dual-use
chemical/biotechnologyindustrialcapacitybuildingactivitiesthatdonotdefinitivelyinclude
weaponization activities.
This dataset identifies 49 states that either pursued or possessed chemical weapons
between 1945 and 2000. A total of of 33 states pursued chemical weapons during these
years, with 15 of those successfully acquiring weapons. 16 states acquired chemical weapons
prior to 1945, for a total of 31 states possessing chemical weapons at some point during the
study timeframe. Seven of these countries are coded as giving up World War II era chemical
weapons at the war’s conclusion in 1945, thus the only year that they appear in the dataset
is 1945.
Between 1945 and2000, the dataset identifies 21 statesthat either pursuedor acquired
biological weapons. A total of of 16 states pursued biological weapons during these years,
with 6 of those successfully acquiring weapons. 5 states acquired biological weapons prior
to 1945, for a total of 11 states possessing biological weapons at some point during the
study timeframe. All except two of the states that pursued biological weapons also pursued
chemical weapons.
32
Re-coding: a restrictive approach
The Horowitz and Narang (2014) dataset casts a wide net, and employs a relatively broad
understanding of the concepts of “chemical weapons”, “biological weapons”, and “pursuit.”
Thus, the dataset may include some observations of CBW pursuit or acquisition that are
not conceptually appropriate to test a theory of leaders’ motivations for initiating chemical
or biological weapons pursuit. I therefore re-evaluate these codings, and produce a more
limited version of the dependent variable that is more directly focused on the pursuit and
acquisition of deliverable chemical and biological weapons, suitable to use in combat, and
operationally controlled by the government of the state in question.
Thesemorerestrictivecodingsrelyonoriginaldatacollectiononthechemicalweapons
agents and delivery systems pursued (or acquired) by each potential proliferator. I restrict
the dependent variables to include only programs where the state in question was pursuing
chemical and biological agents with the clear goal of weaponization. This coding necessarily
excludes states where a program was suspected to exist (e.g., by U.S. intelligence agencies),
but no information on specific agents or systems was known. However, in cases where the
state did not get to the point of producing actual chemical or biological agents, direct ev-
idence of a specific political decision to pursue o↵ ensive chemical or biological capabilities
is considered su cient to code a state as pursuing weapons. The goal here is to distin-
guish directed, intentional weapons pursuit from broader R&D and exploratory activities,
or defensive research into chemical and/or biological agents.
The re-coded dataset identifies 23 states that either pursued or possessed chemical
weapons between 1945 and 2010. A total of of 17 states pursued chemical weapons during
these years, with 14 of those successfully acquiring weapons. 6 states acquired chemical
weapons prior to 1945, for a total of 20 states possessing chemical weapons at some point
during the study timeframe.
Between 1945 and2010, the dataset identifies 13 statesthat either pursuedor acquired
biological weapons. A total of of 10 states pursued biological weapons during these years,
33
with 3 of those successfully acquiring weapons. 3 states acquired biological weapons prior
to 1945, for a total of 6 states possessing biological weapons at some point during the study
timeframe. Using the more restrictive conception of pursuit, all of the states that pursued
biological weapons also pursued chemical weapons.
While the Horowitz and Narang (2014) dataset is the primary existing dataset used by
scholars studying CBW, (Toulabi, 2021) also attempts to produce a more restrictive dataset
of CBW pursuit and acquisition. Toulabi’s focus is on cross-checking evidence of CBW
programs or possession across multiple independent sources. His codings are fairly similar
to mine, as when there is su ciently detailed information on a state’s CBW activities to
identify specific agents, delivery systems, or weaponization activities, or to confirm an in-
tentional political decision, multiple sources are likely to have access to that information.
In particular, both my codings and Toulabi’s drop many of the same states in the Horowitz
and Narang (2014) dataset, which I judge to have only engaged in exploratory chemical
or biological weapons activities. However, the exact dates of pursuit and acquisition often
di↵ er between my codings and Toulabi’s. Additionally, in keeping with his argument against
the “poor man’s atomic bomb” concept, Toulabi may be overly cautious in identifying de-
veloping countries as pursuing CBW. As a robustness check, I replicate the analysis in the
following chapter using Toulabi’s data for the restrictive coding of the dependent variable,
and see similar results, although the e↵ ects of some of the external security variables are less
consistent.
Potential pitfalls in coding CBW programs
The construction of the more restrictive codings of chemical and biological weapons pursuit
usedforthisstudyhighlightsseveralpotentialpitfallsthatonemayencounterwhenattempt-
ing to determine the presence or absence of a chemical and biological weapons program or
weapons capability. In this section, I summarize these issues and my approach to handling
them as I created the more restrictive version of the dependent variable.
34
First,theterms“chemicalweapons”and“biologicalweapons”canbeappliedtoawide
range of phenomena. For instance, in addition to the development of agents and delivery
systems that could be used on the battlefield or for counterinsurgency operations, South
Africa’s “Project Coast” also included the development of agents (including modified street
drugs) and delivery systems better suited to crowd control, and poisons intended for use in
the assassination of regime opponents (Koblentz, 2013). This example highlights two axes
on which we see variation even within the categories of “chemical weapons” and “biological
weapons”: the lethality of the agents, and the intended method of delivery.
Chemical or biological agents that are debilitating, but not necessarily lethal may
be useful either as a means of domestic repression (crowd control or “riot control,” the
suppression of protests against the governing regime) or as support for conventional military
operations(forinstance,asaforcemagnifieror‘equalizer’whenfacinglargerorbetter-armed
forces). However, in some cases, where the only chemical agents (potentially) in a state’s
arsenal are non-lethal, it can be di cult to verify the actual possession and/or use of the
agents. For instance, in some cases, the sources Horowitz and Narang (2014) rely on to code
weapons programs identify potential programs based only on “unconfirmed reports” of the
use of non-lethal agents in domestic contexts. For instance, speculations about Argentina’s
chemical weapons stockpiles are based on unconfirmed reports of lachrymatory agents (tear
gas) and smoke shells (Burck and Flowerree, 1991).
Furthermore, it can be di cult to track down possessors of tear gas and other “riot
control agents” (RCAs) because of their di↵ erent legal status from other agents. The Chem-
ical Weapons Convention defines RCAs as chemical agents “which can produce rapidly in
humans sensory irritation or disabling physical e↵ ects which disappear within a short time,”
(Article II (7)), and prohibits using these agents “as a method of warfare,” but does not
ban possession, opening the way for stockpiling and use of these agents for domestic law
enforcement purposes (Feigenbaum, 2017).
Beyond more traditional military means of delivery, such as artillery shells, missiles,
35
or gravity bombs, some states pursue chemical and biological agents with the intent of
deliveringtheseagentsinsmaller-scale,moretargetedways. Various(usuallylethal)chemical
and biological agents have been used for covert assassination operations by intelligence and
security services in multiple states. Additionally, where some access to opposition forces or
opposition controlled territory is possible, chemical and biological agents can be delivered by
contaminating supplies, particularly food and/or water sources. If a state pursuing chemical
or biological weapons intends to deliver them solely by these smaller-scale, nontraditional
methods, it will not be possible to use evidence of the development of traditional delivery
systems as a criterion for coding the presence of a program.
Given these challenges, it is possible that my more restrictive codings of pursuit and
possession miss some state programs that are focused on acquiring non-lethal agents, or
delivering agents at very small scales. It is worth noting that exclusion of these types of
programs from analysis — if indeed there are states with such programs that are not coded
as pursuing — would likely bias e↵ ect sizes for regime security threats downward, as these
programs in particular involve agents and delivery systems best designed for response to
internal threats: riot control, counterinsurgency, and assassination.
Second, the Horowitz and Narang (2014) dataset includes some observations of states
that are not actually pursuing their own indigenous CBW capabilities, but rather have some
presence of an ally’s CBW in their territory. For instance, Australia is coded as pursuing
chemicalweaponsfrom1945-1973, apparentlybecausefollowingtheend oftheSecondWorld
War, the US and British militaries left stockpiles of chemical munitions at some bases in
Australia, and disposal of these agents was not completed until the early 1970s (Goodwin,
1998; Plunkett, 2007). It seems odd tocodea country as“pursuing” chemical weapons when
theywereactivelydestroyinganddisposingoftheagentstheydidpossess—andwhichwere
never truly under their operational control. Similarly, the initial coding of Mozambique as
pursuing chemical weapons appears to be based on unverified accusations by guerrilla forces
thattheregimeusedSovietsuppliedchemicalagentstotargetthem, andsourcesspeculating
36
on the existence of a Laotian CW program appear to be based on the use of Soviet supplied
chemical agents (and potentially fungal toxins) against insurgents and civillians in the late
1970s and early 1980s.
2
When creating the more restrictive, re-coded dataset of CBW pursuit and possession
dates, I do not count these situations as instances of weapons pursuit. Allowing an ally
to stage weapons on your territory, or to intervene in an ongoing conflict by providing
weapons are distinct phenomena from pursuing an indigenous weapons capability over which
the proliferator’s government will have full operational control. States in an “exploration”
phase, considering the viability and desirability of CBW capabilities might inquire into the
possibility of obtaining agents and/or delivery systems from allies — and these cases are
captured by Horowitz and Narang’s (2014) more expansive dataset. While the exchange of
chemical and biological weapons between allies is, in and of itself, an important area for
future study, detailed analysis of these dynamics is outside the scope of the current project.
To understand how the exclusion of these cases might a↵ ect the analysis and conclu-
sions of this study, note that they fall into two general categories: 1) the presence of allied
weapons stockpiles left over from the Second World War, which the host states typically
either disposed of or returned, sometimes over a period of years, depending on knowledge
of the remaining stockpiles and disposal capabilities, and 2) ( sometimes alleged) provision
of agents by allies as a means of intervention into ongoing civil conflicts. Cases in the first
category are likely examples of pre-nuclear era logics of CBW weapons and warfare, and are
therefore of limited relevance to this study’s goals, a point which the next section addresses
at greater length. Exclusion of cases in the second category is likely to attenuate the e↵ ects
of regime security drivers of proliferation.
2
It is worth noting that accusations by the US government that the incidents in question were
the result of Soviet supplied CBW have been the subject of much controversy, and numerous
investigations have failed to produce a conclusive determination (Tucker, 2001).
37
Scope and timeframe
One potential limitation of the Horowitz and Narang (2014) data is that it only covers the
years1945-2000. Ithereforeattempttocodethepresenceofprogramsbetween2000and2010
by applying coding criteria similar to those used in the Horowitz and Narang (2014) dataset.
Ichoosetostopcodingin2010because,giventhehighlevelofsecrecysurroundingchemical
and biological weapons programs, it can take years for information about these programs
to surface in the open literature. Thus, information on chemical and biological weapons
programs undertaken in recent years is likely to be particularly unreliable, and e↵ orts to
code proliferation in these years risks systematically under reporting newer, smaller, and/or
better concealed programs.
Additionally, for the purposes of this quantitative analysis, I choose not to extend the
timeframe earlier than 1945, for several reasons. The first is practical: data on some of
the key independent variables (particularly the internal security threat variables) and some
control variables is not readily available before 1945, so expanding the timeframe over which
data on the dependent variable is collected wouldn’t necessarily expand the timeframe over
which the quantitative analysis could be conducted.
The second has to do with the conceptual scope of the theory: because the focus is on
understanding how and why states make particular choices between di↵ erent proliferation
options,itismostapplicableintheerawhenthenuclearoptionwas(potentially)anavailable
alternative. Prior to the nuclear era, states seeking powerful “weapons of terror” to deter or
demoralize state adversaries may have attempted to acquire such a capacity by investing in
large-scale chemical or biological weapons programs. For instance, prior to 1945, a handful
or great powers (e.g., the United States, the Soviet Union, Germany, and Japan) pursued
chemical and/or biological weapons programs with the goal of developing weapons that
could be used at large scale in otherwise ‘conventional’ conflict against opposing state forces
(as happened in the First World War). After the advent of nuclear weapons, the question
becomes why states (particularly those with advanced scientific and industrial capacity),
38
would not pursue nuclear weapons instead. Put simply, it does not make sense to assess the
interaction (or lack thereof) between chemical, biological, and nuclear weapons proliferation
options, when the nuclear option is not available.
CBW exploration, pursuit, and possession data
Table 3.1 provides a comparison between the chemical weapons proliferation data used in
this study, the restrictive (Toulabi, 2021), and the more expansive Horowitz and Narang
(2014) dataset. Table 3.2 provides an analogous comparison for biological weapons prolif-
eration data. Both tables cover weapons programs from 1945-2010, although the Horowitz
and Narang (2014) dataset extends only as recently as 2000 (so observations of pursuit or
possession from that dataset with an end date of 2000 should be considered right-censored).
For readability, countries which possessed chemical weapons during World War II, and gave
them up in 1945 are omitted. These include Czechoslovakia, Germany, Greece, Hungary,
Japan, Poland, and Spain for chemical weapons, and Japan for biological weapons.
Chemical weapons use
Unlike nuclear weapons, chemical weapons, in particular, have been employed in combat
with some regularity. Thus, an investigation of patterns of states’ chemical weapons use
may provide an additional potential source of insight into states’ weapons proliferation mo-
tivations. To this end, I construct a dataset of incidents of chemical weapons use in inter-
and intra-state conflict, including details on agents, delivery methods, and targets. This
information is collected from sources similar to those used for the re-coding of weapons pur-
suit and possession dates: reports from intelligence organizations and think tanks, as well
as historical studies of particular countries’ CBW programs and use.
Table 3.3 summarizes major campaigns of chemical weapons attacks by state actors
from 1946 to the present. Attacks by the same actor against the same opponent within the
39
Table 3.1: Chemical weapons exploration, pursuit, and possession: comparison of datasets.
New data Toulabi (2021) Horowitz & Narang (2014)
country exploration pursuit posession pursuit posession pursuit possession
Afghanistan 1982-1994 1982-1994
Algeria 1999-2000 1999-2000
Angola 1984-1993 1984-1993
Argentina 1971-1993 1971-1993
Brazil 1988-1993 1988-1993
Burma 1988-2000 1988-2000
Canada 1963-1968 1941-1946 1946-1969 1941-1946 1945-1946
Chad 1988-1993 1988-1993
Chile 1988-1993 1975-1976 1974-1976 1988-1993
East Germany 1980-1989 1980-1982 1983-1989
Egypt 1945-1957
1958-1962;
1974-
1963-1974
1958-1962;
1974-
1963-1974 1945-1962 1963-2000
Ethiopia 1980-1993 1980-1993
France 1988-1993 1946-1988 1988-1993 1946-1988 1945-1993
Iran 1983-1986 1987-1991 1985-1986 1987-1991 1983-1983 1984-2000
Iraq 1971-1982 1983-1991 1971-1982 1983-1991 1971-1979 1980-2000
Israel 1952-1954 1955-1956 1956- 1955-1956 1956-2010 1952-1955 1956-2000
Kazakhstan 1991-2000
Laos 1988-1993 1988-1993
Libya 1976-1979 1980-1987 1988-2011 1984-1988 1989-2004 1976-1980 1981-2000
Mozambique 1988-1993 1988-1993
North Korea 1965-1968 1969-1987 1988-2004 1961-1988 1989-2004 1965-1987 1988-2000
Pakistan 1982- 1982-1986 1987-2000
Peru 1988-1993 1988-1993
Philippines 1988-1993 1988-1993
Rhodesia (Zimbabwe) 1975 1975-1976 1977-1979 1976-1976 1977-1979 1975 1976-1980
Somalia 1988-2000 1988-2000
South Africa 1981-1986
1945-1946;
1987-1993
1981-1986
1945-1946;
1987-1993
1945-1993
Sudan 1990-2000 1990-2000
Sweden 1945-1973 1945-1973
Syria 1971-1972 1973-1985 1986- 1979-1984 1985-2010 1971-1972 1973-2000
Thailand 1988-1993 1988-1993
Vietnam 1975-2000 1975-1989 1990-2000
Yugoslavia 1958-1976 1976-1986 1987-1991 1976-1987 1988-1991 1958-1968 1969-2000
China 1946- ? ?-1997 1945-2000
India 1947-1979 1980-1984 1985-1997 ? ?-1997 1947-2000
Myanmar 1981-1991 ? ?-1990
Russia 1916-2010 1916- 1945-2000
UK -1957 -1957 1938-1957
US -1997 -1997 1945-2000
Australia 1939-1946 1945-1973
Saudi Arabia 1986-1990 1988-1989
South Korea 1967-1974 1975-1987 1988-1997 ? ?-1997 1967-1988
Taiwan 1970-1978 1979-1982 1983-1991 1970-1982
40
Table 3.2: Biological weapons exploration, pursuit, and possession: comparison of datasets.
New data Toulabi (2021) Horowitz & Narang (2014)
country exploration pursuit possession pursuit possession pursuit possession
Algeria 1999- 1999-2000
Bulgaria 1988-1993 1988-1993
Canada 1942-1969 1942-1969
Cuba 1988-1993 1988-1993
Egypt 1945-1960 1961-2010 1958-2010 1945-1971 1972-2000
France
1948-1956;
1962-1967
1948-1978 1945-1973
Iran 1981-1981 1982-2010 1981-2000
Iraq
1974-1978;
1985-1987
1988-1991
1974-1978;
1985-1989
1990-1991
1974-1986;
1992-2000
1987-1991
Israel 1948- 1948-
Laos 1988-1993 1988-1993
Libya 1988-2000 1988-2000
North Korea 1964-1987 1988- 1964- 1965-1987 1988-2000
Rhodesia (Zimbabwe) 1973-1979 1976-1979 1975 1976-1980
South Africa 1945-1980 1981-1985 1986-1993 1981-1983 1984-1993 1945-1975 1976-1993
Syria 1990- 1990-2000
Taiwan 1975-1993 1975-1993
China 1950-1957 1958- ? ? 1950-1961 1962-2000
Russia 1935- 1935- 1945-2000
UK 1934-1956 1945-1956 1945-1956
US 1944-1972 1944-1972 1940-1972
41
Table 3.3: Chemical weapons use by state actors, 1946-present.
State actor Campaign dates Target
External target
Israel 1949 Egypt
US 1951-1952 North Korea
UK 1961-1971 Vietnam
US 1962-1971 Vietnam
Egypt 1963-1967 Yemen
US 1965-1969 Laos
Iraq 1980-1988 Iran
Libya 1987 Chad
Iran 1987-1988 Iraq
Internal target
China (Nationalist) 1945-1949; 1958 Chinese People’s Army
Greece 1949 Guerrilla forces
France 1957 Algerian insurgents
Cuba 1957 Guerrilla forces
Laos/USSR 1975-1981 H’Mong civilians
Rhodesia 1977-1979 Guerrillas, civilians
Iraq 1988 Kurdish civilians
Russia 2002 Terrorists, civilian hostages
Syria 2013-2018 Rebel forces, civilians
Assassination
Bulgaria/Soviet Union 1978 Georgi Markov
Russia 2015 Emilian Gebrev
North Korea 2017 Kim Jong-nam
Russia 2018 Sergei and Yulia Skripal
Russia 2020 Alexei Navalny
42
same ongoing conflict are grouped into campaigns (e.g., the more than 30 chemical weapons
attacks against rebel forces and civilians by the Assad regime during the Syrian civil war are
groupedintoonecampaignlastingfrom2013-2018). Campaignsareseparatedbytargettype:
external actors, such as other state’s military forces, or internal actors, such as insurgent
forces or civilians. Assassinations of specific individuals are listed separately, as they tend
to rely on di↵ erent capabilities and institutional structures than chemical weapons use in
combat (for instance, assassinations are often undertaken by the intelligence services, while
WMD use during conflict is typically planned and executed by the military).
In this project, these data are used primarily to construct a key independent variable
in the reactive proliferation pathway, the presence of a rival state with a history of chemical
weapons use. However, simply looking at broad patterns of chemical weapons use provides
some initial support for regime security motivations of chemical weapons proliferation (as
well as more traditional external security motivations). Half of all post-1945 campaigns of
chemical weapons attacks in conflict (9 campaigns out of 18 total) were undertaken against
domestic targets, posing internal security threats to the regime that employed the weapons.
A further five uses of chemical weapons for assassination targeted individuals who opposed
ororwereperceivedaspotentialsourcesofinternaloppositiontotheregimeundertakingthe
assassination. In earlier time periods, chemical weapons were often employed by European
powers to counter internal threats to European rule arising within colonies (e.g., Spain used
mustard gas in Morocco; Italy employed a variety of agents including mustard and chlorine
in Ethiopia). Causal analysis of decisions to employ chemical weapons, as well as fine-
grained analysis of patterns of chemical weapons attacks within broader campaigns is left
for future research. Yet this initial descriptive assessment of post-1945 chemical weapon use
is encouraging for regime security theory.
43
Chapter 4
Drivers of CBW Pursuit: Evidence
from New Panel Data
This chapter leverages cross-country data, including the data on chemical and biological
weapons pursuit, possession, and use discussed in the previous chapter, to conduct a quan-
titative test of the theoretical predictions outlined in Chapter 2. I begin by outlining the
event history analysis approach that I employ, and the variables included in these models.
Next, I present the results from these models and conclude by highlighting key takeaways
from the findings.
Putting the data to work: a quantitative test
Dependent Variables
The main outcomes of interest in this study are chemical weapons pursuit and biological
weapons pursuit, specifically, states’ initiation of a chemical weapons program or a biological
weapons program in a given year. I estimate two versions of all statistical models, the first
using program dates identified through the Horowitz and Narang (2014) dataset’s expansive
approach, and the second using my own more restrictive coding of program dates. Chemical
44
weapons pursuit is coded as 1 if a state had a chemical weapons program, and 0 otherwise.
All observations for a given state are dropped from the analysis of chemical weapons pursuit
risk after that state acquires chemical weapons (as at that point, the risk of pursuit drops to
zero). The same coding rules and restrictions on observations apply for biological weapons
pursuit.
Independent Variables
Regime Insecurity
The primary independent variable of interest, regime insecurity, consists of two components:
coup risk and the risk of domestic uprising.
Coup risk is measured using Sudduth (2017)’s latent coup risk measure. Sudduth
treats coup risk as a function of the military’s willingness to organize a coup (their dissatis-
faction with the leader) and their ability to do so successfully, and uses multiple indicators
of willingness and ability. I also test the robustness of my results to using Belkin and
Schofer (2003)’s additive coup risk score, which includes measures of political legitimacy,
civil-military relations, and recent coups.
It is necessary to disentangle the e↵ ects of coup risk, which provides an internal secu-
rity motivation to acquire CBW, from the e↵ ects of coup-proofing, which a response to coup
risk that can generate external security motivations to acquire CBW. Thus, in some spec-
ifications, I include a measures of military e↵ ectiveness related coup-proofing: Pilster and
B¨ ohmelt (2011)’s Counterbalancing and E↵ ective Number of Forces variables, and Wimmer,
CedermanandMin(2009)’sminorityregimevariable(seealso: Brown, FarissandMcMahon
(2016)).
Risk of domestic uprising is measured using data from the Nonviolent and Violent
Campaigns and Outcomes (NAVCO) Data Project (Chenoweth and Lewis, 2013). In the
primary specification, I include binary indicators of whether there is an ongoing violent
campaign against the government and whether there is an ongoing nonviolent campaign
45
against the government. I disaggregate campaigns by the “primary method” used (violent
or nonviolent), because regimes may be a) more likely to view violent unrest as a serious
threat,andb)morelikelytoconsiderCBWuseanappropriateresponsetoviolentcampaigns.
Ialsoproxyfortheriskofcivilconflictusinganindexofethnicfractionalization
(Drazanova, 2019). While there is debate over the utility of ethnic fractionalization for
predicting civil war, there is evidence of a curvilinear ‘inverse-U’ relationship between frac-
tionalizationandtheriskofethnicconflict(Horowitz,1895).
1
Thus,tocapturethisnonlinear
relationship, I include terms for the fractionalization index and its square.
Rival WMD possession and use
Data on rival CBW possession are based on the two sets of chemical and biological weapons
possession dates discussed in the first section of this chapter: the broader, more inclusive set
of dates from the Horowitz and Narang (2014) dataset, and the more restrictive set of dates
that I re-coded based on specific agents, weaponization strategies, and delivery systems.
Indicators of rival nuclear weapons possession are based on the nuclear weapons possession
dates presented by Bleek (2017).
For data on rival CBW use, I draw on the dataset of incidents of chemical weapons
use in inter- and intra-state conflict introduced in the previous chapter. While I also collect
data on CBW use in assassinations, I do not include these instances of use when creating the
indicator of rival chemical weapons use for analysis, as potential proliferators are likely to be
most concerned with rivals’ demonstrated capacity and willingness to use chemical weapons
in combat, which may depend on di↵ erent chemical or biological agents and/or di↵ erent
1
This ‘inverse-U’ relationship occurs because we are unlikely to see ethnic conflict in societies
where almost everyone is a member of the same ethnic group (at the low levels of ethnic frac-
tionalization), or in societies where the population is split between many small ethnic groups (the
theoretical highest level of fractionalization would occur if every individual was a member of their
own ethnic group). Rather, conflict is likely to occur when the population is ‘polarized’ between
several large groups (which happens at intermediate levels of fractionalization).
46
institutionalstructuresthanuseforassassination. Iconstructanindicatorofrivals’chemical
weapons use, but do not include an analogous indicator of rival’s biological weapons use, due
to the extreme rarity of confirmed incidents of biological weapons use in conflict.
Foreachpotentialproliferator,Iconstructthreedichotomousindicatorsofwhetherany
rivalpossesseschemical, biological, ornuclearweapons, respectively, aswellasacountofthe
total number of adversaries that possess each type of weapon. Data on rivalry is taken from
ThompsonandDreyer(2011). Ineachmodel,Iuseversionsoftherivalchemicalweaponsand
rival biological weapons indicators which are based on the same conception of pursuit as the
model’s outcome variable. (For instance, in models of focused biological weapons pursuit —
the more restrictive coding of the dependent variable — I use the more restrictive possession
dates to create the ‘rival biological weapons’ and ‘rival chemical weapons’ indicators.)
Supply-side constraints
As a measure of the “supply side” constraints faced by states considering nuclear pursuit,
I use Smith and Spaniel (2020)’s measure of “nuclear infrastructure.” This measure uses
an item-response framework to combine a variety of indicators related to a state’s nuclear
and weapons infrastructure. Additionally, to set a baseline for a state’s capacity to build
di↵ erent WMD, the log of GDP per capita is included in some models.
Controls
Iestimatetwoadditionalmodelspecificationsforeachoutcome. Thefirstaddsallofthe
controls employed by Horowitz and Narang (2014), the only extant quantitative study of
the determinants of chemical and biological weapons pursuit. This is a parsimonious model
that nevertheless controls for some of the theorized determinants of WMD pursuit. By
controlling for factors that have been shown to be correlates of CBW pursuit, we can resolve
some concerns that the observed e↵ ect of coup risk is due to omitted variable bias. Below,
Ibrieflysummarizethejustificationsforthevariablesincludedinthemodel.
47
GivencorrelationbetweenpursuitofallthreetypesofWMD,binaryindicatorsforbio-
logical weapons pursuit and possession are added to the model of chemical weapons pursuit,
and vice versa. I also include dichotomous indicators for a state’s pursuit and possession
of nuclear weapons. For the coding of these variables, I use the same pursuit and acquisi-
tion dates as Horowitz and Narang (2014), ‘consensus’ dates used by numerous scholars in
a recent JCR special issue (Gartzke and Kroenig, 2009). In order to control for a state’s
capacity to build di↵ erent WMD, and consistent with many studies of nuclear proliferation,
GDP per capita and GDP per capita squared are included. Two dichotomous variables,
CWC membership and BWC membership, code whether a state has joined the Chemical
Weapons Convention and the Biological Weapons Convention, respectively. Depending on
the outcome of interest (CW pursuit or BW pursuit), the variable measuring membership
in the relevant treaty is included. Two variables are used to control for the influence of the
security environment on WMD pursuit: a dichotomous indicator for the presence of nuclear
armed ally (Jo and Gartzke, 2007; Singh and Way, 2004) and a count of the number of
contiguous states (following Way and Weeks (2014)). Other, more traditional controls for
a state’s security environment are excluded from this more parsimonious model due to con-
cerns about post-treatment bias. For example, domestic unrest may cause a state to engage
in more interstate disputes, if leaders initiate disputes in an attempt to generate a ‘rally
around the flag’ e↵ ect. Thus, dispute involvement may serve as an intervening variable, and
not solely as a pretreatment source of omitted variable bias.
The final model adds a full “demand function” for nuclear proliferation as understood
by the existing literature. This allows us to test the robustness of the e↵ ects of regime
insecurity to the inclusion of variables that have been found to a↵ ect states’ demand for
nuclear weapons (and theorized to have similar e↵ ects on CBW proliferation).
The standard measures of a state’s security environment, omitted in the previous
model, are included here. These include the presence of an enduring rivalry (Thompson
and Dreyer, 2011), and involvement in militarized international disputes (total MIDs over
48
the last five years) (Palmer et al., 2021). Additional domestic level controls include the
level of democracy, as measured by a state’s Polity IV score, economic openness (imports
and exports as a percent of GDP) (Singh and Way, 2004), whether the state’s leader has
experience in a rebel movement (Fuhrmann and Horowitz, 2014), and whether the state is a
personalist dictatorship (Way and Weeks, 2014).
Model
Following existing work on the determinants of WMD proliferation, I use event history
analysis to explore the relationship between regime insecurity and chemical and biological
weapons pursuit. I employ a Cox proportional hazard model to estimate the risk that
a state will pursue chemical or biological weapons in a given year, given that it has not
previously pursued those weapons, conditional on the covariates discussed above. I am
interested both in the factors that are associated with di↵ erences in the likelihood of pursuit
across countries and the time-varying factors that increase or decrease given country’s risk
of initiating weapons pursuit. Therefore, I estimate two sets of models, one that assesses
within-countryvariationintheriskofCBWpursuit,byintroducingfixede↵ ectstocontrolfor
time-invariantcountrylevelattributes, andonethatalsoincludesbetweencountryvariation.
For robustness, I estimate both a multiple failure model, where ‘failure’ (pursuit of chemical
or bioweapons) in each year of pursuit is treated as an independent event, and single failure
models, where states can only ‘fail’ once, by initiating pursuit, and then exit the data.
As many of the independent variables are time-varying, the unit of analysis is the
country-year (rather than country level with single-observation survival times). The dataset
rangesfrom 1945–2010, and includes192countries, withcountriesthat never initiatepursuit
considered censored in 2010. With countries exiting the data after they acquire the weapons
in question, as discussed above, roughly 6,800 observations remain for models of chemical
weapons pursuit, and roughly 7,100 for models of biological weapons pursuit. While data
