80
	 5	Nuclear Latency and
Nuclear Proliferation
Scott D. Sagan
how quickly could individual governments, starting from different levels
of nuclear-related expertise and technology, develop a nuclear weapon if
they chose to do so? This question—which I will call the “nuclear latency”
question—is both exceedingly important and poorly understood. It is important
because an accurate understanding of both underlying state capabilities
and the time needed to utilize such capabilities is necessary to analyze a wide
set of nuclear policy issues: for example, dealing with the Iran nuclear crisis
(how quickly could Tehran make a weapon from its stockpile of low–enriched
uranium?); understanding the relationship between the spread of civilian
nuclear power and the spread of nuclear weapons capability (will new civilian
programs make breakout to military programs easier and more likely?);
evaluating potential NPT reforms (what would be the effects of lengthening
the ninety-day notice in the Article X withdrawal clause?); or assessing the stability
of a world without nuclear weapons (could disarmed states rearm in five
days, five weeks, five months, or five years?). Despite widespread discussion of
these policy issues, however, a set of mirror-image analytic failures has limited
our ability to make clear predictions about nuclear latency and proliferation:
Political scientists working on these subjects have often failed to examine basic
technical factors regarding the nuclear fuel cycle that strongly influence how
quickly states can get the bomb; the more technical literature about nuclear
latency has similarly often failed to examine the political factors that strongly
influence the ability of a government to develop nuclear weapons.
This chapter is both a conceptual minesweeping exercise and a modest
effort to suggest a better way forward. It has four main parts. First, I briefly
3978_SP_potter 1_Book.indb 80 4/27/10 2:17:25 PM
Nuclear Latency and Nuclear Proliferation  81
present examples of how some journalists, diplomats, and scholars have misunderstood
the nature of nuclear latency in their interpretations of IAEA reports,
leading to exaggerations concerning the number of states that currently
could build nuclear weapons in a short period of time. Second, I review and
critique the political science literature on “nuclear capability” and “nuclear
latency.” Even the most sophisticated political science studies on this subject
have too often used misleading measures of the key variables involved
in nuclear technology, focusing on broad measures of industrial capability
and nuclear research reactor experience and not on the specific fuel cycle
technologies and facilities needed to make the fissile materials required for a
nuclear weapon. This has led some political scientists, quite mistakenly in my
view, to denigrate the NPT regime, arguing that efforts to restrict the spread
of sensitive nuclear technology have failed in the past and that further restrictions
in the future are likely to be ineffective or even counterproductive.
Third, I review and critique leading examples from the technical literature on
nuclear latency. These studies have usefully focused on how long it has taken
individual states to develop highly enriched uranium (HEU) or weaponsgrade
plutonium but have unfortunately too often left out of their analyses the
political factors that accelerate or constrain such fissile material development.
These technical studies have also usefully included estimates of how long it
might take a state to develop one or more nuclear weapons once it has the
necessary fissile material and the political leadership has made a decision to
seek a nuclear bomb. However, they have ignored the strategic and domestic
factors that can influence the urgency of a leader’s demand for nuclear weapons
and the domestic political and organizational factors that can influence
whether a state bureaucracy can successfully implement a nuclear weapons
acquisition plan.
The conclusions offer suggestions for improved assessments of nuclear latency
in the future. Instead of chasing the quixotic goal of a single measure of
nuclear latency, scholars should seek to understand how political factors can
influence technological developments in this arena and how reaching various
thresholds in nuclear power technology can affect the politics of proliferation
decisions. Unfortunately, the field’s long-standing intellectual tradition
of dividing the proliferation puzzle into “supply-side” and “demand-side” factors
has reduced attention on the crucial and complex relationship between
the supply of nuclear technology and the demand for nuclear weapons. In the
conclusion, I therefore outline how one might better conceptualize and study
3978_SP_potter 1_Book.indb 81 4/27/10 2:17:26 PM
82  Scott D. Sagan
the nature of nuclear latency and its relationship to the acquisition of nuclear
weapons. In addition to sketching out an interdisciplinary research agenda,
the chapter ends with a discussion of how an improved understanding of nuclear
latency could influence policy debates regarding nuclear disarmament,
managing the fuel cycle, and the future of the NPT.
(Mis)quoting Mohamed
In his September 2004 address to the IAEA General Conference, Director
General Mohamed ElBaradei stated:
Some estimates indicate that 40 countries or more now have the know-how
to produce nuclear weapons, which means that if they have the required fissile
material—high enriched uranium or plutonium—we are relying primarily
on the continued good intentions of these countries, intentions which are
in turn based on their sense of security or insecurity, and could therefore be
subject to rapid change. Clearly, the margin of security this affords is thin, and
worrisome.1
The subtlety of ElBaradei’s argument and conditional nature of this
prediction—­“if they have the required fissile material”—were, however, often
ignored. A National Defense University study, for example, cited ElBaradei
when it claimed that there now exists “a high degree of nuclear latency that
challenges traditional thinking about nuclear threats”: “whereas 30 or 40 years
ago, only a handful of countries were assumed to know how to acquire nuclear
weapons, as many as 35 or 40 nations currently are believed to be in the
know.”2
The Austrian government switched “know-how” into “technical capability”
in its 2007 NPT Preparatory Committee statement: “Approximately
40 countries are said to have the technical capability to produce nuclear weap-
ons.”3
In some cases, the problem went beyond imprecise language to clearly
misleading claims, such as the 2005 pronouncement by Greenpeace: “Through
the IAEA’s worldwide support of nuclear power, 35–40 countries today have
the capability of building atomic weapons in several months, as Dr. ElBaradei
recently admitted.”4
A second commonly cited statement is ElBaradei’s October 2006 comment
about the potential for many “virtual nuclear weapons States” around
the globe:
Verifying enrichment facilities or reprocessing facilities is quite difficult and
the so-called conversion time is very short. So we are dealing with what I call
3978_SP_potter 1_Book.indb 82 4/27/10 2:17:26 PM
Nuclear Latency and Nuclear Proliferation  83
“virtual nuclear weapon States.” One of the issues I have been talking about for
a number of years is the need to develop a new international or multinational
approach to the fuel cycle so as to avoid ending up with not just nine nuclear
weapon States but another 20 or 30 States which have the capacity to develop
nuclear weapons in a very short span of time.5
Again, the conditional nature of this prediction—referring to a future world
if more states develop independent enrichment or reprocessing technology—
was widely ignored. For example, Zia Mian claimed that “Mohamed ElBaradei
of the [IAEA] warned that there are another 20 or 30 ‘virtual nuclear weapons
states’ that have the capacity to develop nuclear weapons in a very short time
span.”6
John Feffer, Marcus Raskin, and Kevin Martin similarly claimed that
there are “20–30 virtual nuclear weapons states, which Mohamed ElBaradei
of the . . . IAEA . . . warns have the capacity to develop nuclear weapons in a
very short time span.”7
Such statements by journalists, diplomats, and scholars clearly exaggerate
the number of states that could build nuclear weapons in a short time period.
