Are U.S. nuclear weapons, in fact, "safe?" The
unequivocal and unanimous conclusion of the nuclear weapons establishment
is affirmative. In his testimony on March 15, 1994, Dr. Harold Smith,
Assistant to the Secretary of Defense for Atomic Energy, told Congress,
I am pleased to report the stockpile today is safe, secure, reliable,
and meets current military requirements. We make that statement
with confidence today and for the immediate future ... Our
stockpile is becoming safer and more reliable simply because we are
retiring older weapons ... Thus, we should enter the 21st
century with a modern, safe, and reliable stockpile consistent with
the demands of START I and with anticipated military requirements. (3)
This statement was made in the presence and with the approval of Dr. Victor
Reis, Assistant Secretary of Energy for Defense Programs, who added,
Right now, as Dr. Smith said, that stockpile is safe and reliable. (4)
Dr. Reis and Dr. Smith are, respectively, a member (one of three) and
the Executive Secretary and Staff Director of the Nuclear Weapons Council.
There is no higher or more integrative authority on this subject. And
what they said has been a consistent theme over the past few years. Smith's
predecessor, Robert Barker, told the Senate in March of 1992,
The Air Force and Navy, in cooperation with the Office of the Secretary
of Defense and the Energy Department, evaluated the safety of all ballistic
missiles that carry nuclear warheads. It was determined that there
is not now sufficient evidence to warrant our changing either warheads
or propellants. (5)
John Deutch, now Deputy Secretary of Defense, reiterated Barker's general
conclusion for the specific case of the W88 warhead on May 3, 1993 when
he told the House Panel on the Military Application of Nuclear Energy
that incorporation of insensitive high explosive (IHE) into that warhead
would not be worth its considerable cost (more than $3 billion). (6)
A few days later, Rear Admiral John T. Mitchell, Director, Strategic
Systems Programs Office, U.S. Navy, was even more blunt. On May
11, 1993, he told a Senate committee that, for the W88 warhead, "We
believe that there would be no gain in safety in changing to insensitive
high explosive." (7)
These comments by Deutch and Mitchell signalled that the safety questions
that had been raised by the Drell Report regarding the W88 (see below)
had been resolved, at least to the satisfaction of the DOD and the Navy. (8)
Deutch, a strong advocate of nuclear weapons design and testing, recently
reiterated the government's consensus on the safety of U.S. nuclear
weapons. On October 5, 1994, Chairman Hamilton of the House Foreign
Affairs Committee suggested to Deutch, following the latter's presentation
of the Nuclear Posture Review, that safety concerns might be a reason
to resume nuclear testing. Deutch demurred, saying unequivocally, "It
is my judgment that all the nuclear weapons that we have are adequately
Dr. Smith unambiguously reiterated the continuing safety theme on March
1, 1995, in testimony before the Senate Appropriations Committee, Energy
and Water Development Subcommittee.
In addition to reducing the sheer size of the active nuclear inventory,
a number of additional actions have been or are being taken to improve
nuclear safety, security, and use control ... I can confidently report
today that the stockpile is safe, secure, and reliable. It also
meets the requirements of DoD and the Services.
The repeated testimony cited here was offered by the highest responsible
careful review of those portions of the Drell Panel's
recommendations that remain outstanding. This testimony states
with abundant clarity that no safety problem currently exists in the
nuclear weapons stockpile. Yet, since the weapons labs still hawk
greater "safety" as a mission, many decisionmakers conclude that nuclear
weapons safety is an unresolved issue. (9)
4. What is the nuclear weapons safety problem?
Those responsible for designing, building, and maintaining
U.S. nuclear weapons do so with a keen appreciation of the dangers inherent
in these weapons and in the materials they contain. It is their
job to make the protection of public safety a paramount concern in every
aspect of their work. Their real goal is to effectively protect
the public, not simply to build safer weapons at any cost. These
two goals are not the same, as can be seen from the following outline,
which supplies a common-sense context for analyzing the safety problem
and maximizing the safety benefits of federal spending.