limitations for the coup-proofing variables (counterbalancing and e↵ ective number of forces)
49
significantly reduce the number of observations for models that include these variables, the
results are generally robust when using the full sample (by omitting these variables).
Results
This section presents the results from Cox proportional hazard models of chemical and
biological weapons program initiation. Full results from all models are shown in Tables 1-8
in the appendix. I begin by discussing the results of the fixed e↵ ects models, which identify
the e↵ ects of changes in the independent variables on changes in a given country’s risk
of initiating weapons pursuit, for each category of potential proliferation drivers (internal
security factors vs. external security factors, and then move on to discuss factors that are
associated with between-country di↵ erences in the risk of proliferation.
Regime Security
Accordingtohypothesis1, increasesincoupriskshouldbeassociatedwithbothanincreased
likelihoodofinitiatingchemicalweaponspursuitandwithanincreasedlikelihoodofinitiating
biological weapons pursuit. Results from the fixed-e↵ ects models are consistent with this
hypothesis: for a given country, an increase in coup risk over time increases both the risk
that the state will start a chemical weapons program and the risk that the state will start a
biological weapons program. This is true for both the more expansive conception of pursuit
and exploration, and the more restrictive coding of focused pursuit. These associations
also appears to be robust to the inclusion of control variables. The e↵ ect of coup risk is
consistently positive and statistically significant across all models, with the exception of the
‘full proliferation demand function’ model of broad biological weapons pursuit. As discussed
above, the introduction of numerous control variables in the full demand function both
limits the sample size (due to data availability) and risks inducing posttreatment bias, both
of which may attenuate the e↵ ect.
50
Figure 4.1: E↵ ects of internal security threat indicators on risk of weapons pursuit. Hazard
Ratios displayed here are from Cox proportional hazard models including both internal and
external security threat indicators. Models are stratified by country.
Figure 4.1 illustrates the substantive e↵ ects of the internal security variables on within
country shifts in the risk of initiating weapons pursuit. For a given state, a one standard
deviation increase in the risk of a coup increases the risk of initiating pursuit of specific
chemical weapons agents and delivery systems (restrictive coding of pursuit) by a factor
of 2.7, and roughly triples the risk of initiating exploratory chemical weapons activities
(expansive coding of pursuit). The e↵ ects on biological weapons pursuit are also substantial:
aonestandarddeviationincreaseintheriskofacoupincreasestheriskofinitiatingbroad
biological weapons pursuit by 67%, and increases the risk of pursuing specific weaponization
51
strategies and delivery systems by several orders of magnitude.
2
Figure 4.2: E↵ ects of ethnic fractionalization on risk of weapons pursuit. Hazard Ratios
displayedherearefromCoxproportionalhazardmodelsincludingbothinternalandexternal
security threat indicators. Models are stratified by country.
According to Hypothesis 2, increases in domestic unrest, particularly violent unrest
should be associated with an increased likelihood of initiating chemical weapons pursuit
and with an increased likelihood of initiating biological weapons pursuit. Results from the
fixed-e↵ ects models show that this association holds for chemical weapons pursuit: for a
2
Note that the significantly larger substantive e↵ ect of coup risk on focused biological weapons
pursuit is omitted from Figure 4.1, in order to preserve a scale that allows comparison of the other
coe cients.
52
given country, an increase in violent protest campaigns against the government increases the
risk that the state will start a chemical weapons program. Regardless of whether weapons
programsarebroadlyconceivedtoincludeexploratoryactivitiesormorerestrictivelylimited
to focused weapons pursuit, this relationship is consistently positive across models, although
for the more restrictive coding, e↵ ects drop below conventional thresholds for statistical
significancewhenextensivecontrolvariablesareadded. Additionally, increasesinnonviolent
protest against the government have a consistently positive, but not always statistically
significant, relationship with the risk of initiating chemical weapons pursuit (across model
specifications and codings of pursuit).
On the other hand, the relationship between domestic unrest and biological weapons
pursuit is less clear-cut. Consistent with Hypothesis 2, both increases in violent protest
campaigns and increases in nonviolent protest campaigns against the government are asso-
ciated with increases in the risk that the regime will begin pursuing or exploring biological
weapons capabilities, broadly conceived (expansive coding). These associations are consis-
tently positive across models, although statistical significance varies depending on model
specification. However, there does not appear to be a consistent relationship between in-
creases in violent unrest and increases in the risk of initiating focused pursuit. Furthermore,
increases in nonviolent unrest are consistently associated with a decreased risk of pursuing
specificweaponizationstrategiesanddeliverysystemsforbiologicalweapons(restrictivecon-
ception of weapons pursuit). This negative relationship is statistically significant across all
fixed-e↵ ects models of focused pursuit.
The substantive e↵ ects of the unrest variables can be seen in Figure 4.1. For a given
state,thepresenceofaviolentcampaignagainstthegovernmentdoublestheriskofinitiating
focused chemical weapons pursuit (restrictive coding of pursuit), and increases the risk of
initiatingexploratorychemicalweaponsactivities(expansivecodingofpursuit)by31%. The
presence of nonviolent campaigns increase the risk of initiating chemical weapons pursuit,
broadly conceived, by 16–69% depending on model specification.
53
For biological weapons, the presence of a violent campaign against the government
increases the risk of initiating exploratory pursuit (expansive coding of pursuit) by 73%,
while the presence of a nonviolent campaign against the government decreases the risk of
initiating focused pursuit (restrictive coding of pursuit) by 61%.
Figure 4.2 illustrates the curvilinear relationship between changes in the level of ethnic
fractionalization and the risk of initiating weapons pursuit. Notably, increases in fraction-
alization increase the risk of initiating specific chemical weaponization activities (restrictive
coding of pursuit) up to about the 60th percentile of the fractionalization index (beyond
that, further increases in fractionalization decrease the risk of chemical weapons pursuit).
This pattern is consistent with a relationship between perceived risk of civil war and chemi-
cal weapons pursuit, given the ‘inverse-U’ shaped relationship between fractionalization and
civil war onset (Horowitz, 1895). However, increases in fractionalization are (somewhat sur-
prisingly) consistently associated with decreases in the risk of biological weapons pursuit
across the range of the fractionalization index.
National Security
Chemical and biological weapons pursuit also appear to be motivated in part by reactive
proliferation dynamics.
As stated in Hypothesis 3a, I expect that states with rivals that possess chemical
and/or biological weapons will be more likely to pursue their own chemical and/or biological
weapons capabilities. Evidence for this hypothesis is mixed. Consistent with the hypothesis,
gaining a rival armed with biological weapons is robustly associated with an increased risk
of initiating chemical weapons pursuit. This is true for both the expansive and restrictive
conceptionsofpursuit. Therelationshipbetweengainingabiological-armedrivalandtherisk
of initiating biological weapons pursuit is less robust (particularly for the more restrictive
coding of pursuit), but also appears to be positive. However, gaining a chemical armed
rival alone is not positively associated with an increased risk of initiating either chemical or
54
Figure 4.3: E↵ ects of external security threat indicators on risk of weapons pursuit. Hazard
Ratios displayed here are from Cox proportional hazard models including both internal and
external security threat indicators. Models are stratified by country.
biological weapons pursuit.
3
However, consistent with Hypothesis 3b, having a chemical armed rival choose to use
those weapons in conflict is associated with an increased risk of initiating chemical weapons
pursuit (both broad exploratory pursuit and pursuit focused on specific weaponization and
delivery systems), and biological weapons exploration. These e↵ ects are positive and sta-
3
A positive relationship does appear in some models of focused biological weapons pursuit
(restrictive pursuit coding), but is sensitive to model specification.
55
tistically significant across fixed-e↵ ects model specifications, with the exception of the “full
demand function” models, where, as discussed above, we need to be wary of the possibility
of post-treatment bias. On the other hand, rival chemical weapons use appears to have a
weak negative e↵ ect on the risk of initiating focused biological weapons pursuit. Increases in
the number of a state’s strategic rivals are also associated with an increased risk of initiating
biological weapons pursuit.
As Figure 4.3 illustrates, these e↵ ects can be substantively quite large: a rival’s ac-
quisition of biological weapons roughly doubles the risk that a state will initiate focused
chemical weapons pursuit, increases the risk that a state will begin broad exploration of
chemical weapons capabilities by a factor of 15, and triples the risk of initiating biological
weapons exploration. A rival’s use of chemical weapons increases the risk that a state will
begin exploring chemical weapons capabilities by a factor of roughly 12.5, increases the risk
of focused chemical weapons pursuit by a factor of 2, and increases the risk of beginning
biological weapons exploration by 32%.
According to Hypothesis 4, gaining a nuclear armed rival should make a state more
likely to initiate chemical and/or biological weapons pursuit. Models of focused pursuit
(restrictive coding) provide strong support for this hypothesis: the e↵ ects of rival nuclear
weapons on both the risk of initiating focused chemical weapons pursuit and the risk of
initiating focused biological weapons pursuit are positive and statistically significant across
fixed-e↵ ects model specifications. Furthermore, these e↵ ects are substantively large: as
illustrated in Figure 4.3, gaining a nuclear armed rival increases the risk of initiating focused
chemical weapons pursuit by a factor of 12, and increases the risk of initiating focused
biological weapons pursuit by 73%.
On the other hand, gaining a nuclear armed rival is only positively related to ex-
ploratory pursuit before controlling for other external and internal security factors beyond
rivals’ WMD possession.
Hypothesis 5 states that increases in available resources and more scientific and tech-
56
Figure 4.4: E↵ ects of nuclear infrastructure index on risk of weapons pursuit. Hazard Ratios
displayedherearefromCoxproportionalhazardmodelsincludingbothinternalandexternal
security threat indicators. Models are stratified by country.
nical infrastructure relevant to the development of nuclear weapons should be less likely
to pursue chemical and biological weapons. Consistent with this hypothesis, for a given
state, increases in overall resources (GDPpc) and nuclear weapons specific infrastructure
both decrease the risk of initiating focused pursuit (restrictive coding) of either chemical or
biologicalweapons,andtheriskinitiatingexploratorypursuit(expansivecoding)ofchemical
weapons (possibly because they facilitate the pursuit of more desirable weapons options).
However, there is a positive but not robust relationship between these variables and the risk
of initiating biological weapons exploration.
Figure 4.4 illustrates the substantive e↵ ects of changes in the available nuclear infras-
tructure on CBW pursuit. A one standard deviation increase in the nuclear infrastructure
index reduces the risk of initiating focused pursuit of either chemical or biological weapons
by roughly 90%.
Between-country Variation
Thus far, I have focused on the results of stratified (fixed e↵ ects-like) models aimed at
identifying time-varying factors that increase or decrease a given country’s risk of initiating
57
Figure 4.5: E↵ ects of internal security threat indicators on risk of weapons pursuit. Hazard
Ratios displayed here are from Cox proportional hazard models including both internal and
external security threat indicators.
weapons pursuit. I now turn to models that are not stratified, in order to discuss factors
that are associated with between-country di↵ erences in the likelihood of pursuit. Regime
security factors appear to explain between-country patterns of chemical weapons pursuit,
but not biological weapons pursuit.
Countries where the regime faces higher levels of coup risk are at increased risk of
chemical weapons pursuit, for both the expansive and restrictive conceptions of pursuit, an
association that is generally robust across model specifications. However, countries with
higher baseline levels of coup risk are less likely to pursue biological weapons (regardless of
58
Figure 4.6: E↵ ects of ethnic fractionalization on risk of weapons pursuit. Hazard Ratios
displayedherearefromCoxproportionalhazardmodelsincludingbothinternalandexternal
security threat indicators.
the coding of pursuit).
Higher levels of violent unrest are robustly associated with chemical weapons explo-
ration, but only weakly associated with focused pursuit. Additionally, nonviolent unrest
is weakly associated with both focused chemical weapons pursuit and exploratory chemi-
cal weapons pursuit. However, neither violent nor nonviolent unrest are associated with
between-country variation in the likelihood of biological weapons pursuit.
Turningtothesubstantivee↵ ectsofthesevariables,showninFigure4.5,aonestandard
deviationincreaseintheriskofacoupincreasestheriskoffocusedchemicalweaponspursuit
59
Figure 4.7: E↵ ects of external security threat indicators on risk of weapons pursuit. Hazard
Ratios displayed here are from Cox proportional hazard models including both internal and
external security threat indicators.
by 28% and the risk of exploratory chemical weapons pursuit by 50%. A one standard
deviation increase in coup risk decreases the risk of focused biological weapons pursuit by
33%. Violent unrest increases the risk of chemical weapons exploration by a factor of 2.7.
Figure 4.6 illustrates the relationship between between-country di↵ erences in the level
of ethnic fractionalization and the risk of initiating weapons pursuit. Controlling for other
internal and external security factors, countries at higher levels of fractionalization appear
to be less likely to pursue chemical or biological weapons.
Turning to national security factors, countries with biological-armed rivals are more
60
Figure 4.8: E↵ ects of nuclear infrastructure index on risk of weapons pursuit. Hazard Ratios
displayedherearefromCoxproportionalhazardmodelsincludingbothinternalandexternal
security threat indicators.
likely to engage in focused chemical weapons pursuit, a relationship that is robust across
models. The relationship between biological-armed rivals and the other forms of CBW
pursuit is also generally positive, but less robust. Countries with chemical-armed rivals are
more likely to engage in focused biological weapons pursuit, a relationship that is robust
across models, but there is no consistent relationship between chemical-armed rivals and the
other forms of CBW pursuit. There is also no consistent relationship between rival chemical
weapons use and between country di↵ erences in the likelihood of CBW pursuit.
As shown in Figure 4.7 having a chemical-armed rival increases the risk of focused
biological weapons pursuit by 62%. Having a biological-armed rival increases the risk of
focusedchemicalweaponspursuitbyafactorofroughly4,andtheriskoffocusedbioweapons
pursuit by a factor of 2.8. Having a biological-armed rival increases the risks of chemical
weapons exploration and bioweapons exploration by 163% and 187% respectively. Countries
with non-WMD armed strategic rivals are also more likely to engage in weapons pursuit, an
association which holds across
Countries with nuclear-armed rivals are also more likely to engage in both chemical
weapons pursuit and biological weapons pursuit. This relationship is particularly robust
61
when looking at exploratory pursuit (the expansive coding of pursuit). Having a nuclear
armed rival increases the risk of focused chemical weapons exploration by a factor of 3.5,
and doubles the risk of bioweapons exploration. For focused pursuit, having a nuclear-armed
rival increases both the risk of chemical weapons pursuit and the risk of biological weapons
pursuit by roughly 50%.
While within country positive shocks to nuclear infrastructure and resources decrease
the risk of initiating weapons pursuit, when assessing di↵ erences between states, higher
baseline nuclear industrial capacity, is associated with a higher risk of chemical weapons
pursuit (both focused and exploratory), and a higher risk of focused biological weapons
pursuit (possibly by making CBW programs feasible at all). There do not appear to be
consistent e↵ ects of per-capita GDP in the between-country models.
Discussion
These results suggest that internal security threats, while often disregarded in theories of
arming,playasignificantroleinmotivatingstatestopursuechemicalandbiologicalweapons
proliferation. In particular, governing regimes facing increases in the risk of a coup may be
more likely to initiate chemical and biological weapons programs. Furthermore, regimes
experiencing violent domestic unrest, and concerned about the risk of civil war, may be
more likely to pursue chemical weapons.
This is not to say that external security factors play no role in motivating these deci-
sions. Indeed,thisanalysiso↵ erstentativeevidencefortwopathwaysthroughwhichexternal
security threats can motivate CBW proliferation. First, acquisition of strategic weapons (in
particularnuclear, andsometimesbiological)byrivalscanpromptstatestopursuetheirown
CBWcapabilities, inaformof“reactiveproliferation.”Second, holdinglevelsofinternaland
external threat constant, states facing negative shocks to industrial and economic resources
may be more likely to pursue CBW — perceived as “cheaper” and easier to acquire — as a
(partial) substitute for other arming options (such as nuclear weapons pursuit). Traditional
62
indicators of states’ security environments perform inconsistently at explaining patterns of
CBW pursuit, suggesting either that CBW proliferation is not primarily driven by conven-
tional balancing dynamics, or that we need to seek out better measures of states’ perceived
external threat environments.
While the substantive e↵ ects of the international security variables can sometimes be
quite large, the internal security factors explain a greater proportion of the variation in
observed CBW pursuit choices. Because WMD acquisition is relatively rare, having a rival
acquire chemical, biological, or nuclear weapons is relatively rare. Thus, even if cases of rival
WMD acquisition are strongly associated with decisions to initiate CBW pursuit, there may
be numerous cases of CBW pursuit that are not associated with rival WMD acquisition.
One notable takeaway is that biological weapons, while still relevant to internal secu-
rity, may be closer to traditional “strategic weapons” than chemical weapons. States appear
to balance other state rivals’ bioweapons acquisition, by pursuing their own chemical and
bioweapons capabilities. Furthermore, biological weapons appear be seen as having limited
utility against domestic threats from the general public. This is consistent with theoretical
predictions, given bioweapons relatively greater destructiveness and technical complexity,
relative to chemical weapons and the potential unpredictability of these weapons e↵ ects
when used in a domestic context.
Theanalysispresentedhereidentifieskeyvariablesthatshapechoicestoinitiatechem-
ical or biological pursuit within a given country. However, these variables are often less
consistently able to explain cross-country variation in CBW proliferation behavior, suggest-
ing a need for further work that identifies the factors that drive di↵ erences in the baseline
likelihood of CBW pursuit between countries.
63
Chapter 5
Libya: Substitution in a
“Coup-proofed” State
ThischapterexaminesthetrajectoryofLibya’schemicalweaponsprogramandthepotential
motivations for chemical weapons pursuit. Libya is a valuable case for analysis for several
reasons. First, Libya experienced both significant regime security threats and significant
national security threats in the time period during and immediately preceding its chemical
weapons program, providing ample opportunity to investigate the mechanisms linking these
threats to decisions about chemical weapons proliferation activities. Second, Libya actively
pursued nuclear weapons in addition to chemical weapons, providing an opportunity to
examine the relationship between proliferation choices in the nuclear and chemical domains.
Finally, in 2003, Libya agreed to dismantle its WMD programs, and provide full accounts of
those programs to inspectors from the US, UK, IAEA, and OPCW. Thus, information on
the chemical and nuclear weapons programs is relatively accessible — which, given the high
levels of secrecy that typically surround states’ CBW activities, would not be the case for
an ongoing program, or a program that had never been declared.
The Libyan case illustrates how regime security factors can contribute to substitution
dynamics. A “coup-proofed” state lacked the institutional capacity to e↵ ectively monitor
64
and manage a large, complex nuclear weapons program, making a smaller-scale chemical
weapons program a desirable option to counter a range of security threats.
The rest of the chapter proceeds as follows: I first briefly provide some historical
background and an overview of Libya’s CBW e↵ orts. Next, I discuss the key factors that
may have shaped proliferation decision-making: the domestic threat environment faced by
the Qaddafi regime, the external security environment faced by the Libyan state, and the
available resources and ongoing nuclear weapons e↵ orts. I then discuss the trajectory of
Libya’s chemical weapons program, the agreement to dismantle the program, and the later
revelationofasmallundeclaredchemicalweaponsstockpileinmoredetail. Finally,Iattempt
to disentangle the influence of internal and external security threats on Libya’s chemical
weapons proliferation decisions, and conclude by linking the evidence from this case back to
the findings from the quantitative analysis presented in Chapter 4.
Background
Libya is located in the Maghreb region of North Africa, and consists of three historical
regions: Tripolitania (the western costal region which includes the capital, Tripoli), Fezzan
(the southwestern dessert), and Cyrenaica (the eastern region bordering Egypt). Primarily
inhabited by Berbers, the region historically fell under the control of various Mediterranean
empires — most recently the Ottoman Empire, which controlled the costal region beginning
in1551. FollowingtheItalo-Turkishwar, in1911, ItalyoccupiedTripolitaniaandCyrenaica,
unifying them as “Italian Libia” in 1929. After Italy’s defeat in the Second World War, the
fate of the Italian colonies was left up to the victors. In 1951, Tripolitania and Cyrenaica,
in combination with the previously French-administered province of Fezzan, were granted
independence, forming the Kingdom of Libya.
1
1
For comprehensive histories of modern Libya, see Vandewalle (2018, 2012); Ahmida (1994);
Anderson(1986);Cooley(1982). ForhistoriesoftherecentcivilwarsandtheoverthrowofQaddafi,
65
The Great Powers selected of Idris al-Sanusi, heir to the Sanussiyya movement which
had spearheaded resistance to Italian rule, to head the newly independent monarchy. Idris
ruled from Benghazi, the largest city in Cyrenaica, but enjoyed little legitimacy outside of
the region. In 1969, he was removed in a bloodless coup by the populist, Arab nationalist
Revolutionary Command Council (RCC), led by Muammar al-Qaddafi. Qaddafi would go
on to rule Libya for 42 years, before being overthrown in a civil war which broke out amidst
the wave of Arab Spring protests in 2011. Completing claims to authority following the 2014
elections led to another civil war, which lasted until a ceasefire agreement in 2020.
While the country is geographically large, its small population is mostly concentrated
in costal cities. The majority of the population is Arab or Berber, and there are about 140
tribes in Libya, roughly 30 of which carry significant political influence. The majority of the
Italian settler population left at the time of independence, or in the early years of Qaddafi’s
rule. However, migrant labor plays an important role in the Libyan economy. Mostly
consisting of dessert, and previously subject to brutal colonial rule, Libya was extremely
poor at the time of independence, but the discovery of oil in 1959 brought a vast influx of
revenue. With the world’s 10th largest oil reserves, the Libyan economy quickly became
almost entirely dependent on oil exports.
Libya’s CBW capabilities and programs: an overview
While Libya pursued chemical and nuclear weapons, and at least engaged in exploratory
biological weapons activities, the chemical weapons program was their only successful WMD
e↵ ort. Libya is coded as pursuing chemical weapons beginning in 1980 and possessing them
beginning in 1988. However, exploration of a chemical weapons option may have begun as
early as 1976 (Horowitz and Narang, 2014).
see the volumes reviewed in Anderson (2020).
66
The central component of Libya’s chemical weapons program was the production of
sulfur mustard, also known as mustard gas, a blister agent first weaponized in World War I.
Sulfur mustard causes severe burning and blistering of the skin and mucous membranes, and
can cause death if burning of the skin or lungs is extensive. It also damages DNA, and if not
initiallyfatal,canleadtocancerinthelongerterm(Rall,Pechuraetal.,1993). Additionally,
the Libya attempted — less successfully — to produce several nerve agents, including soman
and sarin. Libyan o cials acknowledged testing activity that included potency tests of the
sulfur mustard and field trials of aerial bombs containing nonlethal agents, but denied any
more extensive outdoor testing of live agents (Tucker, 2009).
Following the rollback of the o↵ ensive chemical weapons program, which began in
December 2003, Libya’s declarations to the Organisation for the Prohibition of Chemical
Weapons (OPCW) have revealed the extent of the program. This included three chemical
weapons production facilities, at Rabata, Sebha, and Tarhuna, and 25 metric tons of sulfur
mustard, stored plastic containers in bunkers at two separate military bases near Tripoli.
Additionally, the declarations included more than 1400 metric tons of imported precursors
for the production of mustard and nerve agent.
2
A further two thousand metric tons of
dual-use chemicals intended for use in the production of mustard and sarin.
3
Fordeliverysystems, theLibyanprogramhadfocusedonaerialbombs. Libyadeclared
more than 3,500 such bomb casings, each designed to hold 48 liters of mustard, which would
be dispersed on impact by a small central conventional explosive (Tucker, 2009). These
munitions were stored unfilled in peacetime.
In addition to the large-scale production of sulfur mustard, the Libyans had produced
2
The scheduled chemical precursors Libya declared included: phosphorus trichloride,
dichloroethane, thionyl chloride, thiodiglycol, ethylene oxide, dimethylamine, sodium sulfide,
sodium fluoride, and pinacolyl alcohol.
3
These chemicals were not scheduled under the Chemical Weapons Convention, as they have
legitimate peaceful uses. However, Article II of the CWC requires declaration—and verifiable
destruction—of even unscheduled precursors acquired for the purposes of developing or producing
o↵ ensive chemical weapons.
67
small quantities of soman and sarin. In an e↵ ort to scale up production, they had imported
specialized corrosion-resistant equipment suitable for work with nerve agents, and industrial
quantities of some precursors. However, o cials reported that e↵ orts to scale up production
had been stymied by technical challenges, and international sanctions eventually made it
di cult to acquire large quantities of key precursors (Tucker, 2009).
US and British inspectors were surprised to discover that Libya’s biological weapons
programapparentlyneveradvancedpasttheearlyexplorationstage.
4
Theremayhavebeen
some (unsuccessful) e↵ orts to procure materials and equipment from foreign suppliers, and
possiblysomesmallscalelaboratoryexperiments(Tucker,2009). WhilesomeLibyano cials
admitted to inspectors that an o↵ ensive biological weapons program had been considered,
one o cial reported that the program “never went beyond the planning stage” and even
those e↵ orts were terminated in the early 1990s because Qadafi considered an o↵ ensive
biological weapons program “too dangerous,” (Robb et al., 2005, p. 255). Lower-level
o cials confirmed the end of the program in the 1990s, but maintained that activities had
been purely defensive in nature (Tucker, 2009), and all potential bioweapons facilities visited
by US and British inspectors were ultimately determined to have legitimate pharmaceutical
purposes (Prados, 2005).
The internal threat environment
Libya’s pursuit of chemical weapons took place during the rule of personalist leader Muam-
mar al-Qaddafi. Qaddafi, as head of the Revolutionary Command Council (RCC), came
to power in a 1969 coup that overthrew King Idris I, who had ruled since Libya gained
independence in 1951. The regime faced internal security concerns almost from the outset.
4
Horowitz and Narang (2014) code Libya as having pursued biological weapons from 1988 to
2000, based on their broad conception of “pursuit.” However, their sources do not appear to give
concrete reasons for choosing those particular dates.
68
Concerned about the possibility of a coup, Qadda e↵ ectively dissolved the Defense Min-
istry in 1970, placing the military under his direct control (Braut-Hegghammer, 2016). At
the same time, in an e↵ ort to buy o cer’s loyalty, he poured massive sums of money into a
conventional arms buildup (Gaub, 2013).
While Qaddafi initially pursued a range of popular policies aimed at providing public
goods such as education, healthcare, and housing, reducing the influence of Italy (Libya’s
former colonizer) and seizing Italian assets, and improving the social position of women, by
1973 he and other top leaders were concerned that popular support for their revolutionary
endeavors was not measuring up to their expectations (Anderson, 1982). In response, the
governmentlauncheda“culturalrevolution,” whichtookplacebetween1973and1977. This
series of “reforms” was a radical transformation of the state, which further concentrated
power in Qaddafi’s hands and e↵ ectively dismantled or disempowered most existing state
institutions. The constitution was eliminated, political dissent was outlawed, and grassroots
“popularcommittees”wereestablishedtoreplacestateagencies(Braut-Hegghammer,2016).
When these committees produced insu cient public engagement — and when a 1975 coup
plot by two members of the RCC revealed ideological divisions — Qaddafi dissolved the
RCC set up and a shadow set of institutions, the Revolutionary Committees, to surveil the
public, monitor o cial state institutions, and in some cases, replace them. (Cooley, 1982;
Gaub, 2013; Braut-Hegghammer, 2016).
In the mid-1980s, worsening economic conditions combined with frustrations over
Qaddafi’s adventurist foreign policy to produce increasing discontent within the military.
The regime was forced to cut back on spending that had served to buy loyalty. Hundreds of
millions of dollars in planned military construction were cancelled, promotions were frozen,
and generous housing, commissary, and travel privileges for o cers were limited. Further-
more,constanttensionswiththeUS,Tunisia,andEgypt,andthemaintenanceofanongoing
military presence in Chad were wearing on the military, and o cers resented Qaddafi’s ideo-
logical disdain for formal military hierarchy (Schumacher, 1986). These frustrations spurred
69
three coup plots in 1983, 1984, and 1985, all of which were foiled (Gaub, 2013).
Economic woes were exacerbated by the international sanctions, and in the mid-1990s
unemployment soared to 30% and inflation to 50%, forcing the regime to impose austerity
measures(Collins,2004;Takeyh,2001). Thestandardoflivingforthegeneralpublicsu↵ ered,
while inequalities between the richest few families with ties to the regime and the rest of
the population grew, leading to widespread public discontent. Riots erupted, opposition
groups became more active, and an islamist insurgency in Cyrenaica (an eastern region of
the country) that erupted in 1995 took three years to subdue (Braut-Hegghammer, 2016;
Takeyh, 2001). Complicating the regime’s response, there had been another coup attempt
in 1993, so rather than use the formal military to crush domestic opposition, Qaddafi relied
on the Revolutionary Guards — a separate unit of troops a liated with the revolutionary
committees (Gaub, 2013). The 1993 coup attempt was followed by a purge of the military,
the arrest of thousands, and numerous executions (Bell and Witter, 2011). This brutal
repression led many regime opponents to flee the country (Emadi, 2012).
In combination with domestic economic reforms, the lifting of economic sanctions
that followed Libya’s 2003 renunciation of its WMD programs and sponsorship of terrorism
improved Libya’s economic fortunes. With material conditions for the general population