But they also point to a deeper conceptual confusion about the nature of
nuclear latency. How could one measure it? What are the relationships among
acquiring research or power reactors, constructing uranium enrichment
or reprocessing facilities, developing fissile material, and acquiring nuclear
weapons? A logical place to start would be the political science literature on
nuclear latency. Unfortunately, such literature on the subject has not been very
helpful, in large part because it has failed to focus on the nuclear fuel cycle.
The Poverty of Political Science
on Nuclear Latency
This failure is due in part to the tendency in social science scholarship of the
past decade to focus on the “demand side” of the nuclear proliferation puzzle.
This emphasis on the motives of governments to develop or refrain from developing
nuclear weapons was understandable. Much of the earlier political
science literature had focused on the effects of arms control treaties, export
controls, and other technical constraints on the supply of nuclear materials
and technology, and an increase in attention to why some governments
wanted nuclear weapons, and why some governments did not, was clearly
needed.8
Moreover, the emergence of nuclear “proliferation rings,” such as
the A. Q. Khan network and other technology smuggling efforts, encouraged
fears that it would be increasingly difficult to prevent weapons proliferation
3978_SP_potter 1_Book.indb 83 4/27/10 2:17:26 PM
84  Scott D. Sagan
through “supply-side” constraints.9
Finally, as will be demonstrated shortly,
the quantitative literature on the “correlates of nuclear proliferation” began to
present surprising evidence on the rapid growth in the number of states that
were “nuclear weapons capable.”
Such considerations have led some prominent political science scholars to
be highly skeptical of efforts to control the supply of nuclear technology for
the purposes of nonproliferation. For example, in his innovative book The Psychology
of Nuclear Proliferation, Jacques Hymans argues that “given the widespread
diffusion of nuclear capacities, supply-side control measures against
potential proliferant states are clearly of declining utility”:
A stricter regime will likely do nothing to change proliferation intentions. It is
highly unlikely that more stringent controls will dissuade oppositional nationalist
leaders from seeking the bomb and in today’s freewheeling global market
they probably will be able to obtain the materials they need to build it, albeit
perhaps more slowly and with difficulty. More problematically, it is highly
unlikely that more stringent controls will dissuade sportsmanlike nationalist
leaders from resisting the nonproliferation regime. Indeed, the harsher the
regime becomes, the more likely that both types of nationalists will resent and
resist it . . . In short, the construction of ever-higher supply-side hurdles to
civilian nuclear development, far from “strengthening” the non-proliferation
regime, is in fact likely to leave the regime even weaker than it is today.10
Hymans admittedly does not want “to abandon the NPT regime,” although
he does “second-guess the continual urge to ‘strengthen’ it with ever-heavier
supply-side controls.”11
Yet Hymans is by no means alone among the political
scientists who have studied nuclear proliferation in criticizing continuing efforts
to control nuclear technology through the NPT. Dong-Joon Jo and Erik
Gartzke also find a strong positive correlation between a state getting “latent
nuclear weapons production capabilities” and the initiation of a nuclear
weapons program, and they maintain that “the inhibiting effect of the NPT
is overcome by the stronger technological diffusion effect,” concluding that
“enthusiasm for the NPT among proliferation opponents thus appears to be
misplaced.”12
Harsh Pant has been even more dismissive of the NPT, claiming
that the treaty “was never sustainable and has had little, if any, effect on the
pace of nuclear proliferation.”13
Such conclusions regarding the NPT are not warranted based on historical
evidence. The underlying assumption, that more and more states have
3978_SP_potter 1_Book.indb 84 4/27/10 2:17:26 PM
Nuclear Latency and Nuclear Proliferation  85
become “nuclear weapons capable,” depends crucially on how one defines and
measures “nuclear latency.” An important weakness, however, in the ways in
which political scientists have measured nuclear latency becomes clear only
after one delves deeply into the methodology used in their studies.
What does it mean for a state to be “capable” of producing nuclear weapons?