The Nuclear Weapons Safety Problem in Context
The Overall Goal: Protect the
A. from dangers other than those from nuclear weapons
B. from nuclear weapons dangers (DOE: "reduce the nuclear
1. nuclear attack
a. by an existing nuclear-weapons state
b. by a proliferant nation or group
2. risks to workers from nuclear weapons testing,
manufacture, waste management, deployment, decommissioning
and decontamination, and cleanup
3. risks to public health from the nuclear weapons
operations listed above
a. risks to current populations
b. risks to future generations
4. any environmental risks not included in 3, such
as loss of tribal lands and sites, environmental damage
that is not human health damage per se
5. other, indirect, nuclear dangers (e.g. threats
to democracy from counterterrorism activities)
6. nuclear weapons accidents
a. unintentional nuclear detonation
b. dispersal of plutonium
Based on this outline, there are three relevant hierarchical levels
of safety goals. The most fundamental goal is maximizing overall
. This means to minimize morbidity and early
mortality, from whatever cause. Since it is clear that there is
a finite amount of money available to the federal government to do this,
it is certain that seeking to maximize safety from one type of danger
without regard to cost, e.g. nuclear weapons, would damage overall public
safety, not to mention impede other important goals of government. Choices
will have to be made, then, and an optimum, not a maximum, level of
safety chosen for each particular program, nuclear weapons included.
A subset of public safety is safety from nuclear weapons
this -- which is called by DOE "reducing the nuclear danger" -- means
to minimize morbidity and mortality from nuclear weapons operations
taken as a whole:
the production, storage, processing, and manufacturing of weapons materials,
and the design, production, maintenance, deployment, and disassembly
of the weapons themselves. It includes the public health aspects
of waste management, as well as environmental restoration or the lack
of it. It includes safety from any intentional use of nuclear
weapons, and from nuclear accidents.
But since only a very limited amount of money is available for this
task, it is possible, even certain, that seeking to reduce any one aspect
of the nuclear danger without regard to cost -- either cost in dollars
or in environmental or proliferation risks -- could well increase, not
decrease, the overall nuclear danger.
Finally, a subset of safety from the dangers of nuclear weapons is safety
from nuclear weapons accidents
, which is served by the incorporation
of safety features into the design of nuclear weapons systems.
It is also served by operational changes that decrease the likelihood
of accidents or the public health exposures from accidents, should any
Thus the technical or design aspect of weapons safety
is an important goal, but it is a subservient one. A sense of
proportion is required. An optimum, not a maximum, amount of nuclear
weapons safety is the inevitable and proper goal.
At present, however, the DOE weapons labs are promoting the quest for
greater nuclear weapons safety as if it were an absolute good -- as
if it had no conflict with the other goals of the agency or with the
other goals of the government as a whole. The damage that this
quest could do to those larger goals is discussed briefly below, following
a summary of the technical aspects of nuclear weapons safety.
5. Design aspects of nuclear weapons safety
In a nutshell, the nuclear weapons safety problem
as it affects the design laboratories consists of minimizing the probability
of two general kinds of untoward events:
1) unintentional nuclear detonation of a weapon, either from
a) accidental activation of the firing circuits (e.g. by lightning
or other electromagnetic pulse) or from
b) accidental detonation of the high explosive (HE) in the primary
from an impact, fire, or other non-electrical cause; and
2) dispersal of plutonium due to an accident of any kind.
The following discussion reviews each of these safety concerns in turn,
and concludes with an overview of problems related to aging of weapons.
A. Accidental activation of the firing circuits is a problem
that has been solved
Electrical safety in U.S. weapons is addressed by means of a protection
policy known as Enhanced Nuclear Detonation Safety (ENDS), which is
achieved by a technology called Enhanced Electrical Isolation (EEI).
The Drell Panel describes this system in detail.
The ENDS is designed to prevent premature arming of nuclear
weapons subjected to abnormal environments. The basic idea of
ENDS is the isolation of electrical elements critical to detonation
of the warhead into an exclusion region which is physically defined
by structural cases and barriers that isolate the region from all
sources of unintended energy. The only access point into the
exclusion region for normal arming and firing electrical power is
through special devices called strong links that cover small openings
in the exclusion barrier ... Detailed analyses and tests
give confidence over a very broad range of abnormal environments that
a single strong link can provide isolation for the warhead to better
than one part in a thousand. Therefore, the stated safety requirement
of a probability of less than one a million ... requires
two independent strong links in the arming set, and that is the way
the ENDS system is designed ... both strong links have to
be closed electrically -- one by specific operator-coded
input and one by environmental input corresponding to an appropriate
flight trajectory -- for the weapon to arm.
ENDS includes a weak link in addition to two independent strong links
in order to maintain assured electrical isolation at extreme levels
of certain accident environments, such as very high temperature and
crush. Safety weak links are ... designed to fail,
or become irreversibly inoperable, in less stressing environments
than those that might bypass and cause failure of the strong links.