improving, and many regime opponents killed, subdued, or chased out of the country, the
mid- to late 2000s were a period of relative domestic peace and stability (Bell and Witter,
2011). However, in February 2011, in the wake of the Arab Spring, large-scale protests
broke out in Cyrenaica, eventually escalating into a civil war that would result in Qaddafi’s
overthrow.
The external threat environment
Libya faced a fairly high level of external threat from the start of Qaddafi’s rule until the
2003 agreement that allowed for the normalization of diplomatic relations with much of the
70
West. Qaddafi purported to subscribe to a foreign policy of pan-Arab nationalism and anti-
imperialism, an approach that both alienated the West, and was unsuccessful in fostering
support from neighboring states (Anderson, 1982).
Qaddafi’s e↵ orts to stir up anti-American sentiment in the Arab world, and his fervent
opposition to Israel and support of Palestine created tensions almost form the beginning of
his rule.The US recalled it’s ambassador from Libya in 1972, and imposed an arms embargo
in 1973 (Emadi, 2012). Relations with neighboring Egypt also deteriorated when Egyptian
President Sadat began to align more closely with the US (and become more accommodating
toward Israel) following the 1973 October War. In 1977, diplomatic tensions broke out into
a brief war — which Libya lost — in July 1977, following Qaddafi’s attempt to expel all
Egyptians from Libya.
Frustrated with a lack of support from other Arab leaders, in the mid-1970s, the
Qaddafi regime began funding dissidents and insurgents opposing regimes he deemed re-
actionary. Support consisted primarily of financing, training, and arms exports, although
troopsweresenttoUgandain1979inan(unsuccessfulanddomesticallyunpopular)e↵ ortto
prop up Idi Amin’s failing regime (Anderson, 1982). The Libyan military was also engaged
in conflict with Chad from 1978 to 1987, over a strip of territory along the border between
the two countries (Gaub, 2013).
5
Relations with the United States, which had been strained throughout the 1970s, as
the US accused Qaddafi of supporting international terrorist groups, deteriorated further
in December 1979 when a mob burned the US Embassy in Tripoli, leading the US to end
diplomaticrelations. Tensionsgrewinthefollowingyears, withUSandLibyanforcesgetting
into a minor skirmish in the Gulf of Sidra in 1981, and the US imposing an oil embargo and
other sanctions in 1982 (Anderson, 1982). In 1986, another incursion into the disputed
5
Control of this 60-mile, strip of dessert, known as the Aouzou strip, had been contested since
colonial times. Libya occupied the territory in 1972, a move permitted via a secret agreement with
Chadian President Tombalbaye — but Tombalbaye then lost power in a 1975 coup.
71
Gulf led to a day of fighting between US and Libyan Forces. Two weeks later, based on
intelligence pointing to Libyan involvement in the bombing of a West Berlin disco, the
US bombed targets in Tripoli and Benghazi, including Qaddafi’s house and o ce building
(although Qaddafi himself was unharmed) (Schumacher, 1986).
Libya’sinvolvementinthe1988bombingofanairlineroverLockerbie,Scotland,killing
270 people was the culmination of a string of terrorist incidents that prompted the UN to
impose severe economic sanctions, and Libya’s failure to extradite the bombing suspects let
to further sanctions in 1992 (Tucker, 2009; Takeyh, 2001). Furthermore, with the collapse of
the Soviet Union, Libya lost the main counterweight to its superpower rival (Takeyh, 2001).
While the external security situation began to improve somewhat in 1999, following
the extradition of the Lockerbie suspects, the Bush Administration’s 2001 declaration of a
“global war on terror” raised fears within the Qaddafi regime. Though there is much debate
over the sources of counterproliferation success in Libya (Busch and Pilat, 2017), these fears
likely contributed to Qaddafi’s willingness to accept the 2003 WMD rollback agreement,
which heralded a period of unprecedented freedom from external threats.
In addition to having the United States as a nuclear-armed rival, Libya was positioned
in a regional context rife with chemical weapons activity. Egypt, with which Libya was
involvedinarivalryfrom1973to1992,hadactivechemicalandbiologicalweaponsprograms,
and had used mustard gas and phosgene during the Yemen War in the 1960s (Terrill, 1994).
Elsewhere in the region, Iraq began pursuing chemical weapons in 1971 and Syria in 1973,
and and Iraq began using some of those weapons against Iran in the early 1980s.
Economic resources and nuclear infrastructure
Libya’s vast oil reserves fueled rapid growth and development in the 1970s and provided
ample resources for Qaddafi to invest in military adventurism, sponsorship of terrorist and
insurgent organizations, and a conventional arms buildup, as well as the nuclear and chem-
72
ical weapons programs (Bell and Witter, 2011). However, in 1981 oil prices began a sharp
decline that would last for most of the decade. Libya’s economy su↵ ered even more than
those of other oil-dependent states, partly because US sanctions limited access to some spe-
cializedoilproductionequipmentandknowledge, reducingproductivityparticularlyatfields
previously operated by American companies, and partly because, with declining revenues,
the regime was forced to cut back on previously lavish productivity incentives (Barnum and
Fearey, 2016). As discussed above, multilateral UN sanctions, imposed in 1992, significantly
worsened Libya’s economic situation.
While Libya invested substantial resources into nuclear weapons pursuit, the program
failed to make much progress over its more than 30 year duration (Braut-Hegghammer,
2016). Libya’s nuclear e↵ orts began in the 1970’s, placing particular emphasis on attaining
technical expertise and training from abroad. Over the course of the decade, the regime
approached numerous nuclear suppliers. Concerns about Gadhafi’s nuclear intentions led to
rejectionsfromtheU.S.,France,China,andothers. AninitialnuclearconnectiontoPakistan
was established in the early 1970s, with Libya providing significant financial support to the
Pakistani nuclear program. Under a 1975 agreement, the Soviet Union provided Libya with
the Tajoura Nuclear Research Center, including a 10MW light-water reactor. In return,
Libya ratified the Nonproliferation Treaty and agreed to establish a safeguards agreement
with the International Atomic Energy Agency (IAEA) (Bowen, 2017, p. 25-31).
After failed attempts to discover exploitable domestic uranium deposits, Libya im-
ported more than 1,200 tons of yellowcake uranium ore from Niger between 1978 and 1981,
some portion of which may have been secretly transferred to Pakistan (Director General,
2004, Annex 1). Uranium enrichment and plutonium separation e↵ orts began in the early
1980s (Bowen, 2017, p. 32-36). In the mid-80s Libya purchased enrichment equipment
including a modular Uranium conversion facility and specialized furnace from Japan and
vacuumpumpsfromEurope. Theyalsoreceiveduraniumhexafluorideanduraniumtetraflu-
oride from a nuclear weapons state, presumably the Soviet Union, in return for shipments
73
of yellowcake (Director General, 2004). New enrichment activities slowed in the early 1990s,
and in 1995, the regime began to move enrichment to new, decentralized locations for con-
cealment purposes. In 1997, the regime established contact with Abdul Qadeer Khan, and
began using his network to acquire technical training, centrifuges, uranium hexafluoride gas,
and other materials (Bowen, 2017, p. 36-38). They had established a complete nine cen-
trifuge cascade and were at varying stages of completion on 19 and 64 centrifuge cascades by
April 2002 when security concerns led the government to dismantle the cascades and move
the equipment to less central locations. The A. Q. Khan network also provided Libya with
incomplete design and fabrication documentation for the physics package of a late 1960s
Chinese weapon (Director General, 2004, p. 3, 5, 7).
Timeline of the Chemical Weapons Program
Libya’s chemical weapons activities appear to have begun in the early 1980s, with attempts
to reach out to foreign suppliers for the provision of CW materials and equipment. As early
as 1980, the West German intelligence service (BND) was receiving reports raising concern
over Libya’s interest in o↵ ensive chemical weapons — and the potential involvement of
German firms in those e↵ orts (Bundestag, 1989). While initial concerns centered around the
possibility of a concealed mustard agent production facility at the Abu Kammash chemical
complex, the BND later concluded that the plant was a “normal chemical factory,” and no
evidence of weapons activities at Abu Kammash came to light in Libya’s declarations to the
OPCW and US and British inspection teams. However, these reports also included evidence
of Libyan e↵ orts to buy precursor chemicals from various European suppliers, and former
employees’ of German chemical companies receiving large sums of money in exchange for
time spent in Libya providing unknown consulting services.
Libyan e↵ orts to build an actual chemical weapons production facility began to come
to fruition in 1984, when the Libyan government entered into an agreement with Ihsan
74
Barbouti International (IBI). Under the terms of the agreement, IBI would act as a middle-
man to coordinate with foreign suppliers. That September, IBI facilitated a secret contract
with a West German firm, Imhausen Chemie, to design and build a chemical weapons pro-
duction facility at Rabata. Imhausen Chemie’s Director, J¨ urgen Hippenstiel-Imhausen was
arrested in 1989, following the release of the infamous “Sch¨ auble report,” which documented
the BND’s knowledge of West-German firms involvement in constructing Libya’s chemical
weapons facilities. Hippenstiel-Imhausen admitted responsibility for the construction of the
RabataCWfacility, andpledguiltytoviolatingvariousWestGermanexportandtaxlawsin
association the company’s provision of secret chemical weapons assistance to Libya (Terrill,
1994). The plant at Rabata would go on to be the primary production site identified in
Libya’s 2004 declaration to the OPCW (NTI, 2011).
Libya’s initial declaration to the OPCW on March 5, 2004 described the 23 declared
tonsofsulfurmustardasproducedatRabatabetween1980and1990(Kerr,2008). However,
itislikelythatmuch,ifnotallofthatproductiondidnotoccuruntilthelate1980s. Basedon
reports from the (mostly Thai) construction workers that worked on the plant, construction
of the facility at Rabata began in November 1984, and equipment was installed 1986 (Burck
and Flowerree, 1991; Wiegele, 1992). In the summer of 1987 the BND informed the West
German government that construction at the Rabata plant was nearly complete and the
facility would be ready to start weapons production within a few weeks. However, this
assessment may have been premature, as BND reports and communications with the US
Embassy suggest that by September 1988 both countries’ intelligence services concluded
that the plant was still not operational — but would be soon (Bundestag, 1989). Production
likely began sometime in late 1988 or early 1990. In January 1989 US o cials reported that
production of small quantities of chemical agents had begun, but that full scale production
had not, and by early 1990, numerous reports from US government conclude that limited
production began sometime the previous year (Burck and Flowerree, 1991).
At Rabata, IBI and Imhausen Chemie built a large industrial complex, “Pharma 150,”
75
for the manufacture of commercial pharmaceuticals. The chemical weapons plant was con-
cealedin“Building17”ofthiscomplex. Mostofthespaceinthisdual-usefacilitycontaineda
variety of chemical reactors used for commercial pharmaceutical manufacturing, but roughly
athirdofthespacewaswalledo↵ and concealed from the rest of the facility. This space
housed a dedicated production line for sulfur mustard. US and British inspectors reported
that this lineconsisted ofhigh quality, specialized equipment that would also havebeen suit-
able for the production of nerve agents,
6
with highly advanced automated process control
systems. One US o cial involved in the 2004 inspections assessed the Rabata plant as more
sophisticated than Soviet and Iraqi chemical weapons facilities (Tucker, 2009).
Sulfur mustard was synthesized at Rabata using the Myer process, which involves re-
actingaprecursor,thiodiglycol(TDG),withachlorinatingagent(Rall,Pechuraetal.,1993).
The Libyans initially purchased thiodiglycol from abroad, but turned to manufacturing it
themselves in the commercial part of Building 17 after UN sanctions made it impossible to
import in su cient quantities. This process involved reacting two simpler precursor chemi-
cals and then piping the result into the concealed production line.
InadditiontohelpfromIBIandImhausen,Libyalikelyalsoreceivedchemicalweapons
assistance from a Japanese company, Japan Steel Works, which built a metalworking plant
at the Rabta complex (Clark and Mitchell, 2018; Herdman, 1993). This metalworking fa-
cility was outfitted with high-precision Japanese-made metalworking tools, which would be
suitable for manufacturing artillery shells, aerial bombs, and corrosion-resistant containers
for the storage of chemical weapons agents. While Japan Steel Works insisted that these
machines were intended for the production of irrigation pumps and desalination equipment,
several factors led US and British intelligence services to be suspicious of these o cial pur-
poses. These included the co-location and apparent joint planning of the metalworking and
chemical/pharmaceutical manufacturing facilities, enhanced safety and security features of
6
Part of what made this equipment potential suitable for producing nerve agents was the use of
a corrosion-resistant, high-nickel type of steel known as Hallestoy.
76
the buildings, and, most importantly, the delivery of special steels typically used in the
production of bomb casings (Herdman, 1993).
Chemical weapons production at the Rabta plant appears to have been shut down
in late 1990. Following US threats to bomb the facility in 1989, the Libyan government
staged a “fire” at the facility — an elaborate hoax that involved painting scorch marks
on (undamaged) buildings, burning large piles of tires and other materials, and rushing
ambulances to the scene — with the goal of convincing Western intelligence services that
the facility was damaged and no longer operable (Clark and Mitchell, 2018; Terrill, 1994).
These e↵ orts at deception failed, as later satellite imagery revealed that the facility was still
intact, and the Libyan government eventually shut down the weapons production line at
Rabta and moved the specialized production equipment to storage an “open-air boneyard,”
where it remained until US and British inspectors visited the site in 2004.
After shutting down weapons production at Rabta, Libya reoriented its e↵ orts to-
ward the construction of two additional CW production facilities. In 1992, construction of
“Pharma 200,” an underground plant similar in design to the Rabta plant, was nearing com-
pletion. Like Rabta, it was designed by Imhausen, possibly with assistance from other West
German firms. Equipment was supplied by German, Swiss, and Italian companies, and some
precursor chemicals may have been provided by the Chinese government (Sinai, 1997; NTI,
2011). TheplantwaslocatedattheSebhaOasis, amilitarybaseneartheborderwithChad,
where other facilities had been used for missile testing and yellowcake uranium storage at
various times beginning in the 1980s. However, chemical weapons production never began
at Sebha, and inspectors found the site dormant (Prados, 2005; Sinai, 1997).
In 1994 preparations began to construct a new production plant at Tarhuna, in moun-
tainous territory 65 km south of Tripoli (Tucker, 2009). The Tarhuna complex consisted
of a maze of tunnels carved into the side of a mountain and lined with reinforced concrete.
The entrance was located in a narrow valley, shielding activity from aerial and satellite view,
and the reinforcement, combined with the sandstone of the mountain, hardened it against
77
conventional strikes (Sinai, 1997; Clark and Mitchell, 2018). Once again, the Libyan govern-
mentreliedonacomplexnetworkofinternationalsuppliersandmiddlemenforplantdesigns,
equipment, precursors, and construction equipment and workers. Specialized equipment was
purchased to equip the plant, including chemical reactors and piping lined with glass or
Teflon to prevent corrosion by toxic agents, a state-of-the-art air purification system, and
sophisticated automated process control systems that could be used to synthesize sarin or
soman.
The United States began a public campaign of accusations and pressure against the
Tarhuna facility in 1996. In April, CIA director John Deutch described Tarhuna as “the
world’s largest underground chemical weapons plant,” (Tucker, 2009), and Pentagon o cials
publicly speculated on the possibility of thermonuclear use to destroy the site: “if we wanted
to destroy [the Libyan plant], B-61 will be the nuclear weapon of choice,” (Pine, 1996).
Despite Qaddafi’s insistence that the work at Tarhuna was part of the “great man-made
river,” a water redistribution project, this pressure campaign was apparently successful,
as work on the Tarhuna project was terminated in late 1996, and according to inspectors
in 2004, the specialized production equipment imported for the facility remained outside
Tripoli, stored in its original shipping crates (Mahley, 2004).
A Chemical Weapons Reversal?
On December 19, 2003, Libya publicly announced an end to its covert nuclear, chemical
weapons, and ballistic missile programs, and agreed to dismantle its programs with technical
assistance and verification from U.S., UK, and IAEA experts, to adhere to its commitments
under the NPT, Biological Weapons Convention, Chemical Weapons Convention, and Mis-
sile Technology Control Regime, and to sign an IAEA Additional Protocol (Joseph, 2009).
This announcement came after months of secret negotiations over Libya’s WMD programs
between Libyan, US ,and UK intelligence o cials, beginning in March 2003, the interdiction
78
ofacontainershipboundforLibyacarryingthousandsofpartsforadvanceduraniumenrich-
mentcentrifugesinOctober2003,andtwoinitialvisitsbyUSandUKintelligenceteamswho
visited weapons facilities and interviewed scientists in order to asses Libya’s commitment to
disclosure (Tucker, 2009).
At this point, Libya’s chemical weapons research e↵ ort was still active, but large-scale
production had ended in the early 1990s with the closure of the Rabta facility, and produc-
tionhadnotcomeonlineateitherofthetwonewpotentialfacilities(atSebhaandTarhuna).
Libya joined the Chemical Weapons Convention on February 5th, 2004, submitted an ini-
tial declaration to the OPCW on February 20, and a complete declaration on March 5. By
March19, theOPCWannouncedthattheirinitialinspectionwascomplete, andthatLibya’s
declared stockpile of unfilled aerial bombs had been destroyed (OPCW). Initially, the dead-
line for complete destruction of Libya’s chemical weapons stockpiles, facilities, and delivery
systems was set for April 29, 2007, but as the process encountered delays and technical
di culties, it was extended twice: first in 2006 to December 2010, and then again in 2009
to December 2011 (Krutzsch, Myjer and Trapp, 2014). 51% of the sulfur mustard stockpile
and 40% of the stockpile of precursor chemicals had been destroyed by February 2011. Libya
had destroyed two of the three existing chemical weapons production facilities (Sebha and
Tarhuna), and converted the facility at Rabta to a civilian pharmaceutical production plant
— a somewhat complex legal process that involved petitioning the Executive Council of the
OPCW to amend a provision of the CWC Verification Annex in order to allow new state
parties to apply to convert chemical weapons facilities to civilian purposes (OPCW).
7
The process of destroying agents and precursors, already delayed by issues with the
disposal facility, was further pushed back by the outbreak of civil war in February 2011. A
7
Initially, the provision had stipulated that facilities could be converted in “exceptional cases
of compelling need,” if verification of the conversion was completed within six years of the treaty’s
entry into force. However, that April 2000 deadline had already passed by the time Libya joined
the treaty, and Libya made a successful case to the committee that amending the provision would
provide an incentive for new states to join the treaty.
79
seriesofprotests,partofthe“ArabSpring,”finallyescalatedintoafull-scalerebellioninlate
February, and OPCW inspectors left the country in in March 2011 in anticipation of NATO
airstrikes against Libyan military targets (Vishwanathan, 2012). Opposition forces gained
control of Tripoli in August, Qaddafi fled, and in September, the United Nations recognized
the National Transitional Council, the interim government organized by the rebels, as the
legal representative of Libya (Nesi, 2011).
That October, the transitional government announced that they had discovered a
small “previously undeclared chemical weapons stockpile,” at two previously undeclared
sites in central Libya. Initial reports suggested that the newly discovered materials included
afewhundredmunitionsfilledwithsulfurmustard(possiblycontainingupto1.6metric
tons of sulfur mustard total), and a few hundred kilograms of sulfur mustard stored in
plasticcontainers,aswellasunfilledartilleryshells(Vishwanathan,2012).OPCWinspectors
investigated the declaration, and released a brief statement on January 20, 2012. Inspectors
reported finding the newly declared materials stored at the Ruwagha depot with previously
declared materials that had not yet been destroyed. The o cial report was somewhat vague
in describing the declared materials:
The OPCW inspectors verified the declared chemical weapons, which consist of
sulfur mustard agent that is not loaded into munitions. At the same time, at the
request of the Libyan authorities the inspectors examined munitions, mainly ar-
tilleryshells,whichtheydeterminedarechemicalmunitionsandhencedeclarable.
(OPCW, 2012)
Inspectors declined to specify the amounts of agents and munitions included in the newly
declared materials, although a spokesman described the stockpile of sulfur mustard as “a
fraction of what was in the original declaration,” (Busch and Pilat, 2017, p. 122). The e↵ ect
ofthesecommentswastodeemphasizetheimportanceofthedeclaration—thestockpilewas
relatively small, and the artillery shells “not ready to use” because they were unfilled. Fur-
thermore, the statement made no reference to the two previously undeclared sites mentioned
in the National Transitional Council’s original report.
80
Yet the newly discovered materials raised troubling questions for some analysts, who
noted that, in combination, the undeclared sulfur mustard and artillery shells could have
served as a “small but potentially viable hedge chemical-weapon arsenal” for the Qaddafi
regime (Busch and Pilat, 2017, p. 123). A variety of factors raised suspicions that the new
materials had been intentionally withheld. The discovery of the shells was surprising given
that Libya had previously only declared aerial bombs — it seemed unlikely that these would
have gone unnoticed or improperly counted next to the declared materials. US o cials were
quoted as saying that the shells “were acquired over many years,” and “[w]e are pretty sure
we know” they were designed and produced by Iran for Libya (Smith, Warrick and Lynch,
2011). Additionally, while the OPCW statement appeared to contradict the transitional
government’s initial reports that the newly discovered materials included filled munitions,
the NTC coordinated with the US government to acquire equipment with the specification
that the destruction process be suitable for filled chemical munitions, as well as stored sulfur
mustard and conventional explosives (Terrell, Hagen and Ryba, 2016).
In light of the newly discovered materials, some of the issues encountered by the initial
inspectionteamsbecamemoreconcerning. WhileOPCWrepresentativesandUSandBritish
o cialsinvolvedtheinspectionanddisarmamentprocessweregenerallypositivelyimpressed
with the extent of Libya’s transparency and cooperation, some unanswered questions still
remainedaftertheinitialinspectionsin2004, particularlyhavingtodowithextentofLibyan
e↵ orts to produce nerve agents, and the extent of bioweapons research activities (Tucker,
2009). USandUKo cialshadfeltpressuretostarttheprocessofdismantlingLibya’sWMD
programs quickly, rather than waiting to resolve these lingering questions (Joseph, 2009).
There had also been discrepancies in the reporting of munitions, with the Libyans informing
the initial inspection team of only 1,500 of the 3,563 aerial bomb casings eventually reported
and destroyed. When questioned, the Libyans claimed that they had withheld information
about the two additional locations where the bomb casings were stored because they were
initially skeptical that Qaddafi would go through with the disarmament agreement (Tucker,
81
2009).
All three of the rollback e↵ orts — nuclear, chemical, and missile — had been plagued
by delays. OPCW deadlines for the destruction of materials were repeatedly pushed back,
as discussed above, a shipment of weapons-grade uranium to Russia was delayed and used as
abargainingchip,andtheQaddafiregimefailedtosellit’sstockpileofyellowcakeuranium
ore. HundredsofSCUD-Bmissileshadalsonotbeendestroyedby2011, andtheregimeused
several against rebels during the conflict (Busch and Pilat, 2017, p. 120-121).
8
Apparent
initial cooperation — which had so impressed the inspectors — may have actually been a
strategy to avoid complete declaration and lower the bar for verification. Certainly, coop-
eration had helped gain the Libyans some face-saving concessions early in the disarmament
process. For instance, while US and British inspectors suspected that multiple reactors in
the dual-use section of the Rabta plant had been used in the production of the precursor
TDG, and would need to be destroyed as part of the facility conversion procedure stipu-
lated by the CWC, Undersecretary of State Bolton “decided to give the Libyans a break”
by declaring only one reactor to be specialized chemical weapons equipment (Tucker, 2009).
The OPCW, in declining to be very specific about its verification of the newly discovered
materials and downplaying their importance, may have been trying to avoid embarrassment
or negative scrutiny of the original disarmament e↵ orts.
The NTC oversaw the elimination of Libya’s remaining Schedule I chemical weapons
(including the newly discovered materials), with the last stockpiles of sulfur mustard and
filled munitions destroyed by early February 2014, and, despite delays due to resurgent civil
conflict, the remaining precursor chemicals were disposed of by January 2018 (Dawson and
Linaki, 2019).
8
US o cals had apparently viewed these missiles as not a major source of concern, assessing
that most were likely not fully operational.
82
Analysis
As discussed above, throughout most of his regime, Qaddafi experienced significant security
concerns, in terms of both threats to his continued rule and external threats to the Libyan
state. The frequent co-occurence of increases in regime security concerns and increases in
national security concerns can make it challenging to identify which factors were driving
changes in chemical weapons pursuit behavior. For instance, Libya began considering of-
fensive chemical weapons pursuit and engaging in exploratory research in the latter half of
the 1970s, shortly after the Qaddafi regime survived its first major coup attempt in 1975.
However, relations with both the United States (a superpower) and Egypt (a contiguous
neighbor) had also started to worsen a few years earlier. While the regime security threat is
more proximate to the start of exploration, the brief war with Egypt that broke out in 1977
and the increase in diplomatic tensions with the United States following the 1979 embassy
burning preceded the start of Libya’s chemical weapons pursuit in the early 1980s.
9
A significant intensification of Libyan chemical weapons pursuit e↵ orts — the 1984
contract to begin construction at Rabta — did occur against the background of dramatically
worsening domestic discontent, particularly within the military, which Qaddafi recognized
as a key power center and the most likely source of existential threats to his regime (Gaub,
2013). Coups were attempted in 1983 and early 1984, and another attempt would follow
in 1985. While it is the case that relations with the US were gradually worsening in the
early 1980s (with the first Gulf of Sidra incident 1981 and the oil embargo in 1982), the
most significant escalation of tensions did not occur until 1986, well after work at Rabta had
begun. The more likely external security concern that may have motivated the increase in
chemical weapons e↵ ort at this time is Iraq’s use of chemical weapons during the Iran-Iraq
9
TheseincidentswerebothinstigatedbytheLibyangovernment,andcouldpotentiallybeviewed
as diversionary tactics intended to generate more popular support for the regime. Scholars have
interpreted Qaddafi’s attempts to “export the revolution” as motivated in part by frustrations over
his inability to fully consolidate domestic popular support. See, e.g., Anderson (1982).
83
war. Although the first reports of Iraqi CW use against Iranian forces arose in November
1980, the Iraqi strategy did not involve regular, large-scale use of chemical weapons until
1983 (Ali, 2001).
While sulfur mustard production at Rabta was shut down by the end of 1990, appar-
ently in response to US threats to bomb the facility, and amidst broadly worsening external
security conditions, as the international community turned against Libya in the wake of the
Lockerbie bombing, the early 1990s saw Libyan e↵ orts to expand CW production to new,
less vulnerable facilities at Sebha and Tarhuna. The more ambitious work at Tarhuna, in
particular, began in 1994, in the context of significant domestic unrest in response to eco-
nomic hardship and austerity measures, and followed in the wake of another coup attempt
in 1993.
Further clarity on the factors motivating Libya’s chemical weapons pursuit can be
gained by examining the timing of chemical weapons activities, and the chemical weapons
program’s relation to the nuclear program. If both weapons programs were intended to
deter external adversaries, or to backstop an adventurist foreign policy, why did the Libyan
regime’s chemical weapons exploration and pursuit decisions come years after the analogous
decisions on nuclear weapons? Understanding the reasons why Libya — despite decades of
e↵ ort — made little progress towards actually acquiring nuclear weapons can help answer
this question.
Beginning in 1973, Qaddafi undertook a sweeping e↵ ort to transform — and largely
dismantle—Libyanstateinstitutionsinordertoconsolidatehisholdonpower. Asdiscussed
earlier in this chapter, this “cultural revolution” involved the dissolution of most existing
state bureaucracies and laws, the establishment of grassroots “popular committees” to re-
placethem,andthecreationofinformal“revolutionarycommittees”directlyunderQaddafi’s
control to surveil and sometimes replace the rest of the state apparatus. Braut-Hegghammer
(2016) documents how this hollowing out of the state dramatically weakened the regime’s
institutional capacity to e↵ ectively set goals for the nuclear program and monitor progress
84
toward those goals. These issues were compounded by the vast resources available for the
program due to Libya’s oil wealth. With limited oversight, and few financial constraints,
scientists were free to spend much of their time on basic research, rather than focusing on
the specific technical challenges of getting a weapons program up and running. Some high
level o cials leveraged lucrative contracts with foreign suppliers for personal profit — and
were therefore invested in the nuclear weapons program’s continuation, but not particularly
invested in reporting on its problems and (lack of) progress.
Regime security concerns created several other issues for the nuclear program. First,
having brutally repressed student protests, Qaddafi saw universities as a breeding ground
for opposition to the regime, and was therefore reluctant — despite wide-ranging and lav-
ish social spending programs — to invest in higher education (Anderson, 1982; Braut-
Hegghammer,2016). Thismeantfeweravailablescientistsandtechnicianswhowerequalified
to work on the nuclear program. Furthermore, in order to remain in power, Qaddafi needed
to balance between rival factions amongst the elite — those who supported Qaddafi’s ambi-
tious foreign policy, and those who preferred to prioritize domestic economic development.
When cash was pouring in, in the 1970s and early 80s, it was possible to please both factions
—tospendonguns and butter. But, declining oil revenues beginning in the mid-1980s be-
gan to force tradeo↵ s. The nuclear program only began to make real tangible progress in the
late 1990s.
10
Not only did Libya begin receiving materials form the A.Q. Khan network, but
also domestic security concerns had lessened: many dissidents had fled or gone underground
afteruprisingsinthemid-1990sweremetwithbrutalrepression, andQaddafihadsomewhat
resolved debates within the regime in favor of the more pragmatist faction (Takeyh, 2001).
The chemical weapons program was able to avoid some of these pitfalls — and was
ultimately more successful. It was much smaller in scale, and therefore easier to manage
in the absence of e↵ ective institutions. US and British inspection teams reported that
10
Although, according to assessments by inspectors after Libya agreed to dismantle the program,
less progress than many Libyan o cials and scientists had believed.
85
the program comprised “fewer than a dozen” chemists and chemical engineers, who had
“belonged to a relatively small, elite group,” (Tucker, 2009). Because so few scientists were
needed,itwaspossibletofindpeoplewiththenecessarytraining,andmanyofthoseinvolved