Hymans’s assessment of what he calls “latent nuclear capabilities” is based on
the methodology and data set used in Stephen Meyer’s pioneering 1984 book
The Dynamics of Nuclear Proliferation and Richard Stoll’s update of the Meyer
data through 1992. Meyer’s study carefully measured a set of ten technical and
economic indicators—national mining activity, indigenous uranium deposits,
metallurgists, steel production, construction work force, chemical engineers,
nitric acid production, electrical production capacity, nuclear engineers,
physicists, chemists, and explosive and electronics specialists—to produce
what he called “a list of nations with latent capabilities to manufacture nuclear
weapons.”14
Not being able to measure directly whether the quantity or quality
of a state’s nuclear engineers and its explosive and electronic specialists
were sufficient to build a nuclear weapon, Meyer used two proxy indicators:
whether the state had been operating a research reactor for three years (the
proxy for nuclear engineering expertise) and whether the state manufactured
automobiles or assembled automobiles and manufactured radios and television
sets (the proxy for explosive and electronics specialists). Based on this
particular set of indicators, Meyer found that thirty-four states held the latent
capability to build nuclear weapons in 1982.15
Stoll updated the Meyer data
set, but with a hidden yet significant change in coding rules, in the mid-1990s
(see Fig­ure 5.1): While Meyer measured indigenous uranium sources, Stoll
assumed that all states had access to nuclear materials, arguing that they were
now freely available in the global marketplace. Stoll thus simply assumed away
the crucial issue of whether a state had access to uranium that, once enriched,
could be used in a nuclear weapons program. Based on the resulting data set,
Stoll argued that forty-eight states had a latent nuclear weapons capability in
1992, noting that:
A country is said to have a latent capacity when it has sufficient technical,
in­dustrial, material, and financial resources to support a wholly indigenous
weapons program. Even though a state may have a latent capacity, it must still
make an explicit decision to develop the particular facilities necessary to create
weapons. However, once a state has a latent capacity, it is very difficult—perhaps
impossible—to deny it nuclear weapons, since it is in essence self-sufficient.16
3978_SP_potter 1_Book.indb 85 4/27/10 2:17:26 PM
86  Scott D. Sagan
Hymans accepted the logic of that argument and claims that the Stoll data
demonstrate that there was a “yawning gap between technical potential and
military reality” in terms of the number of states that have capability to produce
nuclear weapons compared to those that have actually done so.17
(I depict
the Stoll data in Figure 5.1.)
A state can not make a nuclear weapon, however, unless it has HEU or
plutonium from a large reactor, and Stoll’s hidden assumption that any state
could acquire uranium on the open market, coupled with his use of research
reactor experience as the measure of required nuclear engineering expertise,
essentially assumed away those two technical constraints. Moreover, looking
at his data set reveals that even though Stoll argued that each of his ten criteria
were “necessary conditions for the production of nuclear weapons,” North
Korea, which the IAEA discovered had taken spent fuel rods containing
50
45
40
35
30
25
20
15
10
5
0
Nuclear-capable
states
States with nuclear
weapons
1 2 3 4 5 6 7 8 9
Numberofstates
Year
Figure 5.1.  Stoll’s nuclear latency estimate.
Source: Data from Richard Stoll, “Latency Capacity Proliferation Model”; available at
http://es.rice.edu/projects/Poli378/Nuclear/Proliferation/
3978_SP_potter 1_Book.indb 86 4/27/10 2:17:27 PM
Nuclear Latency and Nuclear Proliferation  87
plutonium from the Yongbyon reactor in 1989, 1990, and 1991,18
was not considered
capable of building a nuclear weapon according to Stoll’s model in 1992.
North Korea lacked the necessary nitric acid production capability, chemical
engineers, and electronic/explosives specialists (as measured by domestic
automobile and radio/television industry). This observation obviously raises
questions about whether these particular measures of nuclear latency really are
“necessary conditions” for a state to develop nuclear weapons.
Jo and Gartzke’s 2007 Journal of Conflict Resolution study “The Determinants
of Nuclear Weapons Proliferation” improves on the Stoll coding scheme
by dropping three of the Stoll and Meyer indicators (construction workforce,
steel production, and previous national mining activity) on the grounds that
they are “too easily available to be thresholds” and modifying the coding for
necessary “uranium deposits” (which, as we have seen, were assumed to be
Table 5.1. Comparison of Meyer/Stoll and Gartzke and Jo
nuclear capability indicators.
Definition of indicator
Indicator Meyer/Stoll Gartzke and Jo
Mining activity Some fraction of labor
force in mining activity
————­——
Uranium deposits Known uranium deposits
(Meyer)
Assumed market access
(Stoll)
Known uranium deposits or
produced uranium already
Metallurgists Production of crude steel Production of crude steel or
aluminum
Steel Production of crude steel ————­——
Construction work force Production of steel and
cement
————­——
Chemical engineers Production of nitric acid
or sulfuric acid
Production of nitric acid or
sulfuric acid
Nitric acid production
capacity
Nitric acid production or
sulfuric acid production
and nonorganic nitrogenous
fertilizer production
Nitric acid production or
sulfuric acid production
and nonorganic nitrogenous
fertilizer production
Electricity production
capacity
Installed electrical
capacity of 200MWe
Installed capacity of 200MWe
or produces equivalent of
50,000 metric tons of oil
Nuclear engineers/
physicists/chemists
Three research-reactor
years
Three research-reactor years
Electronics/explosives
specialists
Manufacture of motor vehicles
or assembly of motor
vehicles and manufacture
of radios or TVs
Manufacture or assembly of
motor vehicles and manufacture
of radios or TVs
3978_SP_potter 1_Book.indb 87 4/27/10 2:17:27 PM
88  Scott D. Sagan
available for all states by Stoll) to include either a state with uranium deposits
on its territory or one that has acquired “produced uranium” for a research
or power reactor.19
(See Table 5.1 for a comparison of the various coding
schemes.)
Yet, although Jo and Gartzke correctly note that “states that lack the basic
material capabilities will be excluded from the group of potential proliferators,”
their model actually does not treat nuclear materials (enriched uranium or
plutonium) as a necessary, but not sufficient, condition for building nuclear
weapons.20
Instead, their model implicitly assumes that “where there is a will,
there is a way” and that a latent nuclear weapons state, or even a state that has
already made nuclear weapons, may not actually have the necessary materials.