The ENDs system provides a technical solution to the problem of
preventing premature arming of nuclear weapons subject to abnormal
environments ... ENDS was developed at the Sandia National
Laboratory in 1972 and introduced into the stockpile starting in 1977.
(13) (emphasis added)
While there are some older weapons in the U.S. arsenal that do not contain
ENDS, these weapons are currently being retired. (14)
With these retirements, the problem of electrical safety
of U.S. nuclear weapons has been solved. Note that the one-in-a-million
standard applies in the case of accidents, which are themselves infrequent,
and not in routine operations. The applicable specification for
the probability of an accidental explosion during normal
including all environments in the stockpile-to-target sequence, must
be less than 10-9
per warhead lifetime. (15)
Since the probability of a highly abnormal environment,
i.e. an accident, is now much lower than it was during the Cold War
with its intensive airborne transport of nuclear weapons, projects to
develop firing circuits with still greater isolation possess only very
Such work is, in any case, unlikely to affect the nuclear components
-- the physics package -- of the warhead. It is primarily Sandia
which designs and maintains the arming and safing systems of warheads,
not the two physics labs. Modifications to these electrical systems
are relatively routine and need not trigger major stewardship expenses.
B. The possibility of an accidental nuclear explosion due
to impact or fire has been the subject of intensive study and is extremely
The second aspect of preventing an accidental nuclear explosion consists
in ensuring that impacts, fires, explosions, and any other causes not
covered by the electrical safety system cannot set off any weapon's
high explosive in such a way that any significant nuclear yield results.
Recognition of this danger led to the adoption of the so-called
"one-point-safety" standard in 1968. This quantitative standard
requires all weapons in the stockpile to be "one-point safe," which
is defined as achieved if the probability of a nuclear explosion with
a yield of four pounds TNT-equivalent or greater from detonation of
the HE at any single point is less than one in a million in an accident.
this safety performance must be intrinsic to the design, i.e. it must
obtain in the absence of any mechanical safing device. (17)
Using more detailed computational analysis than had previously been
available, the Drell Panel found that "unintended nuclear detonations
present a greater risk than previously estimated (and believed) for
some of the warheads in the stockpile." (18)
To solve these problems, the Drell Panel recommended a major
competitive effort at the weapons laboratories to design new warheads.
Yet other than an implicit recommendation to quickly retire the SRAM-A
sure the entire stockpile has ENDS, the unclassified Drell Report contains
no specific recommendations for improving the nuclear detonation safety
of the U.S. arsenal.
However, the report did recommend a broad and in-depth review of the
safety of the Trident II (D5) missile system, given the fact that the
W88s used there do not contain IHE and are mounted in a ring around
the third-stage rocket motor, which contains a detonatable propellant.
Kidder concurred with this recommendation and suggested that the
Trident I (C4) W76 system be closely examined as well.
The results of the examination, which set these worries to rest, were
provided to Congress in the testimony quoted above. Meanwhile,
operational changes in the way Trident missiles were loaded into submarines
were immediately implemented. Trident missiles are no longer loaded
into their launching tubes with their warheads in place, which means,
according to Drell, that there is now "no worry" about a dockside warhead
C. To prevent plutonium dispersal, operational changes are
Plutonium -- capable of causing cancer deaths from doses in the microgram
range -- can be dispersed into the environment in any accident in which
the conventional explosive in a nuclear weapon burns or explodes. If
the explosive involved is IHE, an explosion is highly unlikely, since
IHE is remarkably difficult to detonate. In the case of a fire,
the plutonium will burn along with the IHE. Warheads made with
HE may also burn in a fire rather than explode, and in fact this happened
six times at U.S. Air Force bases between 1958 and 1965 when nuclear
warheads were involved in fires.
The only good news here is that, in the absence of an explosion, the
mean particle size of the plutonium oxide produced is larger and less
likely to be inhaled, and is dispersed over far less area, resulting
in many fewer potential casualties. (21)
The Air Force in fact claims that these six accidents resulted
in only localized contamination, which was cleaned up in some fashion
in each case. (22)
All in all, between 1950 and 1980 there were 32 serious nuclear weapons accidents
("Broken Arrows"). None have occurred since 1980. During
that 30-year period there were two accidents that involved explosions
with plutonium. (23)
These were airplane crashes at Palomares, Spain in 1966 and at Thule,
Greenland in 1968. Luckily, these occurred in relatively unpopulated
areas, and no major public exposures resulted. It seems likely,
however, that significant danger was experienced by the cleanup crews,
which were probably not well trained or equipped, both in these cases
as well as in the six accidents in which weapons burned.