in the chemical weapons program had advanced degrees from foreign universities. While the
purchase of chemical precursors and specialized equipment from foreign suppliers — often
at inflated prices — was costly, the overall scale of expenditures was much less than was
required for the nuclear program. Even when oil revenues plunged in the mid 1980s, Qaddafi
was able to direct the necessary resources into the chemical weapons program, achieving a
chemical weapons capability by the end of the decade, while the nuclear scientists drifted,
struggling to e↵ ectively absorb enrichment technology from foreign suppliers.
Further insight comes from examining the Qaddafi regime’s apparent choice to retain
asmall,undeclaredchemicalweaponsstockpileevenafterthe2003agreementtofullydis-
mantle Libya’s WMD programs. With that agreement, the level of external security threats
Libya faced declined dramatically: the regime had managed to reach a deal with the Bush
administration, whose“GlobalWaronTerror”andinvasionofIraqoverallegationsofWMD
pursuit had been a primary source of external security concerns since 2001. The extradition
and trial of the Lockerbie suspects had led to the lifting of UN sanctions, and the normal-
ization of diplomatic relations with much of the world, while domestic economic reforms and
movement away from anti-imperialist nationalism were facilitating Libya’s integration into
world markets.
Indeed, at that point, the most damaging action the Libyan government could have
taken, from a national security perspective, would have been reneging on the disarmament
agreement. Even if Libya was concerned that the US and UK would fail to hold up their end
of the deal in terms of providing sanctions relief, it hardly seems that risking the discovery
of noncompliance would improve the outcome. Furthermore, the undeclared stockpile was
small enough to likely have been a poor deterrent to attack by a major power. However, the
stockpile could potentially have been an e↵ ective last line of defense against insurgents or
86
disloyal factions of the military. Thus, Qaddafi may have been concerned — rightly, given
the events of 2011 — about threats to the regime from domestic opponents, and chosen to
maintain a small “hedge” chemical weapons stockpile for security against those threats.
Conclusion
As the Qaddafi regime faced both significant risk of a coup d’´ etat and significant tensions
with both regional rivals and nuclear powers, the Libyan case illustrates how internal and
external threats can interact to shape WMD pursuit choices. Specifically, it demonstrates
how governance choices made to guard against threats from within the state — fragmenting
or dismantling state institutions — can make chemical weapons acquisition a more feasible
optionthannuclearweapons, eveniftheultimategoalistodeterexternaladversaries. Libya
retention of a small, undeclared CW stockpile after the 2003 agreement, also illustrates how
internal security threats — both coup risk and domestic unrest — might prompt rulers to
maintainchemicalweaponscapabilities, evenevenifdoingsomightriskincurringexistential
threats from external state actors.
With frequent coup attempts, nuclear and chemical weapons armed rivals, and neigh-
bors that employed chemical weapons in conflict, Libyan chemical weapons pursuit repre-
sents a “typical” case in the context of the findings presented in Chapter 4. Exploring the
mechanisms driving chemical weapons proliferation choices in this case adds nuance to the
substitution argument, illustrating how regime security threats can hamper nuclear pursuit,
makingchemicalweaponspursuitanattractivesecondchoice. Thisshouldpromptarethink-
ing of the term “poor man’s atomic bomb;” states that chose to pursue chemical weapons
instead of (or in addition to) nuclear weapons may not necessarily be “poor” in terms of
economic resources or development, but rather lacking in e↵ ective state capacity, or facing
negative economic shocks (relative to their own previous situation).
This case also suggests the need for further research in several areas. First, when
87
internal and and external security conditions are interrelated, as is often the case, how do
we determine which issues are driving WMD pursuit choices. For instance, Qaddafi turned
to an adventurist foreign policy after failing to fully consolidate support for his revolution at
home,sometimesfurtherprovokedtensionsasadiversionarytactictodistractfromunrestat
home. Second, the Qaddafi regime cultivated an extensive network of foreign suppliers that
contributed to both the nuclear and chemical weapons e↵ orts. What role does third party
support play in CBW acquisition more generally — is Libya’s reliance on foreign assistance
anomalous, or part of a typical pattern for chemical weapons pursuit?
88
Chapter 6
Rhodesia: Counterinsurgent CBW
This chapter examines Rhodesia’s chemical and biological weapons programs and the po-
tential motivations for CBW weapons pursuit. Rhodesia is a valuable case for analysis for
several reasons. First, due to the extremely high levels of internal security threat — par-
ticularly violent domestic unrest — faced by the Rhodesian regime, the case provides an
opportunity to examine the mechanisms linking those security threats to CBW pursuit de-
cisions. Second, as a rare case of both successful chemical weapons pursuit and successful
biological weapons pursuit, the Rhodesian case allows us to tease out di↵ erences in how the
internal security environment shapes decisions about the two di↵ erent types of weapons. Fi-
nally, given that the Rhodesian regime was ultimately removed from power and the chemical
and bioweapons programs ended, information on the programs is relatively accessible.
The case of Rhodesia highlights how domestic threats can drive a desperate regime
to pursue and use chemical and biological weapons, and how easily a workable chemical
weapons capability for counterinsurgency purposes can be acquired.
Therestofthechapterproceedsasfollows: Ibeginbybrieflyprovidingsomehistorical
background information, before more extensively discussing the security threats faced by the
Rhodesiangoverningregimeduringitsperiodofindependence. Next,Iprovideanoverviewof
RhodesianCBWprograms,beforediscussingbothchemicalweaponsactivitiesandbiological
89
weapons activities in detail. I then analyze how security threats — particularly the ongoing
insurgency — motivated and shaped CBW pursuit decisions, and conclude by linking the
evidence from Rhodesia back to the findings from the quantitative analysis presented in
Chapter 4.
Background
Rhodesia was an unrecognized state, governing the territory that is now Zimbabwe from
1965–1979. Prior to European colonization, Zimbabwe, with a wealth of natural resources,
wasaprosperousandimportanttradingregion. CecilRhodes’BritishSouthAfricaCompany
gained control of the region in the late 19th century, conquering Mashonaland in 1890,
and Matabeleland in 1893. In 1923, the Britain annexed “Southern Rhodesia,” the area
equivalent to modern Zimbabwe, and granted it the status of a self-governing colony. Under
the1923constitution, thelegislaturewasdominatedbywhiteEuropeans, andthe1930Land
Apportionment Act set aside large areas of the most desirable land for purchase only by the
white minority (Kennedy, 1987).
In 1953, Britain consolidated Southern Rhodesia, Northern Rhodesia (present day
Zambia), and Nyasaland (present day Malawi) into the Central African Federation, placing
economic and foreign policy under central control, but leaving other domestic issues to
the individual territories. In the face of African nationalist pressure for decolonization,
Britain dissolved the federation in 1963, and granted Zambia and Malawi independence
under majority rule (Hargreaves, 2014). This action set the stage for Rhodesia’s unilateral
declaration of independence (UDI) in 1965, and the outbreak of a 15-year conflict with
nationalist insurgents, which I discuss more extensively in the next section.
During the colonial period, the economy experienced rapid growth and development,
initially based on primary resource extraction, with the immigration of (largely British
working-class) skilled labor facilitating the emergence of a strong manufacturing sector af-
90
ter the Second World War. Commercial cattle farming was also an important pillar of the
economy (Barber, 1961). Even with high levels of European immigration, whites remained a
smallminority;by1960thepopulationwasstill93%black(Wills,1967). Thecountry’swhite
population was also fairly transient, with high levels of emigration as well as immigration
(Brownell, 2008).
The Security Environment
Throughoutthefourteenyearsofitsindependentexistence, theRhodesiangovernmentfaced
significantandworseningthreattotheirrulefromdomesticinsurgents. Theinternalsecurity
situationwasdeeplyintertwinedwithexternalpressures. Beginningin1960,theautonomous
British colony of Rhodesia experience a sustained period of unrest and tensions between the
white minority government and increasingly militant black nationalist activists (Raeburn,
1978). Despite a wave of decolonization in the early 1960s, in which many former British
colonies gained independence, Britain refused to grant Rhodesia independence because the
government had not specified a mechanism for transition to majority rule. As negotiations
dragged on without a resolution, Rhodesian Prime Minister Ian Smith issued a unilateral
declaration of independence (UDI) on November 11, 1965 (Martinez, 2002).
Britain immediately declared the UDI illegal, and brought the issue before the UN
Security Council, which labeled Rhodesia an “illegal racist minority regime” and recom-
mended sanctions on imports of petroleum and military hardware. In December 1966, the
UN clarified that the sanctions were mandatory, and further banned the purchase of Rhode-
sian exports by member states. However, the e↵ ect of the sanctions was limited over the
next decade as both the Portuguese colonial government in Mozambique and the white mi-
nority government in South Africa continued to allow trade with Rhodesia (Moorcraft and
McLaughlin, 2010).
1
Nevertheless, Rhodesia was unable to gain international diplomatic
1
UN non-members such as Switzerland and West Germany also continued trade, and some UN
91
recognition.
Conflict with black nationalists intensified after the UDI, and the Rhodesian govern-
ment spent the remainder of its existence fighting the Rhodesian Bush War, also known as
the Zimbabwe War of Liberation or the Second Chimurenga. The conflict was primarily
between the unrecognized Rhodesian government, and two nationalist organizations com-
mitted to liberating Rhodesia from minority white rule: the Zimbabwe African National
Union (ZANU) and the Zimbabwe African Peoples Union (ZAPU). Both received external
support, ZANU from the Chinese government, and ZAPU from the Soviets, and each had
a military wing, ZANLA
2
and ZIPRA,
3
respectively. These guerrilla forces operated from
within Rhodesia, as well as from camps in neighboring Zambia and (later in the conflict)
Mozambique (Gould, 2005).
The first phase of the conflict, from the UDI to 1968, consisted primarily of small-scale
guerrilla incursions into Rhodesia from bases within newly independent Zambia. The na-
tionalists’ initial strategy was to use terrorist tactics to incite fear in the white population
and spur the black population into a nationalist uprising, forcing the British to intervene,
retake control of the colony, and grant independence under majority rule. However, these
small operations were easily crushed by the British South Africa Police (BSAP), no nation-
alist uprising materialized, and by 1966 it became clear that the British would not try to
regaincontrolbyforce(Cilliers,2015). Thisrealizationpromptedtheguerrillastoundertake
more regular incursions into Rhodesian territory, with the aim of retaking the territory by
force. The guerrillas began recruiting more aggressively from Zimbabwean refugee popula-
tions living in Zambia and Tanzania, and in 1967 ZIPRA entered into an alliance with the
African National Congress (ANC, a South African nationalist organization operating from
exile) (Martin, Johnson and Mugabe, 1981). While ZIPRA did briefly manage to establish
member states, such as Japan and Iran, violated the sanctions.
2
Zimbabwe African National Liberation Army
3
Zimbabwe People’s Revolutionary Army
92
astringofbasesinsideRhodesia,theguerrillaswerelargelyoutmatchedbytheRhodesian
forces. By 1969, most guerrillas operating inside Rhodesia had been killed or captured, and
the black population appeared to mostly still support the Rhodesian government (Cross,
2017).
There was a lull in the fighting from 1969-1972. During this time the Rhodesian
government’s political position was relatively strong. Sanctions had failed to produce severe
economic consequences, and diplomatic isolation had apparently not prompted a desire for
compromise. In1971,Britainevenbegantonegotiateacompromiseindependenceagreement
thatwouldhaveallowedfortheindefinitecontinuationofminorityrule,althoughthedealwas
scrapped when it became apparent that it would be considered unacceptable by Rhodesia’s
black African population (Ryan, 2004). Rhodesia also joined Alcora, a secret defensive
alliance established between Portugal and South Africa, both of which were facing African
nationalist insurgencies (Miller, 2016).
However, the security situation deteriorated rapidly as the Portuguese colonial gov-
ernment began to lose control of Mozambique. In 1972 and 1973, guerrillas began to ramp
up operations from rebel-controlled areas of Mozambique. ZANLA’s work building up both
armscachesandanextensivenetworkofsupportersandinformantsinnortheasternRhodesia
also began to pay o↵ ,andinsurgentmovementswithinRhodesiaintensified(Cross,2017).
With Mozambique’s independence in June 1975, Rhodesia was exposed to an additional
1,300 km of hostile border from which guerrillas could operate (Rogers, 2001). Recognizing
that preservation of white minority rule in Rhodesia seemed increasingly unlikely, South
Africa cut o↵ most overt military support in a bid to appease the international community
and buy time for their own minority regime (Moorcraft and McLaughlin, 2010).
The years from 1976 until the war’s conclusion in 1979 saw the most intense fighting.
The Rhodesian government faced an extensive insurgency, was almost entirely surrounded
by hostile states, and was being pushed to compromise by its only ally. Insurgents were
operatingdeepwithinRhodesia,andZANLAoperationshadsucceededinclearingwhitesout
93
of vast areas of the country along the border with Mozambique. The Rhodesian leadership
recognized the futility of hoping for indefinite minority rule, but kept fighting in hopes of
attaining the most favorable possible compromise settlement. By 1976 “no one would say so
inpublic—exceptfortheverypublicactofemigrating—butinprivatemanywereprepared
to admit that even if defeat was unlikely, victory was impossible,” (Godwin, Hancock et al.,
1993).
While the response to the insurgency had initially been led by the BSAP, it now
required extensive regular military involvement. Rhodesian forces were better trained and
equipped than the insurgents, but vastly outnumbered. Not only did the black African
population outnumber the Europeans by more than ten to one, only a fraction of those
eligible for conscription answered the call, and many emigrated to South Africa instead. In
direct skirmishes, Rhodesian forces would win, but they were losing their ability to control
wide swaths of the country. In this context, Rhodesia began to rely heavily on “pseudo-
operations”—psychologicalwarfare,assassination,andothercovertoperations(Bale,2006).
Such operations were carried out Selous Scouts, created in 1973 under the command of
the Central Intelligence Organisation (CIO) Special Branch. The Selous Scouts consisted
primarily of co-opted insurgents who would infiltrate local communities in search of real
insurgents (Martinez, 2002; Parker, 2006).
On March 3, 1978, the Rhodesian government reached an Internal Settlement agree-
ment with the moderate African nationalist parties which were not involved in the fighting.
Under this agreement, elections were held in April 1979, and Abel Muzorewa became the
first black prime minister of the newly renamed Zimbabwe-Rhodesia. However, control of
the police, security forces, judiciary, and civil services remained with the white minority.
Fighting continued, as ZANU and ZAPU refused to participate in a government that did
not constitute true majority rule. Sanctions remained in place, and the majority of the in-
ternational community declined to recognize the new government because it did not include
the two main nationalist parties. A ceasefire was finally reached with the negotiations at
94
Lancaster House in late 1979, facilitated by the British government. Under the Lancaster
House Agreement, Rhodesia briefly reverted to British control, while elections were held un-
derinternationalsupervision. ZANU,ledbyRobertMugabewontheFebruary1980election,
and Zimbabwe became independent on April 18 (Gregory, 1980).
RhodesianCBWCapabilitiesandPrograms: anoverview
During its brief independent existence, the Rhodesian government pursued both chemical
and biological weapons. Consideration of an indigenous chemical and biological weapons
program likely began in 1975 (Horowitz and Narang, 2014), with focused pursuit beginning
in 1976. By 1977, Rhodesia possessed chemical weapons and was employing them in an
ongoing fight against insurgents.
Identifying the dates of Rhodesian biological weapons possession is somewhat more
complicated. There are accounts of Rhodesian forces deploying cholera against insurgents as
early as 1973, and again in 1976. However, the source of the pathogen is unclear, and there
is no available evidence suggesting it was the result of an indigenous bioweapons e↵ ort. It is
likelythateitherculturesorthefinishedagentwereprovidedbyathirdparty.
4
Additionally,
Rhodesian attempts to use cholera were not particularly successful, raising questions about
whether the regime had actual acquired a working bioweapons capability. Rhodesian forces
may also have successfully deployed anthrax in late 1978 or early 1979.
The Rhodesian chemical and biological weapons programs began in late 1975 or early
1976, at the suggestion of Robert Symington, a professor of anatomy at the University of
Rhodesia Medical School. Symington prepared a proposal identifying chemical and biolog-
ical agents that could be used against the guerrillas, and presented it to then Minister of
4
The most likely suppliers are South Africa, which provided a variety of aid to the Rhodesian
government, or Portugal, which Rhodesian forces were aiding against nationalist guerrillas in its
colony, Mozambique.
95
Defense P.K. van der Byl, who approved of the idea and advocated for it to Prime Minister
Ian Smith. Smith authorized the plan and delegated responsibility for implementation to
Rhodesia’s Central Intelligence Organisation (CIO). In contrast to more traditional chemi-
cal weapons programs, the proposal was mostly focused on identifying poisons that could be
easily developed from commercially available agricultural and industrial chemicals (Cross,
2017, p. 249). Similarly, whiletheproposaldocumentsmentionafewbiologicaltoxins, there
is no discussion of more traditional bioweapons agents.
Theprimarychemicalagentsusedwerethreecommerciallyavailablepoisons: parathion,
thallium, and warfarin. Parathion is an organophosphate insecticide, easily absorbed by the
skin, butalsohighlytoxicwheninhaledoringested. Itisaneurotoxinwhichtypicallycauses
death through respiratory failure (with a secondary cardiac component), and symptoms ap-
pear within minutes or hours, depending on the dose and method of exposure (Roberts and
Reigart, 2013). Thallium is a heavy metal typically used as a rat poison, which is tasteless,
odorless, and toxic when ingested. Initial gastrointestinal symptoms, which appear within
12-48 hours, could easily be attributed to endemic diseases such as malaria. Neurological
symptoms, including paralysis and respiratory failure, begin after approximately five days
(Moeschlin,1980). Warfarinisananticoagulantusedasarodenticide,whichalsohasmedical
applications—insmalldoses—totreatsomeheartconditions. Severehemorrhagingbegins
within days to a few weeks after ingestion (Rettie and Tai, 2006). These chemical agents
were delivered by manufacturing poisoned supplies (mostly food, beverages, and clothing),
anddistributingthemtoinsurgentsthroughcovertoperations(GouldandFolb,2000;Cross,
2017, p. 106-110).
Work with biological weapons agents included Vibro cholerae and Bacillus anthracis,
the causative agents of cholera and anthrax, respectively, as well as Clostridium botulinum,
the bacteria that produces botulinum toxin. V. cholerae, typically water-borne, causes se-
vere illness in roughly 1 in 20 cases of infection. Symptoms includes severe diarrhea, which
can lead to dehydration and death (Deen, Mengel and Clemens, 2020). B. anthracis occurs
96
naturally in the soil of some regions, and infects livestock animals, such as cattle. Human
infection typically occurs as a result of skin contact with infected animals or consumption of
infected meat products, although inhalation of anthrax spores can also cause disease. Symp-
tomsdependonthemethodofinfection; themostcommon, skincontact, produceslesionson
the skin, while ingestion and inhalation produce gastrointestinal and respiratory symptoms
(Inglesby et al., 2002). Botulinum toxin, one of the most poisonous known substances, kills
by inducing respiratory paralysis (Horowitz, 2005). V. cholerae, C. botulinum, and possi-
bly B. anthracis were likely cultivated only in small batches, given the available facilities.
The preferred delivery method for cholera an botulinum toxin was introduction into water
sources, and Rhodesian forces may have experimented with dissemination of anthrax via
aircraft.
The small team working on the program mostly consisted of medical or veterinary stu-
dents at the University of Rhodesia, selected by Symington. Three main sites were involved.
Most of the development work was conducted at a laboratory at Symington’s residence,
funded by the BSAP Special Branch (Sti↵ ,2002). Poisoneditemsweremanufacturedand
distributed at Fort Bindura, 89 km north of the capital, Salisbury (now Harare). A more
isolated facility at Mount Darwin, 19 km north of Bindura, was used for clinical trials on
human subjects and experimentation with biological agents.
Rhodesia’s CBW programs appear to have received substantial assistance from the
white minority regime South Africa, and the Rhodesian experience seems to have shaped
South Africa’s subsequent CBW e↵ orts. Funding for Rhodesian special operations was fun-
neled through South Africa,
5
and many of the raw materials for the chemical weapons pro-
gram (including both chemicals and goods to contaminate) were purchased through South
Africa (due to international sanctions against Rhodesia). Additionally, Symington had
personal connections with several South African scientists, including Jan Coetzee, Lothar
5
The original sources of the funds is unknown, although significant amounts of funding may
have originated from Saudi Arabia.
97
Neethling, and Wouter Basson (who would later head South Africa’ s CBW e↵ ort, “Project
Coast”), who may have provided agents for testing, or advice on the choice of agents (Cross,
2017, p. 156). Samples of C. botulinum and B. anthracis, and a range of toxins and chemical
poisons likely also originated from South Africa.
Chemical and biological weapons activities ceased in late 1979, with the end of the
war and the (brief) return of the colony to British control before Zimbabwean independence.
Materials and records from the Bindura and Mount Darwin facilities were transferred to
South Africa by early 1980, and many of the participants in the program later emigrated to
South Africa as well (Cross, 2017, p. 135).
Chemical Weapons Activities
By 1976, Symington’s team began conducting experiments with chemical agents, primarily
at Mount Darwin. Trials were conducted using human subjects — insurgents detained by
the Selous Scouts — to determine the lethal dosages of di↵ erent agents, and the e↵ ects
of application to di↵ erent body parts. The test subjects’ bodies were disposed of down
mine shafts, and many were discovered after the war (Martinez, 2002). At least one report
detailingfindingsfromtheseexperimentswasprovidedtoSouthAfricanJanCoetzeein1979
(Cross, 2017). However, there was also a lot of reliance on trial and error “field testing” to
determine the most e↵ ective agents, dosages, and dissemination strategies.
Beginningin1977, preparationanddisseminationofchemicalagentstookplacemostly
at Fort Bindura, an outpost previously used by the Selous Scouts, but which was no longer
afocalpointofoperationsbythatpointinthewar. Thechemicalweaponsactivitywas
intentionally located away from larger, more active Selous Scout facilities to avoid attracting
attention (Cross, 2017; Gould and Folb, 2000). Chemical agents were delivered primarily by
means of contaminating supplies, such as clothing, food, beverages, and medicine, and then
disseminating them to guerrilla forces.
98
The production of contaminated clothing began in April, and contamination of food,
beverages, and medicines in May or June (Cross, 2017, p. 74). Methods were fairly rudi-
mentary, and the focus was on processing commercially available chemicals into forms that
were more readily concealed and more quickly absorbed into the body. Liquid parathion was
delivered to the fort in large drums. Several approaches were used to eliminate the chemical
odor that could raise suspicion about the contaminated goods. The liquid was either poured
onto tin sheets and dried in the sun to produce a powder which was then brushed onto
clothing, or the parathion was dissolved in methanol and clothing dipped in the resulting
solution before being sun dried (Gould, 2005, p. 37-38). Dimethyl sulfoxide (DMSO), a
solvent which is highly e↵ ective at penetrating skin, was often added to aid absorption of
the poisons — a technique that South Africa’s “Project Coast” would later adopt. Thallium
was injected into canned foods and beverages using micro-needles. Warfarin was mixed into
bulk foods such as corn meal (Gould and Folb, 2000; Cross, 2017, p. 106-110). The team
may also have experimented with contamination of bottled water with cyanide, arsenic, and
other poisons, and powdered organophosphates such a parathion may also have been mixed
into food and beverage, and brushed onto medicines such as aspirin (Gould, 2005, p. 36).
Saftey measures used by those processing the chemical agents were not well docu-
mented, but appear to have been limited, despite the high risks posed by inhalation of the
finelypowderedpoisons. Symingtonwashospitalizedwithparathionpoisoningin1979,lead-
ing the team to stop using parathion for contamination, and to adopt protective measures
such as the use of surgical masks, gloves, and rubber aprons. The focus shifted to telodrin,
an organochloride insecticide, which was used to contaminate cigarettes (Cross, 2017, p.
108-109).
Contaminated goods were delivered by “contact men,” either co-opted insurgents or
operatives disguised as insurgents. Dissemination operations were typically managed by the
Projects Section of the British South Africa Police (BSAP) Special Branch, although Selous
Scoutsweresometimesinvolved(Gould,2005). Supplieswerealsodeliveredtoco-optedstore
99
owners, with the expectation that guerrillas would raid and steal the contaminated supplies
(Cross, 2017, p. 114-115). M.J. McGuinness, the Special Branch O cer who oversaw the
chemical weapons work at Fort Bindura, attempted to maintain detailed documentation
of poisoned items distributed and the resulting fatalities (Gould, 2005, p. 37). Operatives
requiredpriorapprovaltobeissuedwithcontaminateditems, and“everycontaminateditem
that left the Bindura Fort, the only location at which they were being manufactured, had
to be signed for and subsequently accounted for by the recipient,” (Gould, 2005, quoting
McGuinness). “Death bonuses” were paid to contact men, based on the estimated number
of insurgent fatalities (Cross, 2017, p. 100, 114).
Chemicalagentswerereportedlyalsousedtocontaminatewellsandotherwatersources
relied on by guerrillas, particularly in the Eastern Highlands and the areas immediately
across the border with Mozambique (Purkitt and Burgess, 2002). One incident, in which
Selous Scouts contaminated a well in guerrilla controlled territory near Mozambique with
an unknown chemical agent, reportedly resulted in over 200 civilian deaths, as the rebels
main water source was also the main water source for the local villagers (Martinez, 2002).
Accidental civilian deaths also often resulted from the distribution of contaminated supplies.
In some cases, insurgents unknowingly distributed contaminated food to villagers, or goods
that had been delivered to local shopkeepers made their way into civilians hands; o cial
documents record at least 900 such deaths (Gould, 2005, p. 42). In some cases, insurgents,
suspecting poisoning but unable to trace the source of the poisoned goods, retaliated by
killing locals (Alexander and McGregor, 2005).
Biological Weapons Activities
ConclusiveinformationonthedevelopmentanduseofbiologicalweaponsagentsinRhodesia
is more limited than information pertaining to chemical weapons. The Rhodesian military
appears to have had access to V. cholerae, the causative agent of cholera, as early as 1973,
100
although how they acquired the agent is unclear. Former members of the Rhodesian Light
Infantry (RLI) described using cholera to contaminate wells and other water sources in
areas of Mozambique near the Rhodesian border where guerrillas were operating.The RLI
maintained maps of contaminated water sources, and units were warned against consuming
water found in those areas (Cross, 2017, p. 111 - 113). The contamination was intended to
force guerrillas to choose routes into Rhodesia that would funnel them into “killing zones,”
where they could easily be picked o↵ by the smaller Rhodesian forces. Those choosing to
travel through the contaminated zones would have to carry their own water supplies, which
would trade of with other supplies, such as ammunition, leaving the guerrillas ill-prepared
to conduct operations once they reached Rhodesia (Cross, 2017). The most extensively
documented of these e↵ orts appears to have resulted in a large cholera epidemic in two
provinces of Mozambique near the Rhodesian border in 1973, which was investigated by
the UN. Over 1,000 FRELIMO (Liberation Front of Mozambique) insurgents died in the
outbreak, and doctors described the outbreak as unusual both in how rapidly it spread and
how suddenly it stopped (United Nations, 1974). It is also possible that the wells had been
poisoned with a chemical agent that produced similar symptoms to those of cholera.
While there was a general policy against the domestic dissemination of cholera, there
are several accounts of the Selous Scouts releasing cholera into portions of the Ruenya river
(whichflowsnortheastacrosstheborderwithMozambique)locatedwithinRhodesiain1976.
McGuinness described multiple attempts to contaminate a sheltered area of the river where
insurgents frequently bathed, which failed to result in any insurgent deaths (likely because
cholera does not survive well in running water) (Cross, 2017, p. 112). Ellert (Ellert, 1989)
describes the introduction of cholera at several points near the border, which he claimed
coincided with a cholera outbreak among nearby villagers. Ultimately, the Selous Scouts did
not pursue this approach further, as it was judged that the pathogen dissipated too quickly
to provide tactical advantage, while posing a risk of uncontrolled domestic spread (Martinez,
2002).
101
Rhodesian production of and experimentation with other biological weapons agents
likely began in early 1979 when Symington’s assistant Victor Noble returned from a visit
to South Africa with samples of C. botulinum and B. anthracis (Cross, 2017, p. 111).
6
These activities appear to have taken place primarily at Mount Darwin. Records show an
attempt to produce botulinum toxin by heating large tanks filled with a mixture of water,
corn, rotten meat and C. botulinum (). Although this approach may have been ine↵ ective —
giventhatC. botulinum needsanaerobicconditionstogrowandproducesu cientquantities
ofbotulinumtoxin—Rhodesianforcesreportedlyusedthetoxintopoisonwatersourcesnear
guerrillacampsinMozambique(Cross,2017, p. 111, 113). Theteamalsoexperimentedwith
usinganthraxsporestocontaminatecigarettesandfood—experimentsthatwerereportedly
unsuccessful because the anthrax strain brought back from South Africa was only infectious
to animals (Cross, 2017, p. 98, 201, 214). Ricin, a toxin produced from castor beans, was
also discussed as a potential tool in a — somewhat far-fetched — scheme to assassinate
ZANU leader Robert Mugabe (Cross, 2017, p. 36, 84).
Cross (2017, p. 111) claims that “no evidence exists to suggest that the Rhodesians
everattemptedtoisolate,cultureorcultivateB.anthracis.” However,allegationspersistthat
Rhodesian forces used anthrax near the end of the war, and it appears that at a minimum,
the option was considered by various decision-makers. This suggests that decision-makers
believed the capability to deploy anthrax either already existed or could be easily developed.
Allegations of Rhodesian use of anthrax for biological warfare center around the an-
thrax outbreak which began in November 1978 and continued relatively unchecked until
1984, becoming one of largest documented anthrax outbreaks in history (Wilson et al., 2016;
Martinez, 2002).
7
Between 1978 and 1980, in addition to thousands of infections in cattle,
6
Accounts of the sources of these samples di↵ er, and Noble later claimed that he and Symington
stole the anthrax sample from a laboratory at the University of Rhodesia.
7
The other possible contender for largest outbreak occurred in 1770 in Saint Domingue, during
the Haitian revolution, and may have killed up to 15,000 over six weeks. However, the source of
the illness was not immediately identified as anthrax at the time. See, e.g., Morens (2003); Wilson
102
there were 10,783 recorded human infections, resulting in 182 deaths — all black Zimbab-