That the Jo and Gartzke model does not therefore adequately capture the
necessary conditions for nuclear weapons development can be best seen in
the representation of their data set for 2001 presented as a map in Figure 5.2.21
The Jo and Gartzke data set continues to show that North Korea did not have
a full latent capability to develop nuclear weapons in 2001 (it still lacked sufficient
chemical engineers, nitric acid production capability, and explosives
specialists), even though the North Koreans were a major exporter of longrange
missiles at the time and were known to have separated plutonium from
the fuel rods of the Yongbyon reactor.22
We now also know that North Korea
tested its first nuclear weapon in 2006. The Jo and Gartzke coding rules also
lead to the odd conclusion that South Africa, which built six nuclear weapons
in the 1980s, dismantled the weapons in the 1990s but still maintained from
450 to 600 kg of HEU under IAEA safeguard inspections in 2002, nevertheless
lacked the full capability to build nuclear weapons.23
(South Africa lacked
sufficient chemical engineers and nitric acid production capacity, according
to their data.)
By not focusing attention on enriched uranium and plutonium, the weak
proxy measures of nuclear latency used in this work clearly lead to bizarre results.
Trinidad and Tobago (which “only” lacks uranium deposits, “produced
uranium,” and any research reactor experience) is assessed to have a higher
degree of nuclear weapons latency in 2001 than is North Korea, only five years
away from detonating its first nuclear weapon. In Africa, Egypt, which had
only two research reactors in 2001, is assessed to have a higher degree of nuclear
weapons latency than is South Africa.
In addition, Jo and Gartzke usefully try to assess the nuclear diffusion
effect in an attempt to understand whether “knowledge of how to construct
nuclear weapons has spread with the passage of time” and how much this has
3978_SP_potter 1_Book.indb 88 4/27/10 2:17:28 PM
Nuclear Latency and Nuclear Proliferation  89
influenced proliferation.24
Their measure of “nuclear diffusion,” however, is
devoid of substantive content. Because Jo and Gartzke had no direct measure
of the spread of nuclear knowledge over time, they simply assumed that “diffusion
equals the log transformation of the number of years since 1938.”25
This
variable may measure some temporal factor that affects changes in proliferation
behavior over time but does not, contrary to their claim, “suggest that the
NPT may actually contribute to the quickening pace of nuclear diffusion.”26
In conclusion, these problems point to serious weaknesses in the political
science literature analyzing nuclear latency and the effects of the NPT. All
too often this literature measures “proxy variables” that are easily available,
rather than collecting the data that reflect the variables of real interest. All
too often there are hidden, but crucial, assumptions that have a strong impact
on the findings in ways that are not acknowledged. Most importantly, the political
science literature, like the misinterpretations of ElBaradei’s statements
cited earlier, often conflates two analytically different phenomena under
the same labels of “nuclear weapons capability” or “nuclear latency”—first,
what should properly be called “nuclear self-sufficiency” (a measure of how
independent a potential long-term nuclear weapons program could be); and,
Key
7 (Full latency)
5–6
3–4
0–2
Figure 5.2.  Nuclear latency according to Jo and Gartzke.
Source: Data from Dong-Joon Jo and Erik Gartzke, “Dataset for ‘Determinants of Nuclear
Proliferation: A Quantitative Model,’ ” Journal of Conflict Resolution 51, no. 1 (2007), available at:
http://jcr.sagepub.com/cgi/content/full/51/1/167/DC1
3978_SP_potter 1_Book.indb 89 4/27/10 2:17:30 PM
90  Scott D. Sagan
second, “nuclear latency” (a measure of how quickly a state could develop a
nuclear weapon if it chose to do so from its current state of technological de-
velopment).27
The former may have value in estimating whether a state could
develop nuclear weapons eventually, over an extended period of time, even if
it received no technical assistance or nuclear materials from other states. This
is a phenomenon, however, that has never happened in world history, as all
nuclear weapons states have received some degree of assistance from others, as
Itty Abraham has compellingly demonstrated: Even the first nuclear weapons
states, the United States and the Soviet Union, were not self-sufficient, having
received assistance from German scientists, scientists from closely allied nations,
and spies from other states’ programs.28
In short, by focusing our attention away from the acquisition of the fissile
materials needed to make a nuclear weapon, the political science literature has
led to an exaggerated estimate of how many states currently have the technical
capability to build nuclear weapons. The criticisms of the NPT regime cited
above—that it has been ineffective or even may have led to the widespread diffusion
of latent nuclear weapons capability—are therefore unwarranted. The
NPT regime clearly has flaws, and we know of many cases of states that have
violated their commitments, but the NPT has not (at least not yet) led to a
world in which there are dozens of non–nuclear weapons states that could
easily become nuclear weapons states in a short period of time.
Technology, the Fuel Cycle,
and Nuclear Weapons
The biggest problem in these political science studies was their failure to focus
on the fuel cycle technology necessary to produce the fissile materials needed
for a weapon. Without a large nuclear reactor to produce plutonium (and a
reprocessing facility) or the capability to produce enriched uranium, no state
could build its own nuclear weapon. Does a state have nuclear power reactors
or other reactors that produce plutonium? How large are its research reactors,
and do they run on low enriched uranium (LEU) or HEU? Does it have the
technological capability to produce, separate, or reprocess plutonium? Does
the state have an enrichment facility to produce HEU? How long would it
take to build such facilities and then to build a nuclear bomb with the materials?
Such questions should be at the core of any assessment of a state’s latent
nuclear weapons capability.
Fortunately, there is a strong technical literature on nuclear power and
proliferation that addresses such questions. Yet even the most thorough of
3978_SP_potter 1_Book.indb 90 4/27/10 2:17:30 PM
Nuclear Latency and Nuclear Proliferation  91
these studies ignore important political factors that strongly influence proliferation
decisions and their implementation. I will review three important
studies of nuclear technology to illustrate the point.