Can the possibility of nuclear weapons accidents in which plutonium is
dispersed be eliminated? The answer, of course, is no. Even
with IHE, with fire-resistant pits (FRPs, which have a refractory shell
plutonium), and with speculative "super-safe" designs in which the fissile
material is somehow kept separate from the HE or IHE until the arming
sequence -- there will always be a finite chance of plutonium dispersal
in the event of a fire or other accident. And this finite chance
will continue to be much greater than the one-in-a-million standard adopted
for electrical isolation and for one-point safety.
Still, the dangers from plutonium dispersal, while quite serious, are
far less than those from a nuclear detonation. Claims by lab officials
that a plutonium dispersal accident could be "worse than Chernobyl" are
at least two orders of magnitude off base. (25)
In order to prevent plutonium dispersal, the Drell committee recommended
that "all nuclear bombs loaded onto aircraft -- both bombs and cruise
missiles -- [be built] with both IHE and FRPs." On its face, this
had some appeal, since some 84 percent of serious nuclear weapons accidents
have involved aircraft.
Unfortunately, equipping all airborne U.S. weapons with IHE and FRPs would
require redesigning and rebuilding thousands of nuclear weapons, entailing
dozens of nuclear tests and the construction of new nuclear weapons factories,
such as a replacement for Rocky Flats. Kidder's more practical recommendation
was, instead, to implement cost-saving operational changes that would
reduce the risk of accidents which could result in plutonium dispersal
to near zero. These changes basically consist of not putting nuclear
weapons on aircraft in peacetime, eliminating the possibility of any aircraft
accident leading to plutonium dispersal. Carson Mark, Director of
the Theoretical Division at Los Alamos for 26 years, had argued a month
before Drell that operational limitations on warheads (e.g. no routine
deployment for airborne weapons) would be far more cost-effective than
redesigning them to, for example, incorporate IHE. (26)
Removing nuclear weapons from aircraft has now largely been effected.
In September of 1991, not long after Kidder's study, President Bush decided
to take all U.S. strategic bombers off alert, meaning that all U.S. airborne
strategic nuclear weapons have now joined U.S. tactical nuclear weapons
in secure storage bunkers, out of harm's way. President Bush's initiative
essentially solved the safety issues for airborne nuclear weapons. (27)
Since, according to Drell, the consequences of a plutonium explosion are
roughly one hundred times worse than a plutonium fire, the addition of
IHE to a weapon removes about 99 percent of the plutonium dispersal danger.
FRPs could remove part of the remaining 1 percent of the danger.
FRPs add no degree of safety if the explosive in the warhead detonates,
so there is little point in adding FRPs to a weapon that does not also
have IHE. FRPs cannot reliably withstand a rocket propellant fire,
which could be much hotter than a jet fuel fire (about 2000 degrees centigrade
versus 1000 degrees), so there is little point in adding FRPs to ballistic
So are FRPs worth the expense? Even before President Bush took nuclear
weapons off airplanes, Assistant Secretary of Energy for Defense Programs
Richard Claytor told Congress that "for weapons such as the B-61 family
and the W-80, which already have IHE, this [addition of FRPs] will be
a very costly upgrade to accomplish a modest improvement in safety." He
added that "for tactical systems, where weapons are normally stored in
bunkers, the reduction in risk may be very small." (28)
The Air Force's official response to the Drell Panel also panned the marginal
benefits of FRPs.
Qualitative assessment indicates that [the] safety risk associated with
incorporating FRP into bombs and cruise missile warheads which already
have ENDS and IHE would exceed the safety gain realized by FRP, [and
so such weapons] should not be modified to incorporate FRP. (29)
The Drell Panel called for an aggressive study of "super-safe" designs,
such as designs in which the plutonium was physically separate from the
IHE or HE. In response, Kidder pointed out that such designs had
been under study for at least 15 years (by 1991) without practical result.
Furthermore, any designs finally created would very likely be quite
complex, which means that they might have serious reliability problems.
In any case they would require numerous nuclear tests. This
recommendation, like the one calling for all airborne warheads to have
IHE and FRPs, was evidently not considered practical by the Nuclear Weapons
Council, leading to the testimony cited previously.