weans. Rhodesian forces’ documented interest in applications of anthrax as a bioweapon,
the outbreaks’ timing at period of the war’s peak intensity, and the unusual scale and pat-
terns of the epidemic combine to suggest that it may have been deliberately incited, rather
than naturally occurring (Nass, 1992; Martinez, 2002; Wilson et al., 2016). While anthrax
was endemic to the region, it was rare in Rhodesia prior to 1978, with an average of only
six human cases per year between 1926 and 1978 (Wilson et al., 2016). Not only did the
1978 outbreak result in unprecedented infection and fatality numbers, it occurred at an un-
precedented geographic scale, rapidly spreading to five of Rhodesia’s six provinces, including
areas that had seen no previous anthrax infections (Cross, 2017, p. 179). Furthermore — in
contrast to most anthrax outbreaks, which are highly focalized — the outbreak had multiple
geographic centroids, appearing to “hop” between regions (Wilson et al., 2016). Despite this
extensive level of geographic spread within Rhodesia, no neighboring country experienced
higher than usual case numbers. Furthermore, only 11 human infections (and only 4 deaths
ofcattle)occurredin Europeanfarmingareas, andtherewerenodeathsofwhiteRhodesians
(Nass, 1992).
It is, however, di cult to conclusively prove that the outbreak was not naturally
occurring. Numerous studies have analyzed the epizootic and epidemic, coming to di↵ erent
conclusions about the likelihood of deliberate involvement by Rhodesian forces, and o↵ ering
explanations for how many of the unusual features of the outbreak could be consistent with
natural occurrence (Carus, 2001).
8
First, the ongoing conflict had led to a breakdown of
veterinaryservicesintheruralTribalTrustLandswheremostofthediseasespreadoccurred.
In areas under rebel control, regular cattle vaccination and “cattle dipping” (immersion
of cattle in chemical solutions to kill parasites on the skin) ceased, and the incidents of
et al. (2016).
8
For examples of studies that suggest human involvement, see Nass (1992) and Wilson et al.
(2016). For examples of studies suggesting natural occurrence, see Cross (2017, Chapter 4) and
Velsko (2015). For a general analysis of the epidemic, see Davies (1982, 1983, 1985).
103
other cattle diseases increased as well (Burgess and Purkitt, 2001).
9
European commercial
cattle ranchers, on the other hand, were not located in areas of heavy conflict and rebel
control, and therefore still had access to veterinary care, and almost all commercial farmers
vaccinated their cattle during the epidemic (Cross, 2017, p. 190). The ongoing conflict
also limited movement across borders, as Rhodesian forces patrolled the border areas in the
hope of stopping guerrilla incursions, which could explain the lack of spread to neighboring
countries. Additionally, the disease’s puzzling “hops” between regions could be explained by
patients travelling to access medical treatment, or by consumption or handling of infected
meat from areas where the disease was already spreading. Both of these behaviors could
have been prompted by the conflict. As services broke down in contested or rebel-controlled
areas, it became necessary to travel for medical care (Cross, 2017, p. 184-185). A significant
black market in food, including contaminated meat, also emerged at this point in the war,
in response to “Operation Turkey” an e↵ ort by Rhodesian security forces to control food
supplies limit food surpluses (thus limiting the population’s ability to provide provisions to
the insurgents) (Cross, 2017, p. 195).
Some statements by former Rhodesian o cers corroborate the allegations that the
anthrax outbreak was the result of bioweapons use. For instance, Mangold and Goldberg
(2001) quote an o cer as saying:
It is true that anthrax spoor was used in an experimental role in the Gutu,
Chilimanzi, Masvingo and Mberengwa areas, and the anthrax idea came from
PSYOPS. The use of anthrax spoor to kill o↵ the cattle of tribesmen... was
carried out in conjunction with the psychological suggestion to the tribespeople
thattheircattleweresickanddyingbecauseofdiseaseintroducedintoZimbabwe
from Mozambique by the infiltrating guerrillas. (p. 222)
In a similar vein, Jim Parker, a former Special Branch o cer, recounts a 1979 conversation:
9
The actual impact of reduced access to veterinary services on the anthrax outbreak is unclear.
Prior to the outbreak, anthrax vaccination was not common in Rhodesia, due to the low incidence
of the disease (Martinez, 2002). However, once the outbreak began, vaccination could have been
an e↵ ective intervention to slow spread (Cross, 2017, p. 190).
104
The late Doctor Sandy Kirk was based at the fort. He told me that Selous
Scouts teams had recently deployed anthrax spores to infect the cattle of black
tribesmen in the Nkayi [Nkai] and Lupane areas of Matabeleland... The purpose
of the exercise, he said, was to limit the availability of food that troops could
forageifalargeZIPRAforceinvadedthecountry. Outbreaksofanthraxincattle
intheLupaneandPlumtreeareashadoccurredinthepast, soitwouldn’tappear
unusual if similar outbreaks happened in those adjoining areas. (Parker, 2006,
p. 171-172).
However, such statements cannot be taken as conclusive evidence tying the 1978-80 anthrax
outbreaktoRhodesianforces,giventhevagueness,particularlywithrespecttodissemination
methods, conflicting details, and second-hand nature of the accounts, as well as the desire of
o cers potentially involved to remain anonymous. These accounts suggest that if anthrax
was deployed domestically by Rhodesian security forces, the intention was not necessarily to
produce a large-scale human epidemic. Rather, the anthrax may have been deployed as part
of an experimental field test, or with the goal of limit food supplies (in line with Operation
Turkey), lowering morale, and damaging relations between insurgents and local populations.
This is consistent with assessments that the scale of the operation necessary to contaminate
the large area spanned by the outbreak would probably have exceeded Rhodesia’s capacity
(Cross, 2017, p. 198-199). In particular, it does not appear that there was an available
laboratory or facility that could have produced su cient quantities of anthrax spore, and
there is no available evidence of Rhodesian research into dissemination methods appropriate
to such an operation.
Anthrax was used as a biowarfare agent at least once during the conflict, but was
deposited outside of Rhodesian borders. In a 2002 interview, McGuinness described the
Special Air Service (SAS) dropping anthrax from an aircraft in an area of Botswana where
guerrillas were operating, just across the border near Plumtree (Gould, 2005; Cross, 2017,
p. 200, 209). The mission was reportedly assigned to the SAS after both McGuinness and
Selous Scouts commander Ron Reid-Daly rejected the idea as too dangerous. The potential
useofanthraxwasraisedanddebatedattheNationalJointOperationsCenter, butideasfor
use inside Rhodesian borders were rejected due to concerns about the possibility of spread
105
to white-owned commercial cattle farming, upon which the Rhodesian economy was heavily
dependent (Gould, 2005; Cross, 2017, p. 200).
Analysis
Chemical and biological weapons pursuit (and use) decisions in Rhodesia were clearly moti-
vated by a deteriorating domestic security situation that posed an existential threat to the
white minority governing regime. While Rhodesia did face significant external pressure and
diplomatic isolation, these pressures were directly tied to the domestic situation, and the
goal of external actors was to motivate a change in the particular governing regime, rather
than to attack or oppose the vital interests of the state as a whole.
The first reported deployment of cholera by Rhodesian forces, in 1973, occurred at
a turning point in the conflict, when Portuguese loss of control in Mozambique would sig-
nificantly empower the insurgents fighting in Rhodesia. The decision to initiate indigenous
chemical and biological weapons programs in late 1975 or early 1976 came as the insur-
gency was significantly increasing in intensity, and the Rhodesian government was realizing
that they were unlikely to achieve victory through conventional means. Expansion of the
biological weapons program — and the alleged use of anthrax — in late 1978–early 1979
accompanied a further intensification of fighting, as the government accepted that their In-
ternal Settlement Agreement would not be accepted as a compromise, and they therefore
needed to improve their bargaining position for further peace negotiations.
The Rhodesian case illustrates the potential attractiveness of chemical and biological
weapons for counterinsurgency operations. Rhodesian forces were able to use these weapons
in ways that helped to overcome the numerical asymmetry between the government and
insurgent forces — for instance, by poisoning water sources in order to funnel guerrilla’s into
designated “kill zones.” In many of the months where there is available data on casualties
resulting from the chemical weapons program, those casualty rates are higher than the casu-
106
alty rates from conventional operations (Cross, 2017). Beyond simply generating casualties,
the chemical and biological weapons e↵ orts appeared to be successful at lowering morale, in-
stilling fear, and disrupting relationships between guerrillas and the local population. While
the benefits of CBW were not enough for the Rhodesian government to secure victory, they
may have helped prolong the regime’s survival for several years.
While Rhodesian biological weapons activities illustrate the plausibility of bioweapons
pursuit for internal security purposes — despite their often greater levels of destructiveness
relative to chemical weapons — it also provides insight into how internal security decision-
making surrounding bioweapons can di↵ er from that for chemical weapons. Decision makers
were clearly concerned about the risks of domestic bioweapons use: the use of cholera was
supposed to be restricted to areas outside of Rhodesia’s borders, and the National Joint
Operations Center rejected suggestions for domestic anthrax use, citing concerns about
spread to the commercial cattle ranches. This highlights the relative di culty of acquir-
ing a bioweapons capability that strikes the right balance between being destructive and
being controllable. Not only were there concerns about the ability to control pathogens’
spread within Rhodesia, some decision-makers had concerns about their own forces ability
to handle biological agents safely — as demonstrated by McGuinness and Reid-Daly’s re-
jection of the Plumtree anthrax operation. Furthermore, many of the confirmed attempts
at biological weapons use (such as the release of cholera into the Ruenya river), were not
particularly successful. Even if the 1978-80 anthrax outbreak was the result of deliberate
action by Rhodesian forces, the chemical poisoning program produced far more cumulative
fatalities than Rhodesian bioweapons use.
On the other hand, this case demonstrates the relative ease with which chemical
weapons can be acquired — particularly if the goal is counterinsurgency rather than strate-
gic deterrence or large-scale operations in international conflict. A small team of amateurs
working with makeshift equipment and commercial materials were able to produce a rela-
tively e↵ ective chemical weapons capability in a matter of months. After an initial push to
107
build up a stockpile of contaminated supplies, most of those involved in the program only
worked on it intermittently (Cross, 2017, p. 105).
The small scale of the program allowed it to remain concealed even from some within
the Rhodesian leadership. While the Prime Minister and Defense Minister authorized the
program,SelousScoutscommanderRonReid-Dalybelieveditwasunlikelythatthechemical
weapons program was ever discussed at National Joint Operational Command (Gould, 2005,
p. 40). Police Commissioner Peter Allum — known for trying to limit the worst harms
to civilians — only learned of the program in early 1978 through a Ministry of Health
investigation into unexplained deaths among black civilians. Allum suspected poisoning,
and immediately ordered the chemical weapons program stopped, but McGuinness and CIO
director Ken Flower ignored him, and the program continued without his knowledge (Gould,
2005; Ellert, 1989; Cross, 2017, p. 134-135).
Conclusion
As a state experiencing high levels of violent domestic unrest and initiating chemical and
biological weapons pursuit in response to increases in the intensity of that unrest, Rhode-
sia illustrates one of the primary mechanisms linking regime security threats to chemical
and biological weapons pursuit. Given the findings in Chapter 4, Rhodesia represents a
“typical” or “on the regression line” case for chemical weapons pursuit: increases in vio-
lent unrest prompt the regime to explore and pursue chemical weapons. While biological
weapons pursuit is less clearly associated with domestic unrest in the large-n analysis, this
case demonstrates that unrest can motivate bioweapons pursuit, but that — in line with
theoretical expectations — the greater potential destructiveness and increased di culty of
acquiringreliablebioweaponscapabilitiescomplicatebioweaponsdecision-makingininternal
security contexts.
It can be di cult to disentangle the e↵ ects of internal and external security threats on
108
proliferation decision-making, given that states frequently experience both types of threats
simultaneously. However, the Rhodesian case provides an example of a program that was
clearly designed and initiated with the goal of countering internal security threats. The
initial proposal discussed the counterinsurgency utility of the potential agents it identified,
and, where those involved in the program have been willing to share their recollections, they
have been clear about the intended counterinsurgency goals of Rhodesian CBW operations.
ThiscasealsohighlightstheneedforfurtherresearchonsomedimensionsofCBWpro-
liferation,particularlytheroleplayedbyexternalassistance,andthepotentialfor“di↵ usion”
of proliferation between allies. Not only did South Africa provide materials and scientific
advice to the Rhodesian CBW e↵ ort, South Africa’s own later CBW programs were shaped
by lessons learned from the Rhodesian experience, and the influx of Rhodesian emigrants,
documents, and materials into South Africa at the time of Zimbabwean independence.
109
Chapter 7
Conclusion: CBW Research Moving
Forward
Unconventional weapons proliferation decisions are often complex and multi-causal, with
a range of security, political, technical, economic, and normative factors shaping choices
over weapons proliferation options. This dissertation identifies a set of pathway through
which security threats influence chemical and biological weapons proliferation. In doing so,
itmovesbeyondanalyticalframeworksthatmerelyaskhowchemicalandbiologicalweapons
pursuit are related to nuclear weapons pursuit. Rather, it identifies a key factor — internal
security threats faced by a state’s governing regime — that can both motivate chemical or
biological weapons pursuit in states with little or no interest in nuclear weapons pursuit, and
incentivise substitution or diversification of weapons programs in states that are interested
in nuclear weapons. Such internal security threats are not widely incorporated into theories
of arming, and highlight the unique role that chemical and biological weapons in can play
in regimes’ security strategies.
Aquantitativeeventhistoryanalysis,basedonnewpaneldataonCBWpursuit,acqui-
sition, and use, suggests that internal security threats, while often disregarded in theories of
arming,playasignificantroleinmotivatingstatestopursuechemicalandbiologicalweapons
110
proliferation. In particular, governing regimes facing increases in the risk of a coup may be
more likely to initiate chemical and biological weapons programs, and regimes experiencing
violent domestic unrest may be more likely to pursue chemical weapons.
Additionally, this analysis o↵ ers tentative evidence for two pathways through which
external security threats can motivate CBW proliferation. The first is “reactive prolifera-
tion”: states to pursue their own CBW capabilities in response to a rival’s acquisition of
strategic weapons (in particular nuclear, and sometimes biological). The second is “sub-
stitution”: in the face of negative shocks to industrial and economic resources, states may
begin pursue CBW as a hedge against the failure of a costlier, more technically challenging
nuclear weapons program.
Two case studies illustrate how the two di↵ erent types of regime security threats shape
chemicalandbiologicalweaponspursuitchoices. InLibya,Qaddafi’sfearofbeingoverthrown
by opposing elements within the governing elite led him to dismantle much of the existing
state bureaucracy. This “coup-proofed” state struggled to e↵ ectively monitor and manage a
large, complex nuclear weapons program, making a smaller-scale chemical weapons program
adesirableoptiontocounterarangeofsecuritythreats. InRhodesia, anincreasingly
desperate struggle against a numerically asymmetrical insurgency prompted chemical and
biologicalweaponspursuit, inthehopesincitingfearamongsttheinsurgentsandthecivilian
population and generating casualties covertly and with relatively little manpower.
Understanding these mechanisms provides a necessary foundation for further inquiry
into the causes and consequences of chemical and biological weapons proliferation. Interna-
tional relations scholars have built up a substantial body of knowledge regarding the causes
and consequences of nuclear weapons pursuit. In comparison, the research on chemical and
biological weapons proliferation has been less systematic and cohesive. Not only does this
dissertation advance our understanding of CBW proliferation, it also provides potential in-
sight into other tradeo↵ smadeinstates’armingdecisions,andintothesubstitutabilityof
111
foreign policy tools more generally.
1
Policy Implications
Appreciationofthedynamicsthatdrivechemicalandbiologicalweaponsproliferation—and
howtheyrelatetothedynamicsdrivingnuclearproliferation—iscrucialtotheformationof
e↵ ectivenon-proliferationpolicythatavertsunintendednegativeconsequences. Forinstance,
if CBW, like nuclear weapons, are pursued primarily for strategic external security purposes,
then attempts to stem nuclear proliferation may inadvertently spur the spread of other
WMD. On the other hand, if possessors of CBW see them as fulfilling di↵ erent goals than
nuclearweapons,wemayneedtodevelopnewunderstandingsofarmscontrolanddeterrence
that better apply to CBW. In particular, given the important role that internal security
threats can play in driving decision to initiate CBW pursuit, powerful states seeking to
undermine hostile regimes domestically may inadvertently spur CBW pursuit by adversaries
—acautiontothosewhopintheircounterproliferationhopesonthepursuitofregime
change in rogue states.
Even when di↵ erentiated in analysis from nuclear weapons, chemical and biological
weapons themselves are often grouped together as the “other WMD”. This too can be dan-
gerous — as the analysis presented here demonstrates, the motivations for and goals of
biological weapons pursuit may di↵ er substantially from those for chemical weapons pursuit,
with biological weapons being viewed as more “strategic” in nature by both proliferators
and their rivals. The findings of this dissertation point to the need to disaggregate the study
of motivations for weapons pursuit, not only within overarching categories such as “WMD”
and“CBW,”butalsowithinseeminglymorediscretecategoriessuchas“chemicalweapons.”
Analystsandpolicymakersoftenrelyoncategorizationofweaponstypesasaheuristic
1
On the theoretical importance of studying foreign policy substitutability, see, e.g., Most and
Starr (1984); Starr (2000)
112
for thinking about weapons’ e↵ ects, potential uses, and why actors might want to acquire
them. However, even within traditional weapons categories such as ‘chemical weapons?, it is
worth thinking about the range of weapons characteristics encompassed by a given category,
and the potential variation in motivations for pursuing di↵ erent tools within that category.
For instance, governments pursuing small quantities of a nerve agent such as VX, with the
intent to use it as a means of assassination may be seeking to address di↵ erent security
threats than those stockpiling large quantities of such agents and developing the means to
deliver them via ballistic missiles or gravity bombs.
This insight applies to other types of weapons as well. Consider the broad category
of cyber weapons, the subject of much scholarly and policy attention. Much attention has
been devoted to the potential for “strategic cyber” weapons.
2
However, most actual cyber
attacks are much more limited in nature, and, while they can have important counterforce
implications, cyber capabilities alone often do not provide the same countervalue utility as
traditional strategic weapons (Gartzke, 2013; Li↵ ,2012;GartzkeandLindsay,2017). This
suggests a need to distinguish between di↵ erent types of cyber capabilities when analyzing
the potential causes and consequences of the proliferation of cyber weapons capabilities.
Theoretical Implications
Understanding the role that internal threats can play in driving states’ arming behavior also
has several important implications for international security more generally.
First, my findings add to our understanding of how internal security considerations
can lead states to adopt seemingly suboptimal defense policies. Regimes facing significant
internal security threats engage in omnibalancing, choosing allies that they believe would be
able and willing to provide assistance against domestic threats, rather than those best posi-
2
For example, see: Jasper (2017); Schneider (2020); Andres (2012)
113
tionedtocounterexternalsecuritythreats(David,1991b). Wheredomesticsecurityconcerns
are centered on the possibility of a military coup d’´ etat, regimes engage in counterbalanc-
ing, dividing power between multiple, redundant security forces, and other “coup-proofing”
measuresthatcanhinderconventionalmilitarye↵ ectiveness(BelkinandSchofer,2003). Fur-
thermore, where either elites or society are highly divided, “incoherent states” may fail to
su cientlybalanceexternalthreats,asleadersareunwillingtotakeonthecostsandpolitical
risks of balancing (Schweller, 2010). This dissertation illustrates how regime security threats
— and the measures taken to mitigate them — can a↵ ect the relative cost and feasibility
of di↵ erent types of weapons pursuit, leading to weapons pursuit choices that might appear
puzzling from a purely national security perspective.
Second, the findings presented here point to potential domestic sources for security
dilemmas. Security dilemmas may occur wherein a regime’s pursuit of chemical or biological
weapons to counter internal security threats is be perceived as threatening by external ad-
versaries, who respond by arming or undertaking preemptive attacks. Because chemical and
biological weapons have both internal and external security applications, it can be di cult
for an adversary to discern a state’s intentions in pursuing those weapons, just as it can be
di cult to discern the o↵ ensive or defensive intentions of conventional arming. Furthermore,
as a state’s security environment shifts — for instance, if an adversary engages in reactive
proliferation — its intentions for its CBW may shift as well, and chemical or biological
weapons originally pursued for domestic security reasons may also come to serve national
security purposes.
Avenues for Further Research
The findings here suggest several directions for future research on the topic of chemical
and biological weapons proliferation. First, if states pursue chemical and biological weapons
becausedecision-makersperceivethemasusefultoolstoaddressparticularsecurityconcerns,
114
do CBW, once acquired, actually fulfill those security purposes. Does chemical or biological
weapons acquisition increase the likelihood of regime survival? Does it deter chemical or
biological weapons attacks by adversaries? Furthermore, do governments that possess CBW
actually deploy them to counter internal threats?
More generally, this points to the question of the relationship between CBW prolifer-
ation motivations, and the motivations for CBW use. While leaders might pursue chemical
or biological weapons in response to either internal or external security threats, my findings
suggest that, when facing significant external security threats, regimes concerned about do-
mestic threats could be more likely to choose chemical and/or biological weapons pursuit
over (or in addition to) other options, such as nuclear weapons pursuit. In these cases,
even if domestic use was not the original intention, those weapons are then available for
the regime to use if the domestic security situation escalates. For instance, North Korea,
initially pursued chemical weapons as a deterrent to aggression from the United States and
South Korea, but decision-makers later concluded that chemical weapons were insu cient
to deter nuclear threats, and embarked on a nuclear weapons program. However, the North
Korean regime has since made use of that chemical weapons stockpile in the assassination of
Kim Jong-nam, a perceived threat to Kim Jong-un’s leadership (Bermudez, 2000; Parachini,
2018). Similarly, Syria likely began chemical weapons pursuit in the hopes of deterring Is-
raeli nuclear threats, but the Syrian government has infamously deployed chemical weapons
numerous times during the ongoing civil war (Diab, 1997; Chapman, Elbahtimy and Martin,
2018). Furtherinvestigationofsuchcasescouldilluminatetherelationshipbetweenweapons
pursuit decisionmaking and weapons use decisionmaking.
Additionally, as the case studies illustrate, the independent variables that drive chemi-
cal and biological weapons pursuit can be interrelated. In particular, external threat percep-
tions often involve concerns that powerful rivals will intervene in a state’s domestic a↵ airs,
and at the extreme, attempt to force regime change. On the other hand, leaders facing
domestic challenges often provoke external disputes as a diversionary tactic. Further com-
115
plicating these relationships, external adversaries often use economic sanctions as a tool to
coercechangesinbehaviororprovokeregimechange. Undersomeconditions, suchsanctions
can cause significant economic hardship, provoking further domestic unrest. Disentangling
these relationships can provide deeper insight into the mechanisms by which regime security
concerns influence proliferation choices.
One commonality that emerges between the case studies is both Libya and Rhodesia’s
governing regimes’ disregard for international norms, and opposition to the international
status quo (or, in the case of Rhodesia, opposition to the newly emerging status quo).
Rhodesia resisted global momentum towards decolonization and majority rule, incurring
international opprobrium, while Libya under Qaddafi adopted an interventionist foreign
policy and extensively sponsored rebel organizations and terrorist groups, under the banner
of resisting Israel and Western influence. What role does this revisionism play in chemical
and biological weapons proliferation choices? Is overall disregard for international norms a
necessary precondition for the pursuit of “taboo,” stigmatized weapons? Revisionist states
are more likely to perceive — and provoke — external security threats; are their governing
regimes also more likely to experience security threats from within?
Finally, both of the case studies point to a need to further investigate the prevalence of
foreign assistance in chemical and biological weapons pursuit e↵ orts, and its role in shaping
proliferation outcomes. Libya’s chemical weapons program relied on an extensive network of
foreign suppliers, while Rhodesia’s CBW programs depended on South Africa for materials
and scientific advice. Furthermore, although South Africa was not actively pursuing CBW
duringthetimethattheyprovidedassistancetotheRhodesianprogram,tiestothatprogram
appear to have influenced their own CBW proliferation e↵ orts, which began shortly after
the Rhodesian regime came to an end. Beyond providing supplies and technical assistance,
some countries — most frequently the Soviet Union — have provided finished agents for
other states to use. Understanding the causes and consequences of such assistance appears
to be an important piece of the CBW proliferation puzzle.
116
Bibliography
Ahmida, Ali Abdullatif. 1994. The making of modern Libya: state formation, colonization,
and resistance, 1830-1932. SUNY Press.
Alexander, Jocelyn and Jo Ann McGregor. 2005. “Hunger, violence and the moral economy
ofwarin.”Violence And Belonging: The Quest For Identity In Post-Colonial Africap.75.
Ali,Javed.2001. “ChemicalweaponsandtheIran-Iraqwar: Acasestudyinnoncompliance.”
The Nonproliferation Review 8(1):43–58.
Anderson, Lisa. 1982. “Libya and American foreign policy.” Middle East Journal 36(4):516–
534.
Anderson, Lisa. 1986. The state and social transformation in Tunisia and Libya, 1830-1980.
Princeton University Press.
Anderson, Lisa. 2020. “A Pool of Water: Perspectives on the Libyan Revolution.” Lamma:
A Journal of Libyan Studies 1.
Andres, Richard.2012. “TheEmergingStructureofStrategicCyberO↵ ense, CyberDefense,
and Cyber Deterrence.” Cyberspace and National Security: Threats, Opportunities, and
Power in a Virtual World pp. 89–104.
Bale, Je↵ rey M. 2006. “South Africa?s Project Coast:?Death Squads,? Covert State-
Sponsored Poisonings, and the Dangers of CBW Proliferation.” Democracy and Security
2(1):27–59.
117
Barber, William J. 1961. “The Economy of British Central Africa.”.
Barnum, Miriam and Bryan L Fearey. 2016. “Sanctions as a nonproliferation tool: Lessons
from Libya.” Comparative Strategy 35(4):234–245.
Belkin, Aaron and Evan Schofer. 2003. “Toward a structural understanding of coup risk.”
Journal of Conflict Resolution 47(5):594–620.
Bell,AnthonyandDavidWitter.2011. ROOTSOFREBELLION:PARTI.Technicalreport
Institute for the Study of War.
URL: http://www.jstor.org/stable/resrep07869
Bentley, Michelle. 2013. “War and/of words: constructing WMD in US foreign policy.”
Security Studies 22(1):68–97.
Bermudez, Joseph Jr. 2000. The democratic People?s Republic of Korea and unconventional
weapons. In Planning the Unthinkable: How New Powers Will Use Nuclear, Biological,
and Chemical Weapons., ed. Peter Ren´ e Lavoy, Scott Douglas Sagan and James J Wirtz.
Cornell University Press.
Bleek, Philipp C. 2017. When Did (and Didn’t) States Proliferate?: Chronicling the Spread
of Nuclear Weapons.BelferCenterforScienceandInternationalA↵ airs.
Bowen, Wyn Q. 2017. Libya and nuclear proliferation: stepping back from the brink.Rout-
ledge.
Braut-Hegghammer, M˚ alfrid. 2016. Unclear Physics: Why Iraq and Libya Failed to Build
Nuclear Weapons. Cornell University Press.
Brooks, Risa. 2013. Political-military relations and the stability of Arab regimes. Routledge.
Brown, Cameron S, Christopher J Fariss and R Blake McMahon. 2016. “Recouping after
coup-proofing: Compromised military e↵ ectiveness and strategic substitution.” Interna-
tional Interactions 42(1):1–30.
118
Brown, Fredric. 2017. Chemical warfare: A study in restraints. Routledge.
Brownell, Josiah. 2008. “The hole in Rhodesia’s bucket: White emigration and the end of
Settler rule.” Journal of Southern African Studies 34(3):591–610.
Bueno De Mesquita, Bruce. 1999. Principles of International Politics: People’s Power,
Preferences, and Perceptions. CQ Press.
Bundestag, Deutscher. 1989. Bericht der Bundesregierung an den Deutschen Bundestag
¨ ubereinem¨ oglicheBeteiligungdeutscherFirmenaneinerC-Wa↵ en-ProduktioninLibyen.
Technical report Unterrichtung durch die Bundesregierung.
URL: https://dserver.bundestag.de/btd/11/039/1103995.pdf
Burck, Gordon M and Charles C Flowerree. 1991. International handbook on chemical
weapons proliferation. Greenwood Pub Group.
Burgess, Stephen F and Helen E Purkitt. 2001. The rollback of South Africa’s chemical and
biological warfare program. Technical report AIR UNIV MAXWELL AFB AL CENTER
FOR AEROSPACE DOCTRINE RESEARCH AND EDUCATION.
Busch, Nathan E and Joseph F Pilat. 2017. The politics of weapons inspections: Assessing
WMD monitoring and verification regimes. Stanford University Press.
Carus, W Seth. 1991. ” The Poor Man’s Atomic Bomb?”: Biological Weapons in the Middle
East. Washington, DC: Washington Institute for Near East Policy.
Carus, W Seth. 2001. Bioterrorism and biocrimes: the illicit use of biological agents since
1900. Technical report NATIONAL DEFENSE UNIV WASHINGTON DC.
Chapman, Geo↵ rey, Hassan Elbahtimy and Susan B Martin. 2018. “The Future of Chemical
Weapons: Implications from the Syrian Civil War.” Security Studies 27(4):704–733.
Chenoweth, Erica and Orion A Lewis. 2013. “Unpacking nonviolent campaigns: Introducing
the NAVCO 2.0 dataset.” Journal of Peace Research 50(3):415–423.
119
Chevrier, Marie I. 1993. “Deliberate disease: biological weapons, threats, and policy re-
sponses.” Environment and Planning C: Government and Policy 11(4):395–417.
Cilliers, Jakkie. 2015. Counter-Insurgency in Rhodesia (RLE: Terrorism and Insurgency).
Routledge.
Clark, Robert M and William L Mitchell. 2018. Deception: Counterdeception and counter-
intelligence. CQ Press.
Collier, David and James E Mahon. 1993. “Conceptual ?stretching? revisited: Adapting
categories in comparative analysis.” American Political Science Review 87(4):845–855.
Collins, Stephen D. 2004. “Dissuading state support of terrorism: Strikes or sanctions?(An
analysisofdissuasionmeasuresemployedagainstLibya).”Studies in Conflict & Terrorism
27(1):1–18.
Cooley, John K. 1982. Libyan sandstorm. New York: Holt, Rinehart, and Winston.
Cordesman, Anthony H and Eric E Fillinger. 2001. Weapons of Mass Destruction and the
Global Nuclear Balance: A Quantitative and Arms Control Analysis. Center for Strategic
and International Studies Washington, DC.
Cross, Glenn. 2017. Dirty War: Rhodesia and Chemical Biological Warfare 1975-1980.
Helion and Company.
David, Steven R. 1991a. Choosing sides: Alignment and realignment in the Third World.