The pioneering study on this subject was the 1977 book Swords from Plowshares,
produced by an interdisciplinary team of scholars led by Albert Wohl-
stetter.29
Wohlstetter and his colleagues divided non–nuclear weapons states
into three categories—states with advanced infrastructure and fissile material,
states with a research or power reactor, and states with no nuclear experience
at all. In the first category were the nine states that were estimated to have “full
access to the fissile material required to make a weapon” in 1977: Japan, West
Germany, South Africa, Belgium, Taiwan, Italy, the Netherlands, Canada, and
Sweden. The study argued that each of these states could take the following
four remaining steps to produce a nuclear weapon within one year, assuming
the steps were undertaken in parallel: converting the fissile material in their
possession into metallic form, designing a weapon, fabricating the weapon
and its components, and preparing for and conducting a nuclear test.30
In the
second category, Swords from Plowshares listed fifteen states that had a reactor
in 1977 but no access to fissile material outside of the spent fuel rods: Israel,
Argentina, Switzerland, Egypt, Spain, South Korea, Indonesia, East Germany,
Czechoslovakia, Australia, Pakistan, Iran, Norway, Brazil, and Mexico. The
study assumed that the most practical pathway for such states to get a nuclear
weapon would be to construct a reprocessing plant, which the study estimated
would take four years. It assumed that the four final weaponization steps listed
above could be done in parallel to the reprocessing efforts (though the authors
never explained how a state could produce plutonium metal before the reprocessing
plant was completed and in operation) and thus estimated that any
of these states could make a nuclear weapon in four years.31
Finally, states in
the third category were estimated to need at least six years to build a nuclear
reactor and simple reprocessing facility “from scratch.” (See Table 5.2 for the
countries in each category.) Wohlstetter and his team acknowledged that these
timelines were “engineering estimates based on American experience” and
should therefore only be considered “rough approximations.”32
Swords from Plowshares was an influential and prescient study. Wohlstetter
and his colleagues identified the danger of what they called “nearing the bomb
without breaking promises not to make it” under the NPT.33
They discussed
the need for a more extensive system of IAEA inspections of non–nuclear
weapons states’ facilities and produced one of the first analyses of the pros
and cons of building multinational nuclear fuel facilities. With the passage
3978_SP_potter 1_Book.indb 91 4/27/10 2:17:30 PM
92  Scott D. Sagan
of time, however, one can also see some obvious weaknesses in this pioneering
work. First, by basing the estimates for nuclear latency time lines on the
U.S. experience, the study implicitly assumes that all proliferators are likely
or able to follow the same technological pathways, with the same degree of
bureaucratic and organizational competence and the same degree of wartime
and Cold War urgency. Wohlstetter and his colleagues also assumed that once
a government made a decision to develop nuclear weapons, there would be a
high degree of political consensus behind it and the weapons program would
progress without significant internal or external constraints. The estimate on
the “starting from scratch” scenario seems particularly low and appears to
con­tain a hidden assumption that the world market would encourage such
growth in reprocessing facilities and that the United States and the global
nonproliferation regime would not add additional political hurdles to constrain
that growth.
To be fair, the alarm bell sounded by the Wohlstetter study was a kind of
self-denying prophecy because it was a major contributor to the development
of future nonproliferation innovations, at least in the United States.34
Swords
Table 5.2.  Wohlstetter et al.’s nuclear capability timeframes.
Capability level (NNWS only)
Advanced nuclear
infrastructure plus
“full access to fissile
material needed
for bomb”
Research or power
reactor; no fissile
material except in
reactors and spent
fuel rods
No nuclear
infrastructure
Countries qualifying
(as of 1977)
Japan, West Germany,
South Africa,
Belgium, Taiwan,
Italy, Netherlands,
Canada, Sweden (9)
Israel, Argentina,
Switzerland, Egypt,
Spain, Republic of
Korea, Indonesia,
East Germany,
Czechoslavakia,
Australia, Pakistan,
Iran, Norway, Brazil,
Mexico (15)
All other states
(Assumed) obstacles
remaining
(a) Convert fissile
material to metal;
(b) design weapon;
(c) fabricate
weapon; (d) test
(a) through (d)
plus concurrent
construction of
reprocessing plant
Build reactor and
reprocessing plant
Estimated time to
first bomb
1 year 4 years 6+ years
Source: Data from Albert Wohlstetter, Thomas A. Brown, Gregory Jones, David C. McGarvey,
Henry Rowen, Vince Taylor, and Roberta Wohlstetter, Swords from Plowshares: The Military Potential
of Civilian Nuclear Energy (Chicago: University of Chicago Press, 1977).
3978_SP_potter 1_Book.indb 92 4/27/10 2:17:31 PM
Nuclear Latency and Nuclear Proliferation  93
from Plowshares also contained a useful discussion of the need for more rigorous
agreements among governments that exported nuclear technology on
what could be sold and what could not be sold, even to NPT member states.
Still, this important 1977 technical study could have been supplemented at
the time (and certainly needs to be supplemented now) with both analyses
of alternative weapons development pathways and historical data on actual
political decision making and bureaucratic implementation in different states’
programs that could provide more accurate evidence to support estimates of
future latency timelines.
The second study focused on technology contributing to nuclear latency
is the 2006 Science and Global Security article by Robert Harney and his colleagues
entitled “Anatomy of a Project to Produce a First Nuclear Weapon.”35
The authors identify and provide a complex timeline of the 196 necessary
tasks required to produce a uranium-based nuclear weapon by a state that has
produced or acquired 120 metric tons of yellowcake. The tasks include production
of enrichment-plant feed material (UF6), uranium enrichment (with
timelines for different enrichment methods), production of HEU metal, and
finally the design and construction of actual nuclear weapons.36
Using measures
of time, labor, energy, and necessary money required to complete these
tasks un­der both normal and expedited (“crash”) conditions, the authors
provide an estimate of the earliest possible completion time: under normal
conditions, approximately 338 weeks (six and a half years) would be needed
to produce six weapons, and under crash conditions 260 weeks, or just under
five years, would be required.37
Harney and his coauthors also provide an estimate
of an Iranian withdrawal from the NPT breakout scenario: According
to their model, it would still take 216 weeks (about four years) for a state that
has a prototypic uranium enrichment plant already in place to produce its first
nuclear weapon.38
The Harney and colleagues study usefully highlights the complexity and
time needed to produce uranium metal from highly enriched uranium and
for subsequent steps in the weaponization process. By developing estimates
of both crash programs and noncrash programs, the study brings one political
variable into its analysis. But there are nonetheless limits to its utility as
a guide to understanding nuclear latency in future proliferation scenarios.