Johns Hopkins University Press.
David, Steven R. 1991b. “Explaining third world alignment.” World Politics 43(2):233–256.
Davies, JC. 1983. “A major epidemic of anthrax in Zimbabwe.” Central African Journal of
Medicine 29(1):8–12.
120
Davies, JCA. 1982. “A major epidemic of anthrax in Zimbabwe: The spread of the epidemic
inMatabeleland,MidlandsandMashonalandProvincesduringtheperiodNovember,1978
to October, 1980.” Central African Journal of Medicine 28(12):291–298.
Davies, JCA. 1985. “A major epidemic of anthrax in Zimbabwe. The experience at the
Beatrice Road Infectious Diseases Hospital, Harare.” Central african journal of medicine
31(9):176–180.
Dawson, GrantandEvangeliaLinaki.2019. “TheLegalChallengesintheMissiontoRemove
theRemainingChemicalWeaponStockpilesfromLibya.”Journal of Conflict and Security
Law 24(1):53–70.
Deen, Jacqueline, Martin A Mengel and John D Clemens. 2020. “Epidemiology of cholera.”
Vaccine 38:A31–A40.
Diab, M Zuhair. 1997. “Syria’s chemical and biological weapons: Assessing capabilities and
motivations.” The Nonproliferation Review 5(1):104–111.
Director General, IAEA. 2004. “Implementation of the NPT Safeguards Agreement of the
SocialistPeople’sLibyanArabJamahiriya.”Vienna: International Atomic Energy Agency
.
Drazanova, Lenka. 2019. “Historical index of ethnic fractionalization dataset (HIEF).” Har-
vard Dataverse 1.
Early, Bryan R and Christopher Way. 2012. Launching Nukes and Nuclear Weapons Pro-
grams: The Spread of Ballistic Missile Technology and Nuclear Proliferation. In Annual
Meeting of the American Political Science Association, New Orleans. Vol. 30.
Einhorn, Robert J and Michele A Flournoy. 2003. Protecting Against the Spread of Nuclear,
Biological, and Chemical Weapons: Agenda for action. Vol. 1 CSIS.
121
Ellert, Henrik. 1989. The Rhodesian Front War: counter-insurgency and guerrilla war in
Rhodesia, 1962-1980. Mambo Press.
Emadi, Hafizullah. 2012. “Libya: The road to regime change.” Global Dialogue (Online)
14(2):128.
Eyre, Dana P, Mark C Suchman et al. 1996. “Status, norms, and the proliferation of con-
ventional weapons: An institutional theory approach.” The culture of national security:
Norms and identity in world politics pp. 79–113.
Fearon, James D. 2018. “Cooperation, conflict, and the costs of anarchy.” International
Organization 72(3):523–559.
Feigenbaum, Anna. 2017. Tear Gas: From the Battlefields of World War I to the Streets of
Today. Verso Books.
Frankenstein,William,GhitaMezzour,KathleenMCarleyandLRichardCarley.2015. “Re-
mote assessment of countries? nuclear, biological, and cyber capabilities: joint motivation
and latent capability approach.” Social Network Analysis and Mining 5(1):5.
Fuhrmann,MatthewandMichaelCHorowitz.2014.“Whenleadersmatter: Rebelexperience
and nuclear proliferation.” The Journal of Politics 77(1):72–87.
Gartzke, Erik. 2013. “The myth of cyberwar: bringing war in cyberspace back down to
earth.” International Security 38(2):41–73.
Gartzke, ErikandJonRLindsay.2017. “Thermonuclearcyberwar.”Journal of cybersecurity
3(1):37–48.
Gartzke, Erik and Matthew Kroenig. 2009. “A strategic approach to nuclear proliferation.”
Journal of Conflict Resolution 53(2):151–160.
Gaub, Florence. 2013. “The Libyan armed forces between coup-proofing and repression.”
Journal of Strategic Studies 36(2):221–244.
122
Geissler, Erhard and John Ellis van Courtland Moon. 1999. Biological and toxin weapons:
research, development and use from the Middle Ages to 1945. Oxford University Press.
Godwin, Peter, Ian Hancock et al. 1993. ’Rhodesians Never Die’: The Impact of War and
Political Change on White Rhodesia, C. 1970-1980. Oxford University Press, USA.
Goodwin,Bridget.1998. KeenasMustard: Britain’sHorrificChemicalWarfareExperiments
in Australia. Univ. of Queensland Press.
Gould, Chandre. 2005. South Africa’s chemical and biological warfare programme 1981-1995
PhD thesis Rhodes University.
Gould, Chandr´ e and Peter I Folb. 2000. “The South African chemical and biological warfare
program: an overview.” The Nonproliferation Review 7(3):10–23.
Gregory, Martyn. 1980. “Rhodesia: From Lusaka to Lancaster House.” The World Today
36(1):11–18.
Guillemin, Jeanne. 2005. “Biological weapons and secrecy (WC 2300).” FASEB jour-
nal: o cial publication of the Federation of American Societies for Experimental Biology
19(13):1763–1765.
Hagan, Joe D. 1987. Regimes, Political Oppositions, and the comparative analysis of foreign
policy.InNewdirectionsinthestudyof foreignpolicy,ed.CharlesW.KegleyJr.Hermann,
Charles F. and James N. Rosenau. Allen & Unwin Boston, MA pp. 1946–84.
Hargreaves, John D. 2014. Decolonization in Africa. Routledge.
Harris, ElisaD.1990. “ChemicalWeaponsProliferation: CurrentCapabilitiesandProspects
for Control.” Aspen Strategy Group, New Threats: Responding to the Proliferation of
Nuclear, Chemical, andDeliveryCapabilitiesintheThirdWorld(Lanham, MD:University
Press of America, 1990) pp. 71–2.
123
Herdman, Roger C. 1993. Technologies Underlying Weapons of Mass Destruction. Technical
report OFFICE OF TECHNOLOGY ASSESSMENT WASHINGTON DC.
Horowitz, B Zane. 2005. “Botulinum toxin.” Critical care clinics 21(4):825–839.
Horowitz, Donald L. 1895. Ethnic groups in conflict, updated edition with a new preface.
Univ of California Press.
Horowitz, Michael C and Neil Narang. 2014. “Poor Man’s atomic bomb? exploring the rela-
tionshipbetween’weaponsofmassdestruction’.”JournalofConflictResolution58(3):509–
535.
Hymans, JacquesEC.2004. TherootsoftheWashingtonthreatconsensus.InStriking First.
Springer pp. 33–45.
Hymans, Jacques EC. 2006. “Theories of nuclear proliferation: The state of the field.”
Nonproliferation Review 13(3):455–465.
Hymans, Jacques EC. 2012. Achieving nuclear ambitions: scientists, politicians, and prolif-
eration. Cambridge University Press.
Inglesby,ThomasV,TaraO’Toole,DonaldAHenderson,JohnGBartlett,MichaelSAscher,
Edward Eitzen, Arthur M Friedlander, Julie Gerberding, Jerome Hauer, James Hughes
et al. 2002. “Anthrax as a biological weapon, 2002: updated recommendations for man-
agement.” Jama 287(17):2236–2252.
Jackson, Richard. 2007. Regime security. Oxford University Press Oxford pp. 146–63.
James, Larry C and Terry L Oroszi. 2015. “Defining Weapons of Mass Psychological De-
struction.” Weapons of Mass Psychological Destruction and the People Who Use Them
pp. 3–12.
Jasper, Scott. 2017. Strategic cyber deterrence: The active cyber defense option. Rowman &
Littlefield.
124
Jervis, Robert. 1978. “Cooperation under the security dilemma.” World politics 30(2):167–
214.
Jo, Dong-Joon and Erik Gartzke. 2007. “Determinants of nuclear weapons proliferation.”
Journal of Conflict Resolution 51(1):167–194.
Joseph, Robert G. 2009. Countering WMD: the Libyan experience. Vol. 7 National Institute
Press Fairfax, VA.
Kegley, Charles W. 1980. “International and domestic correlates of nuclear proliferation: A
comparative analysis.” Korea and World A↵ airs 4(1).
Kemp, R Scott. 2014. “The nonproliferation emperor has no clothes: the gas centrifuge,
supply-side controls, and the future of nuclear proliferation.” International Security
38(4):39–78.
Kennedy, Dane Keith. 1987. Islands of white : settler society and culture in Kenya and
Southern Rhodesia, 1890-1939. Durham [N.C: .
Kennedy, John F. October 13, 1960. “Third Nixon-Kennedy Presidential Debate.”.
Kerr, Paul K. 2008. Nuclear, biological, and chemical weapons and missiles: Status
and trends. LIBRARY OF CONGRESS WASHINGTON DC CONGRESSIONAL RE-
SEARCH SERVICE.
Koblentz, Gregory. 2003. “Pathogens as weapons: the international security implications of
biological warfare.” International security pp. 84–122.
Koblentz, Gregory D. 2010. Living Weapons: Biological Warfare and International Security.
Cornell University Press.
Koblentz,GregoryD.2013.“RegimeSecurity: ANewTheoryforUnderstandingtheProlifer-
ation of Chemical and Biological Weapons.” Contemporary Security Policy 34(3):501–525.
125
Krutzsch, Walter, Eric Myjer and Ralf Trapp. 2014. The chemical weapons convention: a
commentary. OUP Oxford.
Legro, Je↵ rey. 1995. Cooperation Under Fire: Anglo-German Restraint During World War
II. Cornell University Press.
Li↵ ,AdamP.2012. “Cyberwar: anew?absoluteweapon?? Theproliferationofcyberwarfare
capabilities and interstate war.” Journal of Strategic Studies 35(3):401–428.
Mahley, Donald. 2004. “Dismantling Libyan Weapons: Lessons Learned.” The Arena 10.
Mangold, Tom and Je↵ Goldberg. 2001. Plague wars: the terrifying reality of biological
warfare. Macmillan.
Martin, David, Phyllis Johnson and Robert Mugabe. 1981. The struggle for Zimbabwe: The
chimurenga war. Faber & Faber London.
Martinez, Ian. 2002. “The history of the use of bacteriological and chemical agents dur-
ing Zimbabwe’s liberation war of 1965-80 by Rhodesian forces.” Third World Quarterly
23(6):1159–1179.
McMahon, R Blake and Branislav L Slantchev. 2015. “The guardianship dilemma:
Regime security through and from the armed forces.” American Political Science Review
109(2):297–313.
Meyer, StephenM.1986. The dynamics of nuclear proliferation. UniversityofChicagoPress.
Miller, Jamie. 2016. An African volk: The Apartheid regime and its search for survival.
Oxford University Press.
Miller, Nicholas L. 2014. “Nuclear Dominoes: A Self-Defeating Prophecy?” Security Studies
23(1):33–73.
Moeschlin, Sven. 1980. “Thallium poisoning.” Clinical toxicology 17(1):133–146.
126
Montgomery, Alexander H. 2013. “Stop helping me: When nuclear assistance impedes
nuclear programs.” The Nuclear Renaissance and International Security pp. 177–202.
Moorcraft, Paul L and Peter McLaughlin. 2010. The Rhodesian war: A military history.
Stackpole Books.
Morens, David M. 2003. “Characterizing a ?new? disease: epizootic and epidemic anthrax,
1769–1780.” American journal of public health 93(6):886–893.
Most, Benjamin A and Harvey Starr. 1984. “International relations theory, foreign policy
substitutability, and ?nice? laws.” World Politics 36(3):383–406.
Mueller, Harald and Andreas Schmidt. N.d. The Little Known Story of De-Proliferation:
Why States Give Up Nuclear Weapon Activities.
Narang, Neil. 2015. “All Together Now? Questioning WMDs as a Useful Analytical Unit
for Understanding Chemical and Biological Weapons Proliferation.” The Nonproliferation
Review 22(3-4):457–468.
Nass, Meryl. 1992. “Anthrax epizootic in Zimbabwe, 1978-1980: Due to deliberate spread?”.
Nesi, Giuseppe. 2011. “Recognition of The Libyan National Transitional Council: When,
How and Why.” The Italian Yearbook of International Law Online 21(1):45–59.
Nordhaus, William, John R Oneal and Bruce Russett. 2012. “The e↵ ects of the interna-
tional security environment on national military expenditures: A multicountry study.”
International Organization 66(3):491–513.
NTI, The Nuclear Threat Initiative. 2011. “Libya Chemical Facilities.”. [Online; accessed
13-September-2021].
URL: https://www.nti.org/analysis/articles/libya-chemical-facilities/
OPCW. 2012. “OPCW Inspectors Verify Newly Declared Chemical Weapons Materials in
Libya.”.
127
Palmer, Glenn, Roseanne W McManus, Vito D?Orazio, Michael R Kenwick, Mikaela
Karstens, Chase Bloch, Nick Dietrich, Kayla Kahn, Kellan Ritter and Michael J Soules.
2021. “The MID5 Dataset, 2011–2014: Procedures, coding rules, and description.” Con-
flict Management and Peace Science p. 0738894221995743.
Parachini, John. 2018. Assessing North Korea’s Chemical and Biological Weapons Capabili-
ties and Prioritizing Countermeasures. RAND.
Parker, Jim. 2006. Assignment Selous Scouts: Inside Story of a Rhodesian Special Branch
O cer. Alberton: Galago Publishers.
Paul, ThazhaVarkey.2009. The tradition of non-use of nuclear weapons. StanfordUniversity
Press.
Pilster, Ulrich and Tobias B¨ ohmelt. 2011. “Coup-proofing and military e↵ ectiveness in
interstate wars, 1967–99.” Conflict Management and Peace Science 28(4):331–350.
Pine, Art. 1996. “Only A-Bomb Could Destroy Libya Plant, Scientist Says.” Los Angeles
Times .
Plunkett, Geo↵ .2007. Chemical warfare in Australia. Australian Military History Publica-
tions.
Powell, Jonathan. 2012. “Determinants of the Attempting and Outcome of Coups d?´ etat.”
Journal of Conflict resolution 56(6):1017–1040.
Prados, John. 2005. “How Qaddafi came clean.” Bulletin of the Atomic Scientists 61(6):26–
33.
Press, Daryl G, Scott D Sagan and Benjamin A Valentino. 2013. “Atomic aversion: Exper-
imental evidence on taboos, traditions, and the non-use of nuclear weapons.” American
Political Science Review 107(1):188–206.
128
Price, Richard. 1995. “A genealogy of the chemical weapons taboo.” International Organi-
zation 49(1):73–103.
Price, Richard, M. 1997. The Chemical Weapons Taboo. Ithaca, NY: Cornell University
Press.
Purkitt, Helen E and Stephen Burgess. 2002. “South Africa’s chemical and biological war-
fare programme: A historical and international perspective.” Journal of Southern African
Studies 28(2):229–253.
Raeburn, Michael.1978. We are Everywhere: Narratives from Rhodesian Guerillas.Random
House (NY).
Rall, David P, Constance M Pechura et al. 1993. “Veterans at risk: The health e↵ ects of
mustard gas and lewisite.”.
Rettie, Allan E and Guoying Tai. 2006. “The pharmocogenomics of warfarin.” Molecular
interventions 6(4):223.
Ritchie, Elspeth Cameron, Patricia J Watson and Matthew J Friedman. 2015. Interventions
following mass violence and disasters: Strategies for mental health practice. Guilford
Publications.
Robb, Charles S, Laurence H Silberman, Richard C Levin, John McCain, Henry S Rowen,
Walter B Slocombe, William O Studeman, Charles M Vest, Patricia Wald and Lloyd Cut-
ler. 2005. The commission on the intelligence capabilities of the united states regarding
weaponsofmassdestruction: Reporttothepresidentoftheunitedstates.Technicalreport
COMMISSIONONINTELLIGENCECAPABILITIESREGARDINGWMDWASHING-
TON DC.
Roberts, James R and J Routt Reigart. 2013. Recognition and management of pesticide poi-
129
sonings, 6th edition. United States Environmental Protection Agency, O ce of Pesticide
Programs.
Robinson, JP Perry. 1989. Supply, demand and assimilation in chemical-warfare armament.
In Military Technology, Armaments Dynamics and Disarmament. Springer pp. 112–123.
Rogers, Anthony.2001. Someone else’s war: mercenaries from 1960 to the present. Trafalgar
Square.
Romano Jr, James A and James M King. 2001. “Psychological casualties resulting from
chemical and biological weapons.” Military medicine 166(suppl 2):21–22.
Ryan, Johnny. 2004. “Principled failure: British policy toward Rhodesia, 1971–72.” The
History Review .
Sagan, Scott D. 1997. “Why do states build nuclear weapons? Three models in search of a
bomb.” International security 21(3):54–86.
Sartori, Giovanni. 1970. “Concept misformation in comparative politics.” American political
science review 64(4):1033–1053.
Schneider, Jacquelyn. 2020. “A strategic cyber no-first-use policy? Addressing the US cyber
strategy problem.” The Washington Quarterly 43(2):159–175.
Schumacher, Edward. 1986. “The United States and Libya.” Foreign A↵ airs 65(2):329–348.
Schweller, Randall L. 2010. Unanswered Threats. Princeton University Press.
Seawright, Jason and John Gerring. 2008. “Case selection techniques in case study research:
A menu of qualitative and quantitative options.” Political research quarterly 61(2):294–
308.
Sinai, Joshua. 1997. “Libya’s pursuit of weapons of mass destruction.” The Nonproliferation
Review 4(3):92–100.
130
Singh, Sonali and Christopher R Way. 2004. “The correlates of nuclear proliferation: A
quantitative test.” Journal of Conflict Resolution 48(6):859–885.
Smith, Bradley C and William Spaniel. 2020. “Introducing ⌫ -CLEAR: a latent variable
approach to measuring nuclear proficiency.” Conflict Management and Peace Science
37(2):232–256.
Smith, James A. 2000. “Search for national security in an NBC World.” Colorado Springs,
CO: USAF Institute for National Security Studies, US Air Force Academy .
Smith, R. Je↵ rey, Joby Warrick and Colum Lynch. 2011. “Iran may have sent Libya shells
for chemical weapons.” The Washington Post .
Spiers, E. 1994. Chemical and Biological Weapons: A study of proliferation. Springer.
Spiers, Edward M. 1989. Chemical Weaponry: A Continuing Challenge. Springer.
Starr, Harvey. 2000. “Substitutability in foreign policy: Theoretically central, empirically
elusive.” Journal of Conflict Resolution 44(1):128–138.
Sti↵ , Peter. 2002. “See you in November: The story of an SAS Assassin.”.
Stokes, James W and Louis E Banderet. 1997. “Psychological aspects of chemical defense
and warfare.” Military Psychology 9(4):395–415.
Sudduth, Jun Koga. 2017. “Coup risk, coup-proofing and leader survival.” Journal of Peace
Research 54(1):3–15.
Takeyh, Ray. 2001. “The rogue who came in from the cold.” Foreign a↵ airs pp. 62–72.
Talmadge, Caitlin. 2015. The Dictator’s Army. Cornell University Press.
Tannenwald, Nina. 1999. “The nuclear taboo: The United States and the normative basis
of nuclear non-use.” International organization 53(3):433–468.
131
Terrell, Patrick, Katharine Hagen and Ted A Ryba. 2016. “Eliminating Libya?s WMD
programs: creating a cooperative situation.” The Nonproliferation Review 23(1-2):185–
196.
Terrill, W Andrew. 1994. “Libya and the Quest for Chemical Weapons.” Journal of Conflict
Studies 14(1).
Thompson, William and David Dreyer. 2011. Handbook of international rivalries. CQ Press.
Toulabi, Biejan Poor. 2021. The Myth of the Poor Man’s Atomic Bomb and the Politics
of Proliferation: Knowledge, Method, and Ideology in the Study of Chemical, Biological,
and Nuclear Weapons PhD thesis Vrije Universiteit Amsterdam.
Tucker, Jonathan. 2000. “Motivations for and against proliferation: The case of the middle
east.” Biological Warfare: Modern O↵ ense and Defense pp. 27–52.
Tucker, Jonathan B. 2001. “The “yellow rain” controversy: lessons for arms control compli-
ance.” The Nonproliferation Review 8(1):25–42.
Tucker, Jonathan B. 2009. “The rollback of Libya’s chemical weapons program.” Nonprolif-
eration Review 16(3):363–384.
United Nations, Press Section. 1974. “Statement by Spokesman for Commission of Inquiry
into Reported Massacres in Mozambique.”.
United States Congress, Senate Committee on Armed Services. 1985. Department of defense
authorization for appropriations for fiscal year 1986: Unified commands binary chemicals.
Department of defense authorization for appropriations for fiscal year 1986: Hearings
before the committee on armed services, united states senate, ninety-ninth congress, first
session, on s. 674 U.S. Government Printing O ce. tex.lccn: 85602365.
URL: https://books.google.com/books?id=9ilIAQAAMAAJ
Vandewalle, Dirk. 2012. A history of modern Libya. Cambridge University Press.
132
Vandewalle, Dirk. 2018. Libya since independence. In Libya since Independence. Cornell
University Press.
Velsko, SP. 2015. Using Dempster-Shafer theory to assess explanations for the Zimbabwe
anthrax outbreak of 1978-79. Technical report Lawrence Livermore National Lab.(LLNL),
Livermore, CA (United States).
Vishwanathan, Arun. 2012. “2011 Libyan Disclosures:’Take-away’for CWC verification.”
CBW Magazine: Journal of Chemical and Biological Weapons 5(1):17–22.
Waltz, Kenneth N. 1979. Theory of international politics. McGraw-Hill.
Way,ChristopherandJessicaLPWeeks.2014. “Makingitpersonal: regimetypeandnuclear
proliferation.” American Journal of Political Science 58(3):705–719.
Weller, Nicholas and Jeb Barnes. 2016. “Pathway analysis and the search for causal mecha-
nisms.” Sociological Methods & Research 45(3):424–457.
Wheelis, Mark and Lajos R´ ozsa. 2009. Deadly cultures: biological weapons since 1945.
Harvard University Press.
Wiegele, Thomas C. 1992. The Clandestine Building of Libya’s Chemical Weapons Factory:
A Study in International Collusion. Southern Illinois University Press.
Wills, Alfred J. 1967. “An introduction to the history of Central Africa.”.
Wilson, James M, Walter Brediger, Tom Albright and Julie Smith-Gagen. 2016. “Reanalysis
of the Anthrax Epidemic in Rhodesia, 1978-84.” bioRxiv p. 053769.
Wimmer, Andreas, Lars-Erik Cederman and Brian Min. 2009. “Ethnic politics and armed
conflict: Aconfigurationalanalysisofanewglobaldataset.”AmericanSociologicalReview
74(2):316–337.
133
Zanders, Jean Pascal. 1995. “Towards understanding chemical warfare weapons prolifera-
tion.” Contemporary Security Policy 16(1):84–110.
Zelico↵ , Alan P. 2001. “An impractical protocol.” Arms Control Today 31(4):25.
134
Appendix A
Regression Tables
135
Table A.1: Security threats and CW pursuit: fixed e↵ ects models
Dependent variable:
Risk of Chemical Weapons Pursuit (restrictive)
(1) (2) (3) (4) (5)
Coup risk 0.675
⇤⇤⇤
(0.066) 0.561
⇤⇤⇤
(0.063) 0.636
⇤⇤⇤
(0.064) 0.652
⇤⇤⇤
(0.064)
Ethnic Fractionalization 122.617
⇤⇤⇤
(0.297) 114.635
⇤⇤⇤
(0.298) 24.630
⇤⇤⇤
(0.300) 68.054
⇤⇤⇤
(0.340)
Ethnic Fractionalization
2
16.928
⇤⇤⇤
(0.375) 2.943
⇤⇤⇤
(0.379) 97.921
⇤⇤⇤
(0.382) 189.118
⇤⇤⇤
(0.462)
Violent unrest 0.644
⇤⇤⇤
(0.227) 0.547
⇤⇤
(0.224) 0.074 (0.226) 0.410 (0.256)
Nonviolent unrest 0.656
⇤⇤
(0.305) 0.569
⇤ (0.309) 0.290 (0.320) 0.038 (0.337)
Rival CW 1.760
⇤⇤⇤
(0.057) 1.408
⇤⇤⇤
(0.067) 1.440
⇤⇤⇤
(0.071) 1.095
⇤⇤⇤
(0.074)
Rival NW 2.874
⇤⇤⇤
(0.170) 13.420
⇤⇤⇤
(0.184) 12.401
⇤⇤⇤
(0.201) 12.600
⇤⇤⇤
(0.203)
Rival BW 1.092
⇤⇤⇤
(0.183) 0.817
⇤⇤⇤
(0.190) 1.422
⇤⇤⇤
(0.187) 1.041
⇤⇤⇤
(0.198)
Rival CW use 1.027
⇤⇤⇤
(0.058) 0.170
⇤⇤
(0.069) 0.388
⇤⇤⇤
(0.077) 0.296
⇤⇤⇤
(0.079)
Strategic rivalry 1.452
⇤⇤⇤
(0.196) 13.473
⇤⇤⇤
(0.221) 4.791
⇤⇤⇤
(0.224) 5.752
⇤⇤⇤
(0.231)
Nuclear infrastructure 1.700
⇤⇤⇤
(0.150) 2.762
⇤⇤⇤
(0.152)
log(GDPpc) 17.472
⇤⇤⇤
(0.990) 22.399
⇤⇤⇤
(1.092)
Dispute involvement 0.038 (0.086)
Nuclear ally 0.123 (0.186) 0.680
⇤⇤⇤
(0.199)
Land borders 0.710
⇤⇤⇤
(0.029) 1.340
⇤⇤⇤
(0.036)
CWC membership 12.994 (91.389) 16.045 (366.450)
Polity 0.452
⇤⇤⇤
(0.027)
Economic openness 0.057
⇤⇤⇤
(0.004)
Rebel experience 0.410
⇤ (0.242)
Neopatrimonialism 0.237
⇤ (0.141)
Personalism 0.370
⇤ (0.200)
Observations 4,274 8,855 4,274 4,274 3,250
R
2
0.207 0.114 0.219 0.233 0.270
Max. Possible R
2
0.392 0.266 0.392 0.392 0.436
Log Likelihood 568.336 833.141 534.309 497.495 420.711
LR Test 989.864
⇤⇤⇤
1,074.682
⇤⇤⇤
1,057.918
⇤⇤⇤
1,131.546
⇤⇤⇤
1,020.770
⇤⇤⇤
(df = 203) (df = 203) (df = 208) (df = 213) (df = 219)
Note:
⇤ p<0.1;
⇤⇤
p<0.05;
⇤⇤⇤
p<0.01
136
Table A.2: Security threats and CW exploration: fixed e↵ ects models
Dependent variable:
Risk of Chemical Weapons Pursuit (expansive)
(1) (2) (3) (4) (5)
Coup risk 1.206
⇤⇤⇤
(0.050) 1.169
⇤⇤⇤
(0.052) 0.448
⇤⇤⇤
(0.055) 0.631
⇤⇤⇤
(0.061)
Ethnic Fractionalization 187.526
⇤⇤⇤
(0.265) 218.352
⇤⇤⇤
(0.265) 205.220
⇤⇤⇤
(0.278) 296.389
⇤⇤⇤
(0.304)
Ethnic Fractionalization
2
120.649
⇤⇤⇤
(0.310) 169.539
⇤⇤⇤
(0.309) 182.615
⇤⇤⇤
(0.329) 307.850
⇤⇤⇤
(0.349)
Violent unrest 0.274
⇤ (0.144) 0.272
⇤ (0.148) 0.716
⇤⇤⇤
(0.152) 0.963
⇤⇤⇤
(0.159)
Nonviolent unrest 0.156 (0.226) 0.152 (0.228) 0.166 (0.227) 0.524
⇤⇤
(0.247)
Rival CW 1.609
⇤⇤⇤
(0.121) 4.566
⇤⇤⇤
(0.107) 4.792
⇤⇤⇤
(0.106) 2.679
⇤⇤⇤
(0.122)
Rival NW 1.093
⇤⇤⇤
(0.147) 1.701
⇤⇤⇤
(0.145) 1.406
⇤⇤⇤
(0.145) 0.146 (0.164)
Rival BW 1.058
⇤⇤⇤
(0.211) 2.698
⇤⇤⇤
(0.182) 3.275
⇤⇤⇤
(0.183) 1.375
⇤⇤⇤
(0.203)
Rival CW use 0.967
⇤⇤⇤
(0.121) 2.533
⇤⇤⇤
(0.107) 2.439
⇤⇤⇤
(0.107) 0.044 (0.116)
Strategic rivalry 0.081 (0.131) 2.565
⇤⇤⇤
(0.162) 1.844
⇤⇤⇤
(0.164) 2.606
⇤⇤⇤
(0.176)
Nuclear infrastructure 2.164
⇤⇤⇤
(0.087) 3.156
⇤⇤⇤
(0.103)
log(GDPpc) 17.125
⇤⇤⇤
(0.593) 23.689
⇤⇤⇤
(0.720)
Dispute involvement 0.667
⇤⇤⇤
(0.092)
Nuclear ally 1.655
⇤⇤⇤
(0.151) 0.760
⇤⇤⇤
(0.171)
Land borders 0.518
⇤⇤⇤
(0.029) 1.226
⇤⇤⇤
(0.033)
CWC membership 3.416
⇤⇤⇤
(0.278) 3.777
⇤⇤⇤
(0.321)
Polity 0.050
⇤⇤⇤
(0.011)
Economic openness 0.023
⇤⇤⇤
(0.003)
Rebel experience 0.719
⇤⇤⇤
(0.151)
Neopatrimonialism 0.180
⇤⇤⇤
(0.063)
Personalism 0.182 (0.154)
Observations 4,442 6,876 4,442 4,442 3,255
R
2
0.306 0.094 0.315 0.342 0.396
Max. Possible R
2
0.573 0.495 0.573 0.573 0.607
Log Likelihood 1,075.391 2,008.786 1,046.279 957.050 698.383
LR Test 1,624.933
⇤⇤⇤
678.187
⇤⇤⇤
1,683.156
⇤⇤⇤
1,861.615
⇤⇤⇤
1,642.658
⇤⇤⇤
(df = 190) (df = 190) (df = 195) (df = 200) (df = 206)
Note:
⇤ p<0.1;
⇤⇤
p<0.05;
⇤⇤⇤
p<0.01
137
Table A.3: Security threats and CW pursuit
Dependent variable:
Risk of Chemical Weapons Pursuit (restrictive)
(1) (2) (3) (4) (5)
Coup risk 0.103 (0.087) 0.250
⇤⇤⇤
(0.091) 0.361
⇤⇤⇤
(0.081) 0.376
⇤⇤⇤
(0.087)
Ethnic Fractionalization 10.451
⇤⇤⇤
(1.282) 8.995
⇤⇤⇤
(1.265) 9.695
⇤⇤⇤
(0.294) 9.813
⇤⇤⇤
(0.334)
Ethnic Fractionalization
2
9.232
⇤⇤⇤
(1.530) 7.239
⇤⇤⇤
(1.496) 8.632
⇤⇤⇤
(0.365) 8.145
⇤⇤⇤
(0.432)
Violent unrest 0.589
⇤⇤
(0.236) 0.377 (0.244) 0.239 (0.218) 0.235 (0.236)
Nonviolent unrest 0.658
⇤⇤
(0.277) 0.923
⇤⇤⇤
(0.284) 0.322 (0.299) 0.172 (0.319)
Rival CW 0.215 (0.227) 0.036 (0.201) 0.172
⇤⇤⇤
(0.054) 0.294
⇤⇤⇤
(0.063)
Rival NW 0.529
⇤⇤
(0.267) 0.414 (0.267) 0.097 (0.173) 0.894
⇤⇤⇤
(0.189)
Rival BW 0.686
⇤⇤⇤
(0.266) 1.436
⇤⇤⇤
(0.252) 1.480
⇤⇤⇤
(0.177) 0.592
⇤⇤⇤
(0.193)
Rival CW use 0.846
⇤⇤⇤
(0.238) 0.209 (0.214) 0.049 (0.059) 0.062 (0.073)
Strategic rivalry 1.628
⇤⇤⇤
(0.212) 1.572
⇤⇤⇤
(0.241) 1.541
⇤⇤⇤
(0.218) 0.808
⇤⇤⇤
(0.223)
Nuclear infrastructure 0.350
⇤⇤⇤
(0.111) 0.595
⇤⇤⇤
(0.145)
log(GDPpc) 2.610
⇤⇤⇤
(0.807) 5.906
⇤⇤⇤
(1.105)
Dispute involvement 0.187
⇤⇤⇤
(0.064)
Nuclear ally 0.066 (0.180) 0.252 (0.193)
Land borders 0.042 (0.029) 0.066
⇤ (0.037)
CWC membership 18.267 (803.849) 15.500 (409.099)
Polity 0.246
⇤⇤⇤
(0.026)
Economic openness 0.004 (0.003)
Rebel experience 1.019
⇤⇤⇤
(0.220)
Neopatrimonialism 0.522
⇤⇤⇤
(0.122)
Personalism 0.914
⇤⇤⇤
(0.197)
Observations 4,968 8,833 4,968 4,674 3,307
R
2
0.027 0.038 0.073 0.098 0.152
Max. Possible R
2
0.366 0.267 0.366 0.376 0.431
Log Likelihood 1,064.078 1,200.210 941.241 861.196 658.267
LR Test 133.467
⇤⇤⇤
(df = 5) 339.727
⇤⇤⇤
(df = 5) 379.143
⇤⇤⇤
(df = 10) 482.165
⇤⇤⇤
(df = 15) 546.698
⇤⇤⇤
(df = 21)
Note:
⇤ p<0.1;
⇤⇤
p<0.05;
⇤⇤⇤
p<0.01
138
Table A.4: Security threats and CW exploration
Dependent variable:
Risk of Chemical Weapons Pursuit (expansive)
(1) (2) (3) (4) (5)
Coup risk 0.340
⇤⇤⇤
(0.055) 0.414
⇤⇤⇤
(0.063) 0.402
⇤⇤⇤
(0.084) 0.500
⇤⇤⇤
(0.100)
Ethnic Fractionalization 6.001
⇤⇤⇤
(0.952) 6.636
⇤⇤⇤
(1.007) 6.439
⇤⇤⇤
(1.042) 4.752
⇤⇤⇤
(1.116)
Ethnic Fractionalization
2
4.928
⇤⇤⇤
(1.075) 5.206
⇤⇤⇤
(1.127) 5.947
⇤⇤⇤
(1.209) 3.958
⇤⇤⇤
(1.314)
Violent unrest 0.984
⇤⇤⇤
(0.144) 0.987
⇤⇤⇤
(0.150) 0.487
⇤⇤⇤
(0.165) 0.893
⇤⇤⇤
(0.200)
Nonviolent unrest 0.628
⇤⇤⇤
(0.226) 0.718
⇤⇤⇤
(0.230) 0.378 (0.238) 0.186 (0.267)
Rival CW 0.051 (0.180) 0.198 (0.173) 0.434
⇤⇤
(0.174) 0.445
⇤ (0.240)
Rival NW 1.180
⇤⇤⇤
(0.179) 1.248
⇤⇤⇤
(0.194) 1.082
⇤⇤⇤
(0.202) 0.945
⇤⇤⇤
(0.248)
Rival BW 0.354
⇤ (0.181) 0.488
⇤⇤
(0.205) 0.180 (0.210) 0.076 (0.270)
Rival CW use 0.003 (0.182) 0.377
⇤⇤
(0.177) 0.572
⇤⇤⇤
(0.184) 0.062 (0.239)
Strategic rivalry 1.581
⇤⇤⇤
(0.144) 1.739
⇤⇤⇤
(0.169) 1.337
⇤⇤⇤
(0.178) 0.749
⇤⇤⇤
(0.209)
Nuclear infrastructure 0.754
⇤⇤⇤
(0.128) 0.400
⇤⇤
(0.158)
log(GDPpc) 4.480
⇤⇤⇤
(1.007) 3.233
⇤⇤
(1.569)
Dispute involvement 0.911
⇤⇤⇤
(0.109)
Nuclear ally 1.430
⇤⇤⇤
(0.174) 1.204
⇤⇤⇤
(0.209)
Land borders 0.101
⇤⇤⇤
(0.034) 0.234
⇤⇤⇤
(0.042)
CWC membership 2.753
⇤⇤⇤
(0.236) 2.797
⇤⇤⇤
(0.278)
Polity 0.158
⇤⇤⇤
(0.029)
Economic openness 0.009
⇤⇤⇤
(0.003)
Rebel experience 0.319 (0.216)
Neopatrimonialism 0.265
⇤ (0.156)
Personalism 0.077 (0.190)
Observations 4,424 6,867 4,424 4,424 3,244
R
2
0.035 0.046 0.095 0.154 0.208
Max. Possible R
2
0.571 0.493 0.571 0.571 0.604
Log Likelihood 1,792.708 2,170.442 1,649.319 1,500.410 1,124.996
LR Test 155.498
⇤⇤⇤
(df = 5) 321.820
⇤⇤⇤
(df = 5) 442.278
⇤⇤⇤
(df = 10) 740.095
⇤⇤⇤
(df = 15) 756.747
⇤⇤⇤
(df = 21)
Note:
⇤ p<0.1;
⇤⇤
p<0.05;
⇤⇤⇤
p<0.01
139
Table A.5: Security Threats and BW pursuit: fixed e↵ ects models
Dependent variable:
Risk of Biological Weapons Pursuit (restrictive)
(1) (2) (3) (4) (5)
Coup risk 8.359
⇤⇤⇤
(0.133) 8.918
⇤⇤⇤
(0.133) 5.065
⇤⇤⇤
(0.141) 6.316
⇤⇤⇤
(0.148)
Ethnic Fractionalization 29.922
⇤⇤⇤
(0.297) 20.781
⇤⇤⇤
(0.289) 191.882
⇤⇤⇤
(0.292) 134.946
⇤⇤⇤
(0.409)
Ethnic Fractionalization
2
151.548
⇤⇤⇤
(0.382) 185.517
⇤⇤⇤
(0.366) 546.913
⇤⇤⇤
(0.369) 507.601
⇤⇤⇤
(0.595)
Violent unrest 0.826
⇤⇤⇤
(0.227) 0.307 (0.227) 0.259 (0.236) 3.407
⇤⇤⇤
(0.314)
Nonviolent unrest 0.871
⇤⇤⇤
(0.314) 0.929
⇤⇤⇤
(0.323) 1.079
⇤⇤⇤
(0.370) 0.796
⇤⇤
(0.364)
Rival CW 0.062 (0.043) 0.174
⇤⇤⇤
(0.061) 0.312
⇤⇤⇤
(0.068) 0.488
⇤⇤⇤
(0.078)
Rival NW 0.827
⇤⇤⇤
(0.096) 0.548
⇤⇤⇤
(0.121) 0.400
⇤⇤⇤
(0.131) 0.903
⇤⇤⇤
(0.140)
Rival BW 0.533
⇤⇤⇤
(0.098) 0.131 (0.116) 0.371
⇤⇤⇤
(0.117) 0.270
⇤⇤
(0.130)
Rival CW use 0.065 (0.049) 0.150
⇤⇤
(0.068) 0.120 (0.074) 0.586
⇤⇤⇤
(0.082)
Strategic rivalry 6.064
⇤⇤⇤
(0.193) 15.657
⇤⇤⇤
(0.458) 16.865
⇤⇤⇤
(0.463) 9.273
⇤⇤⇤
(0.465)
Nuclear infrastructure 2.257
⇤⇤⇤
(0.117) 3.379
⇤⇤⇤
(0.132)
log(GDPpc) 60.158
⇤⇤⇤
(0.749) 54.861
⇤⇤⇤
(0.771)
Dispute involvement 0.468
⇤⇤⇤
(0.040)
Nuclear ally 0.121 (0.164) 0.175 (0.186)
Land borders 0.155
⇤⇤⇤
(0.018) 0.280
⇤⇤⇤
(0.020)
CWC membership 1.271
⇤⇤⇤
(0.205) 1.144
⇤⇤⇤
(0.220)
Polity 0.188
⇤⇤⇤
(0.011)
Economic openness 0.025
⇤⇤⇤
(0.003)
Rebel experience 0.027 (0.239)
Neopatrimonialism 4.021
⇤⇤⇤
(0.073)
Personalism 0.217 (0.170)
Observations 5,199 9,234 5,199 4,897 3,510
R
2
0.272 0.192 0.279 0.296 0.332
Max. Possible R
2
0.432 0.398 0.432 0.435 0.475
Log Likelihood 643.673 1,356.826 619.472 538.545 420.681
LR Test 1,651.849
⇤⇤⇤
1,971.819
⇤⇤⇤
1,700.250
⇤⇤⇤
1,715.943
⇤⇤⇤
1,417.960
⇤⇤⇤
(df = 204) (df = 204) (df = 209) (df = 214) (df = 220)
Note:
⇤ p<0.1;
⇤⇤
p<0.05;
⇤⇤⇤
p<0.01
140
Table A.6: Security threats and BW exploration: fixed e↵ ects models
Dependent variable:
Risk of Biological Weapons Pursuit (expansive)
(1) (2) (3) (4) (5)
Coup risk 0.239
⇤⇤⇤
(0.084) 0.518
⇤⇤⇤
(0.103) 1.125
⇤⇤⇤
(0.104) 0.026 (0.094)
Ethnic Fractionalization 71.076
⇤⇤⇤
(0.401) 19.544
⇤⇤⇤
(0.334) 9.704
⇤⇤⇤
(0.324) 13.450
⇤⇤⇤
(0.360)
Ethnic Fractionalization
2
44.077
⇤⇤⇤
(0.390) 139.167
⇤⇤⇤
(0.372) 175.435
⇤⇤⇤
(0.365) 3.600
⇤⇤⇤
(0.412)
Violent unrest 1.053
⇤⇤⇤
(0.185) 0.549
⇤⇤
(0.223) 0.093 (0.224) 0.198 (0.258)
Nonviolent unrest 1.248
⇤⇤⇤
(0.265) 0.603 (0.382) 0.701
⇤ (0.382) 0.194 (0.396)
Rival CW 0.942
⇤⇤⇤
(0.054) 1.237
⇤⇤⇤
(0.060) 0.843
⇤⇤⇤
(0.059) 1.393
⇤⇤⇤
(0.078)
Rival NW 1.128
⇤⇤⇤
(0.150) 1.290
⇤⇤⇤
(0.162) 0.878
⇤⇤⇤
(0.163) 0.223 (0.194)
Rival BW 0.629
⇤⇤⇤
(0.139) 1.112
⇤⇤⇤
(0.145) 0.223 (0.142) 0.982
⇤⇤⇤
(0.214)
Rival CW use 0.177
⇤⇤⇤
(0.060) 0.280
⇤⇤⇤
(0.064) 0.613
⇤⇤⇤
(0.064) 0.050 (0.089)
Strategic rivalry 2.188
⇤⇤⇤
(0.194) 7.475
⇤⇤⇤
(0.253) 6.187
⇤⇤⇤
(0.251) 0.252 (0.186)
Nuclear infrastructure 3.267
⇤⇤⇤
(0.155) 0.010 (0.119)
log(GDPpc) 6.435
⇤⇤⇤
(1.108) 4.741
⇤⇤⇤
(0.789)
Dispute involvement 0.026 (0.075)
Nuclear ally 0.043 (0.176) 0.175 (0.182)
Land borders 0.696
⇤⇤⇤
(0.028) 0.082
⇤⇤⇤
(0.029)
CWC membership 0.438
⇤ (0.238) 0.069 (0.242)
Polity 0.049
⇤⇤⇤
(0.012)
Economic openness 0.004
⇤⇤
(0.002)
Rebel experience 0.096 (0.183)
Neopatrimonialism 0.053 (0.070)
Personalism 0.137 (0.189)
Observations 4,699 7,191 4,699 4,699 3,477
R
2
0.114 0.155 0.229 0.233 0.131
Max. Possible R
2
0.399 0.324 0.399 0.399 0.410
Log Likelihood 910.099 804.339 585.219 571.205 672.857
LR Test 570.331
⇤⇤⇤
1,209.353
⇤⇤⇤
1,220.090
⇤⇤⇤
1,248.119
⇤⇤⇤
486.945
⇤⇤⇤
(df = 195) (df = 195) (df = 200) (df = 205) (df = 211)
Note:
⇤ p<0.1;
⇤⇤
p<0.05;
⇤⇤⇤
p<0.01
141
Table A.7: Security Threats and BW pursuit
Dependent variable:
Risk of Biological Weapons Pursuit (restrictive)
(1) (2) (3) (4) (5)
Coup risk 0.643
⇤⇤⇤
(0.124) 0.399
⇤⇤⇤
(0.121) 0.644
⇤⇤⇤
(0.185) 0.447
⇤ (0.250)
Ethnic Fractionalization 9.506
⇤⇤⇤
(1.095) 7.177
⇤⇤⇤
(1.069) 11.010
⇤⇤⇤
(1.453) 10.558
⇤⇤⇤
(2.066)
Ethnic Fractionalization
2
8.398
⇤⇤⇤
(1.350) 5.806
⇤⇤⇤
(1.372) 10.639
⇤⇤⇤
(1.826) 9.976
⇤⇤⇤
(2.595)
Violent unrest 0.823
⇤⇤⇤
(0.239) 0.269 (0.265) 0.313 (0.274) 0.350 (0.382)
Nonviolent unrest 0.199 (0.302) 0.045 (0.310) 0.401 (0.346) 0.138 (0.374)
Rival CW 0.595
⇤⇤⇤
(0.144) 0.485
⇤⇤⇤
(0.147) 0.542
⇤⇤⇤
(0.177) 0.500
⇤⇤
(0.208)
Rival NW 0.324
⇤⇤
(0.155) 0.388
⇤⇤
(0.178) 0.334 (0.233) 0.878
⇤⇤
(0.404)
Rival BW 0.574
⇤⇤⇤
(0.142) 1.025
⇤⇤⇤
(0.150) 0.246 (0.169) 1.128
⇤⇤⇤
(0.250)
Rival CW use 0.068 (0.144) 0.043 (0.126) 0.125 (0.152) 0.402
⇤⇤
(0.198)
Strategic rivalry 2.225
⇤⇤⇤
(0.201) 3.236
⇤⇤⇤
(0.464) 2.608
⇤⇤⇤
(0.471) 1.585
⇤⇤⇤
(0.523)
Nuclear infrastructure 0.466
⇤⇤⇤
(0.181) 1.194
⇤⇤⇤
(0.274)
log(GDPpc) 3.876
⇤⇤⇤
(1.406) 4.944
⇤ (2.543)
Dispute involvement 0.486
⇤⇤⇤
(0.065)
Nuclear ally 0.554
⇤⇤⇤
(0.211) 0.890
⇤⇤⇤
(0.300)
Land borders 0.015 (0.038) 0.003 (0.051)
CWC membership 2.885
⇤⇤⇤
(0.233) 2.748
⇤⇤⇤
(0.289)
Polity 0.041 (0.065)
Economic openness 0.013
⇤⇤⇤
(0.004)
Rebel experience 1.387
⇤⇤⇤
(0.293)
Neopatrimonialism 0.219 (0.342)
Personalism 1.275
⇤⇤⇤
(0.334)
Observations 5,177 9,210 5,177 4,875 3,495
R
2
0.035 0.077 0.134 0.171 0.230
Max. Possible R
2
0.430 0.396 0.430 0.432 0.471
Log Likelihood 1,359.819 1,956.426 1,081.786 923.659 658.116
LR Test 186.185
⇤⇤⇤
(df = 5) 737.343
⇤⇤⇤
(df = 5) 742.250
⇤⇤⇤
(df = 10) 912.444
⇤⇤⇤
(df = 15) 911.308
⇤⇤⇤
(df = 21)
Note:
⇤ p<0.1;
⇤⇤
p<0.05;
⇤⇤⇤
p<0.01
142
Table A.8: Security threats and BW exploration
Dependent variable:
Risk of Biological Weapons Pursuit (expansive)
(1) (2) (3) (4) (5)
Coup risk 0.115 (0.095) 0.047 (0.092) 0.394
⇤⇤⇤
(0.124) 0.204 (0.181)
Ethnic Fractionalization 3.118
⇤⇤⇤
(1.197) 4.548
⇤⇤⇤
(1.281) 4.803
⇤⇤⇤
(1.281) 4.497
⇤⇤
(1.892)
Ethnic Fractionalization
2
1.889 (1.426) 3.464
⇤⇤
(1.483) 3.497
⇤⇤
(1.518) 0.871 (2.432)
Violent unrest 0.374 (0.229) 0.085 (0.242) 0.153 (0.246) 0.179 (0.340)
Nonviolent unrest 0.229 (0.365) 0.179 (0.374) 0.380 (0.388) 0.267 (0.412)
Rival CW 0.255 (0.176) 0.121 (0.158) 0.543
⇤⇤⇤
(0.188) 0.030 (0.264)
Rival NW 0.602
⇤⇤⇤
(0.170) 0.715
⇤⇤⇤
(0.181) 0.940
⇤⇤⇤
(0.203) 0.904
⇤⇤⇤
(0.310)
Rival BW 0.522
⇤⇤⇤
(0.194) 0.624
⇤⇤⇤
(0.197) 0.446
⇤⇤
(0.206) 0.302 (0.281)
Rival CW use 0.127 (0.180) 0.210 (0.158) 0.032 (0.173) 0.360 (0.270)
Strategic rivalry 1.246
⇤⇤⇤
(0.204) 1.815
⇤⇤⇤
(0.259) 1.881
⇤⇤⇤
(0.265) 1.244
⇤⇤⇤
(0.311)
Nuclear infrastructure 0.352
⇤⇤
(0.151) 0.306 (0.262)
log(GDPpc) 1.701 (1.114) 8.675
⇤⇤⇤
(2.060)
Dispute involvement 0.252
⇤⇤⇤
(0.090)
Nuclear ally 0.107 (0.192) 0.279 (0.273)
Land borders 0.100
⇤⇤⇤
(0.031) 0.107
⇤⇤⇤
(0.040)
CWC membership 0.800
⇤⇤⇤
(0.216) 0.500
⇤ (0.296)
Polity 0.268
⇤⇤⇤
(0.051)
Economic openness 0.004 (0.003)
Rebel experience 0.871
⇤⇤
(0.355)
Neopatrimonialism 0.282 (0.271)
Personalism 1.166
⇤⇤⇤
(0.277)
Observations 4,677 7,178 4,677 4,677 3,462
R
2
0.007 0.037 0.068 0.076 0.145
Max. Possible R
2
0.400 0.325 0.400 0.400 0.411
Log Likelihood 1,178.620 1,274.155 1,029.396 1,008.568 643.455
LR Test 30.808
⇤⇤⇤
(df = 5) 268.863
⇤⇤⇤
(df = 5) 329.256
⇤⇤⇤
(df = 10) 370.912
⇤⇤⇤
(df = 15) 543.697
⇤⇤⇤
(df = 21)
Note:
⇤ p<0.1;
⇤⇤
p<0.05;
⇤⇤⇤
p<0.01
143