First, the authors, like those contributing to the Wohlstetter study, base their
estimates on the U.S. experience with nuclear materials and weapons produc-
tion.39
This could bias the results in both directions, though it is difficult to
3978_SP_potter 1_Book.indb 93 4/27/10 2:17:31 PM
94  Scott D. Sagan
know which bias would be stronger. On the one hand, the study assumes a
high degree of organizational competence on the part of future proliferators,
which would likely make the Harney estimates lower than is likely to occur
in most new proliferant states in developing world. On the other hand, the
study also assumes that future proliferators would copy the later U.S. Cold
War penchant for careful testing, stringent safety, and high reliability, tasks
that take time and would make the estimates derived from the model much
too long. (For example, the study estimates it would take twenty-four weeks
to verify gun velocity, twenty-four weeks to build a delivery vehicle compatibility
mock-up, and forty weeks to finalize the weapons design.40
) From a
U.S. weapons designer’s view, such assumptions may reflect normal peacetime
procedures; from a proliferation prevention perspective, they may constitute
a wishful thinking, best-case analysis. For example, if a new proliferant government
decides to use a simple uranium gun-type device or has access to the
more advanced nuclear bomb designs that were peddled by the A. Q. Khan
network, these time-consuming activities could be greatly reduced. The Harney
study also assumes that a proliferant government has not developed covert
facilities to jumpstart the weaponization process, having most essential
tasks completed before the HEU is produced.
The third, and most historically grounded, technical study on nuclear latency
is a detailed 2005 report from the Pacific Northwest National Laboratory
(PNNL): Nuclear Proliferation Technology Trends Analysis.41
The PNNL
study does not base its estimates on the U.S. experience with different nuclearrelated
technologies but rather gathers and presents the available data on the
experience and time lags seen in the history of many different states’ uranium
enrichment and plutonium production and reprocessing programs. The study
is the most detailed analysis, at least in the unclassified literature, of the history
of successful efforts either to enrich uranium or reprocess plutonium. The
PNNL authors wisely do not “sample on the dependent variable” and therefore
also analyze failed or abandoned programs to enrich uranium or reprocess
plutonium. I reproduce the concluding estimates about nuclear materials
production time lines from the PNNL study in Table 5.3.
This PNNL study is valuable, but its claims to be able to use history to
provide accurate estimates about technical time constraints that continue to
apply today are problematic. The study correctly notes that “the time required
for success varies widely and is strongly dependent on either help from nations
that have already developed the technology or the nuclear and industrial
maturity of the nation.”42
But the authors do not analyze the evolution of
3978_SP_potter 1_Book.indb 94 4/27/10 2:17:31 PM
Nuclear Latency and Nuclear Proliferation  95
international export controls on nuclear technology, which have strengthened
over time and thus influence late developers more than early developers of
some nuclear materials production technologies. The authors maintain that
“based on an evaluation of historical trends in nuclear technology development,
conclusions can be reached concerning: 1) the length of time it takes to
acquire a technology; 2) the length of time it takes for production of special
nuclear material to begin; and 3) the type of approaches taken for acquiring
the technology.”43
History surely can provide an answer to point three, but
the other two questions are historically contingent on the characteristics of
the states involved and can change depending on the spread of other related
technologies, organizational learning, illicit networks of suppliers, and shifts
in foreign government assistance. The authors’ calculations of “average” times
to pilot plants and “average” times to production success are interesting but
have a peculiar ahistorical character to them; and, even if the full ranges of
historical time lines were presented, using the lowest number as a “worst-case
estimate” would be problematic because these time lines include advanced
industrial states as well as less-developed nations. In short, future scholars
Table 5.3.  PNNL nuclear capability timeframes.
Technology
Number of
countries
interested in
technology
Number of
countries with
successful
production
programs1
Average
time to
pilot plant2
Average
time to
production3
Gaseous diffusion enrichment 6 5 ———— 6 years
Centrifuge enrichment 18 7 8 years 14 years
Electromagnetic isotope separation 11 1 2 years 3 years
Chemical isotope separation 3 ———— 6 years 11 years
Aerodynamic isotope separation 3 1 7 years 18 years
Laser enrichment 14 ———— ———— ————
Graphite-moderated production reactors 6 6 1 year 2–11 years4
Heavy-water-moderated reactors 12 5 1 year 2–6 years
Research reactors 14 3 ———— 4–5 years
Reprocessing 19 13 6 years 10 years
Source: M. D. Zentner, G. L. Coles, and R. J. Talbott, Nuclear Proliferation Technology Trends Analysis, Pacific
Northwest National Laboratory, September 2005; available at www.pnl.gov/main/publications/external/technical
_reports/PNNL-14480.pdf
1
More than gram quantities of material produced [note in original].
2
Technological capability demonstrated [note in original].
3
Significant quantities of material produced [note in original].
4
Note that “average” times given in this cell and the two cells immediately below it are in fact ranges, not averages.
3978_SP_potter 1_Book.indb 95 4/27/10 2:17:31 PM
96  Scott D. Sagan
should be more cautious than were the PNNL authors when they concluded
that “the timeframes identified above can be considered representative of current
development efforts.”44
Toward a Political Theory of Nuclear Latency
and Proliferation
These considerations lead me to conclude that any general measure of “nuclear
latency” is likely to be a chimera. Nuclear latency is not like human pregnancy,
in which all women have virtually the same nine month gestation period.
Different non–nuclear weapons states, even those starting from the same
technological threshold, are likely to take different lengths of time to move to
possession of a single nuclear weapon or a usable arsenal.
Where should the political science and technical community go from here
to improve our understanding of nuclear latency and the risks of nuclear proliferation
in the future? And how would such research help us understand how
the NPT works to constrain proliferation? This conclusion outlines a multidisciplinary
research agenda and suggests some principles that should guide
future research to avoid some of the weaknesses of past efforts in this area.
I then conclude with some policy-relevant observations about how new agreements
regarding the fuel cycle facilities and produced fuels could strengthen
the NPT regime.
First, we need more research on the domestic characteristics of the regimes
that have been successful and of those that have been less so in their attempts
to develop uranium enrichment and plutonium reprocessing capabilities. Such
analysis would have to take into account the degree to which the “failures” or
longer time lines for successful states were due to internal characteristics or
external constraints. The simple division between democracies and nondemocracies
may be less helpful in this regard than a focus on the relationship
among the political leadership and the scientific community and military.
Jacques Hymans, for example, usefully theorizes that “neopatrimonial” or
“sultanistic” regimes—governments characterized by extreme personalized
rule, use of state resources to buy off clients, and an absence of checks and
balances—will take longer to develop advanced levels of nuclear technology
and will fail more often in attempts to move from one technological threshold
to the next.45
Hymans argues that North Korea fits this model and compares it
to Romania, where an unsuccessful program was run by Elena Ceauşescu, the
president’s wife, who hired scientists based primarily on whether they would
3978_SP_potter 1_Book.indb 96 4/27/10 2:17:32 PM
Nuclear Latency and Nuclear Proliferation  97
promote her candidacy for the Nobel Prize in chemistry; and Libya, which
was described by the 2005 WMD Commission as “an inept bungler, the court
jester among the band of nations seeking biological and nuclear capabilities.”46
This is a promising approach, yet it is important to note that North Korea,
unlike Romania and Libya, was able to produce fissile materials despite international
sanctions. North Korea tested a nuclear device in 2006, though this
device was not as effective as was apparently expected, and another in 2009,
widely considered to have been successful.47
In addition, it is worth noting that
Saddam Hussein’s regime in Iraq was characterized by massive corruption, a
culture of fear that led to exaggeration of progress by laboratory officials and
military commanders alike, and a decision-making style that, according to his
senior colleagues, “verged on the mystical.”48
Despite such pathological decision
making and leadership, however, the Iraqi covert program was discovered
after the 1991 Gulf War to be much closer to producing a nuclear weapon
than the CIA had estimated.49
Another potential political constraint on nuclear weapons programs can
be the rivalries for power between different leaders in potential proliferators.
In Egypt in the 1960s, for example, Gamal Abdel Nasser started a nuclear
weapons program but did not give it high priority or a large budget, in part because
the head of the nuclear program was a strong ally of Nasser’s chief rival,
Abdel Hakim Amer. As one former military officer later explained: “We didn’t
want to create heroes in the system that a nuclear bomb would create.”50
In short, more research on how regime characteristics influence the ability
to develop both fissile materials and nuclear weapons should be pursued.
We have a strong literature on how regime type can influence decisions about
whether to seek nuclear weapons. But we lack broader studies of how regime
characteristics influence the ability to implement decisions to acquire sensitive
nuclear technology or use such technology to move closer toward developing
a nuclear arsenal. Such research should avoid, however, the common assumption
that governments that seek nuclear weapons options have already decided
to get the bomb. As Itty Abraham argues, the proliferation/nonproliferation
lens through which scholars commonly study nuclear history can blind us to
the diversity of motivations of different bureaucratic actors within states and
even the mixed motives of individual leaders. The nonproliferation literature
commonly refers to “nuclear ambiguity”: the lack of knowledge about whether
a foreign government is pursuing nuclear weapons or only nuclear energy. In
some cases, however, a government would be better described as experiencing
3978_SP_potter 1_Book.indb 97 4/27/10 2:17:32 PM
98  Scott D. Sagan
“nuclear ambivalence” because its leadership is undecided or deeply conflicted
about different options for future nuclear development.51
A second line of research could focus on the time period between when
states acquired weapons-usable fissile materials and when they tested a weapon
or had a suspected nuclear weapons arsenal. Such research would be difficult, of
course, because of the lack of firm information on the dates of nuclear program
initiation and successful acquisition of a nuclear weapon in many historical
cases.52
Even if accurate dates were available, however, scholars should be careful
not to assume that each of these different governments was seeking nuclear
weapons with the same degree of urgency or unity. Even among the states that
did eventually develop nuclear weapons, one finds some governments initiating
nuclear weapons programs on a crash basis in wartime or crisis conditions,
others slowly developing a hedge for an uncertain future in peacetime, and
others initiating a nuclear power program or peaceful nuclear explosive program
with only minimal interest in nuclear weapons applications.
This line of research could also usefully focus on the political and technical
factors that influence the size and characteristics of the nuclear arsenals
sought by new nuclear proliferants. Any analysis of nuclear latency timelines
should therefore be mindful of the assumptions used not only about the starting
point (the capacity of an individual enrichment facility or reprocessing
facility) but also about the end point sought by the state (a single weapon,
an arsenal, a simple nuclear device, a miniaturized warhead?). The characteristics
of the arsenal sought by a government have a major impact on the financial
costs, the technical hurdles, and the time involved in a nuclear program.
For example, although outside observers often claim that it would take Japan
six months to a year to develop nuclear weapons if it chose to do so,53
an internal
Japanese government study after the North Korean test reportedly calculated
that it would take Japan three to five years to construct a breeder
reactor, expand its reprocessing facility, and build a prototype of a miniaturized
warhead capable of fitting onto a missile to counter the DPRK threat.54
Unfortunately, we know little about what would-be proliferators think about
what kinds and levels of nuclear weapons might be needed to meet their security
requirements.