Abstract (if available)

Abstract This dissertation identifies a set of pathway through which security threats influence chemical and biological weapons proliferation. In doing so, it moves beyond analytical frameworks that merely ask how chemical and biological weapons pursuit are related to nuclear weapons pursuit. Rather, it identifies a key factor — internal security threats faced by a state’s governing regime — that can both motivate chemical or biological weapons pursuit in states with little or no interest in nuclear weapons pursuit, and incentivize substitution or diversification of weapons programs in states that are interested in nuclear weapons. Such internal security threats are not widely incorporated into theories of arming, and highlight the unique role that chemical and biological weapons in can play in regimes’ security strategies.
A quantitative event history analysis, based on new panel data on CBW pursuit, acquisition, and use, suggests that internal security threats, while often disregarded in theories of arming, play a significant role in motivating states to pursue chemical and biological weapons proliferation. In particular, governing regimes facing increases in the risk of a coup may be more likely to initiate chemical and biological weapons programs, and regimes experiencing violent domestic unrest may be more likely to pursue chemical weapons. Additionally, this analysis offers tentative evidence for two pathways through which external security threats can motivate CBW proliferation. The first is “reactive proliferation”: states to pursue their own CBW capabilities in response to a rival’s acquisition of strategic weapons (in particular nuclear, and sometimes biological). The second is “substitution”: in the face of negative shocks to industrial and economic resources, states may begin pursue CBW as a hedge against the failure of a costlier, more technically challenging nuclear weapons program. Two case studies illustrate how the two different types of regime security threats shape chemical and biological weapons pursuit choices. In Libya, Qaddafi’s fear of being overthrown by opposing elements within the governing elite led him to dismantle much of the existing state bureaucracy. This “coup-proofed” state struggled to effectively monitor and manage a large, complex nuclear weapons program, making a smaller-scale chemical weapons pro- gram a desirable option to counter a range of security threats. In Rhodesia, an increasingly desperate struggle against a numerically asymmetrical insurgency prompted chemical and biological weapons pursuit, in the hopes inciting fear amongst the insurgents and the civilian population and generating casualties covertly and with relatively little manpower.
Understanding these mechanisms provides a necessary foundation for further inquiry into the causes and consequences of chemical and biological weapons proliferation. International relations scholars have built up a substantial body of knowledge regarding the causes and consequences of nuclear weapons pursuit. In comparison, the research on chemical and biological weapons proliferation has been less systematic and cohesive. Not only does this dissertation advance our understanding of CBW proliferation, it also provides potential insight into other tradeoffs made in states’ arming decisions, and into the substitutability of foreign policy tools more generally.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Pathways to nuclear weapon reversal: exploring mechanisms and understanding non-proliferation policy