Third, it would be useful to have more technical and political science research
on the effectiveness of past efforts to control the spread of uranium enrichment
and plutonium reprocessing. Recent studies by Matthew Fuhrmann
and Matthew Kroenig have contributed significantly to the field by examining
3978_SP_potter 1_Book.indb 98 4/27/10 2:17:32 PM
Nuclear Latency and Nuclear Proliferation  99
the political and economic causes and the security consequences of international
exports of sensitive nuclear facilities, nuclear cooperation agreements,
and sales of nuclear reactors.55
This line of research could be extended to
examine the effectiveness over time, or lack thereof, of strengthened export
controls, the creation of the Nuclear Suppliers Group, and rise and fall of the
A. Q. Khan network.
Finally, there should be more integrated studies of supply and demand for
nuclear weapons. Instead of thinking about these two “sides” of nuclear proliferation
as separate issues, we need to recognize, and therefore study, the
potential for complex connections between supply and demand. Three such
interactions are obvious. First, how hard a government works on a nuclear
weapons program—the resources it commits to the program and whether it
is engaged in a crash effort or normal construction effort—is likely to be affected
by the severity of its demand for a weapon. Second, a high degree of
nuclear capability or latency could influence demand by enabling actors favoring
a nuclear weapon to argue that acquiring a weapon is easier than would
otherwise be the case. Third, a high degree of latency could make it easier for
a pronuclear weapons party or individual leaders to implement a decision to
acquire nuclear weapons if they are in power for only a brief period of time.
NUCLEAR POWER WITHOUT NUCLEAR PROLIFERATION?
Concerns about climate change and growing energy demand have sparked a
global resurgence of interest in civilian nuclear power generation. The potential
increase in nuclear power reactors in experienced states and the spread
of nuclear power to new states will inevitably boost the demand for enriched
uranium to fuel such reactors and increase interest in reprocessing to recycle
plutonium from the back end of the fuel cycle. Will it be possible to have
the expansion of nuclear power without enhancing the risk of global nuclear
proliferation?
A final research task would be to understand more about how the growth
of civilian nuclear power bureaucracies in different states influenced their
decisions to seek or renounce nuclear weapons. Indeed, how best to ensure
that civilian nuclear power bureaucracies maintain a strong interest in opposing
nuclear weapons proliferation may be the proverbial “$64,000 question”
for the future of nuclear nonproliferation. (It will be closer to a $64 billion
question in reality.) On the one hand, leaders of successful nuclear power
enterprises would likely want to maintain strong ties to the global financial
3978_SP_potter 1_Book.indb 99 4/27/10 2:17:32 PM
100  Scott D. Sagan
markets, nuclear power industry, and regulatory agencies and hence seek cooperation
with the nuclear nonproliferation regime. On the other hand, leaders
of less successful or struggling nuclear power enterprises might be more
likely to support nuclear weapons development programs as tools to justify
their existence and budgets within their state. Etel Solingen notes how crucial
the former factor has been in promoting nuclear weapons restraint in East
Asia; Itty Abraham has, in contrast, demonstrated that the weak record in
producing nuclear power encouraged India’s “strategic enclave” in the nuclear
laboratories to lobby New Delhi to acquire nuclear weapons.56
The degree to which the expansion and spread of nuclear power will increase
individual countries’ latent nuclear weapons capability will be largely
determined, however, by who manages and controls uranium enrichment
facilities and plutonium separation and reprocessing facilities. Fortunately,
there are many potential reforms of the international regime, including the
NPT, that could reduce states’ incentives and capabilities to acquire nuclear
weapons even in a world of expanded nuclear power. First, as demonstrated
in ElBaradei’s statements discussed earlier, there is significant interest in the
creation of international nuclear fuel banks, multinational fuel production facilities,
and spent fuel take-back arrangements that would permit expansion
of nuclear power without expansion of national fuel production facilities. Second,
the Nuclear Suppliers Group or even a future NPT Review Conference
could discourage states from exercising their Article X right to withdrawal
from the NPT by making future sales of nuclear fuels and sensitive technology
subject to a “Return to Sender” agreement. U.N. Security Council Resolution
1887 explicitly encourages such a “Return to Sender” policy by requesting that
nuclear supplier states:
require as a condition of nuclear exports that the recipient State agree that, in
the event that it should terminate, withdraw from, or be found by the IAEA
Board of Governors to be in non-compliance with its IAEA safeguards agreement,
the supplier state would have a right to require the return of nuclear
material and equipment [ . . . ] as well as any special nuclear material produced
through the use of such material or equipment.57
Unfortunately, many non–nuclear weapons states fear that efforts to enhance
international controls over the nuclear fuel cycle cut against their “inalienable
right” as specified in Article IV to enjoy the benefits of nuclear energy.
These concerns should, however, be balanced by the hope that both nuclear
3978_SP_potter 1_Book.indb 100 4/27/10 2:17:32 PM
Nuclear Latency and Nuclear Proliferation  101
disarmament and nonproliferation will be enhanced by strong international
control of the fuel cycle. George Perkovich and James Acton have compellingly
argued that weapons states are less likely to agree to complete nuclear disarmament
or go to extremely low numbers in the future if many other states
have their own uranium enrichment or reprocessing facilities and could therefore
develop nuclear weapons openly and more rapidly in an NPT withdrawal
scenario or covertly in peacetime.58
There is, therefore, an important logical
connection between future controls over the nuclear fuel cycle and nuclear disarmament.
The governments of non–nuclear-weapons states should recognize
that entering into negotiations about international control of the nuclear fuel
cycle is part of their Article VI commitment “to pursue negotiations in good
faith on effective measures relating to cessation of the nuclear arms race.”59
If
nuclear disarmament is ever to occur, it is obvious that all the nuclear weapons
states will need to negotiate to reduce and eliminate their arsenals in a mutual
and verifiable manner. It is less obvious, but no less true, that the prospects for
eventual nuclear disarmament will be linked to the success or failure of the
global effort to negotiate serious constraints on the global spread of enrichment
and reprocessing facilities.
3978_SP_potter 1_Book.indb 101 4/27/10 2:17:32 PM