PDF

The paradox of personalism: internal political constraints on nuclear posture in authoritarian regimes

PDF

When bullets are not enough: determinants of chemical weapons usage and abstention

PDF

Perceiving and coping with threat: explaining East Asian perceptions toward China’s rise

PDF

Flinging the boomerang: locating instability and the threat potential of identity-bias in US national security policy

PDF

Goldilocks’ signal for security cooperation in East Asia: China’s rise, hedging, and joint military exercises

PDF

Status, security, and socialization: explaining change in China's compliance in international institutions

PDF

The partial Good Samaritan states: China and Japan in the international relations of autocracy and democracy

Asset Metadata

Creator Barnum, Miriam Marina (author)

Core Title Lesser evils? WMD pursuit beyond nuclear weapons

School College of Letters, Arts and Sciences

Degree Doctor of Philosophy

Degree Program Political Science and International Relations

Degree Conferral Date 2022-08

Publication Date 08/01/2024

Defense Date 06/21/2022

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag biological weapons,chemical weapons,nuclear weapons,OAI-PMH Harvest,reactive proliferation,regime security

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor James, Patrick (committee chair), Hymans, Jacques (committee member), Lo, James (committee member), Meshkati, Najmedin (committee member)

Creator Email mbarnum@usc.edu,miriam.barnum@gmail.com

Permanent Link (DOI) https://doi.org/10.25549/usctheses-oUC111375871

Unique identifier UC111375871

Legacy Identifier etd-BarnumMiri-11057

Document Type Dissertation

Format application/pdf (imt)

Rights Barnum, Miriam Marina

Type texts

Source 20220801-usctheses-batch-965 (batch), University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the author, as the original true and official version of the work, but does not grant the reader permission to use the work if the desired use is covered by copyright. It is the author, as rights holder, who must provide use permission if such use is covered by copyright. The original signature page accompanying the original submission of the work to the USC Libraries is retained by the USC Libraries and a copy of it may be obtained by authorized requesters contacting the repository e-mail address given.

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA

Repository Email cisadmin@lib.usc.edu