Note that IHE was intentionally
not incorporated into the W88 Trident
D5 warhead, because IHE is less energetic and reduces the yield of the
weapon, the range of the missile, and/or the number of warheads it can
carry. A conscious decision was therefore made to not
these particular weapons as safe as possible, because they were, and still
are, judged to be safe enough. In addition to the W88, IHE is also
not present in the W62 and W78 Minuteman III warheads and the W76 Trident
C4 warhead. The W62 and W78 warheads are being retired, which will
leave the Navy's W76s and W88s as the only warheads in the START II arsenal
lacking IHE. (30)
Interestingly, an FRP-equipped cruise missile warhead (the W84, one of
only three such FRP-equipped weapons in the stockpile) has been taken
of the active stockpile, in favor of cruise missile warheads
that lack FRPs (specifically, the W80-0 and W80-1). (31)
Thus the Air Force, the Navy, and the Nuclear Weapons
Council have, on at least three occasions if not also on others, concurred
in decisions that chose warheads for the so-called "enduring" stockpile
that lack some of the possible safety features that could have been incorporated.
The decisions to forego FRP- and IHE-equipped weapons contrast sharply
with the rhetoric coming from the labs calling for so-called "safer" weapons.
Finally, note that incorporation of an FRP and, especially, IHE into a
weapon would require a substantial redesign and would, in effect, amount
to a new weapon. Kidder suggests that roughly three nuclear tests
per warhead or bomb would be necessary to proof-test the former and that
six such tests would be required for the latter. (32)
Thus these are not minor changes, and they would
require perhaps two dozen nuclear tests to accomplish for the entire START
II stockpile. It is certainly not accurate to call such changes
"safety improvements to existing weapons," as is now commonly done.
Table 1 (attached) shows the planned U.S. stockpile and its safety features.
D. Dr. Drell believes that physics package safety improvements
should not be undertaken
Since his 1990 work, Dr. Drell has clarified his position on the advisability
of modifying nuclear weapons physics packages to increase safety. (33)
Writing with Bob Peurifoy, Drell acknowledges that some weapons in the
so-called "enduring" arsenal do not have every safety feature to prevent
plutonium dispersal and that some questions involving multi-point safety
remain for the Trident systems. But these deficiencies should not,
in their view, lead to modification of the stockpile physics packages.
Drell and Peurifoy point out that if multipoint safety was judged to be
inadequate for the Tridents, and monies were actually to be budgeted to
improve them, this could be done by means other than modifying the warhead:
installation of blast shielding around the third-stage motor and decreasing
the number of warheads to compensate for the weight, changing the third-stage
propellant, costing about 4% in range, or omitting the third stage motor
altogether and decreasing the number of warheads from eight to four to
give the same range as at present. (34)
Should some "devices" -- the term used by Drell and Peurifoy for the nuclear
explosives themselves -- be changed to make plutonium dispersal less likely?
They say not.
Most importantly, in the case of a test ban, one should not tamper with
the device hardware once it has been certified ... in the absence of
follow-on stockpile testing, hardware modifications must be avoided
... New, untested devices should not be considered.
From a technical point of view, safety improvements could be achieved
in a continued underground testing program focused on safety. From
a political perspective, however, continued testing of nuclear weapons
may hinder efforts to counter, if not prevent, the proliferation of
nuclear weapons in the years ahead.
In our view, the early achievement of a strengthened and durable worldwide
nonproliferation regime will contribute more significantly to worldwide
nuclear safety than will further improvements in the safety of part
of the US nuclear force. (35)
Drell and Peurifoy then call for the U.S. to "take the initiative"
and "lead a joint effort by the five declared nuclear powers to negotiate
an end to all nuclear weapons testing" (emphasis added). They
also call for "a diverse and coordinated scientific program at the national
weapons laboratories so that they can maintain and certify confidence
in the US nuclear deterrent over a long period without testing."
(36) Maintaining safety is not mentioned as a
challenge for that program, for reasons that will now be examined.
E. There are no safety problems related to aging of weapons
Although authorities agree that nuclear weapons are "safe" now, often
vague reference is made to possible safety problems that could arise
in the future. What these problems might be is never mentioned,
however. After all this intensive review by the Drell Panel and
subsequently by the labs -- who were, it is fair to say, searching for
every possible reason to continue nuclear testing and nuclear weapons
design work in general -- it is difficult to imagine that some heretofore
overlooked safety problem of real significance would suddenly appear.
Likewise, no safety problems are expected to occur in the aging process.
This was the first question posed to Dr. Kidder by the senators, and
he makes it clear that aging does not create safety problems.
Safety problems with nuclear warheads are generally inherent in the
design of the warhead itself, not the result of aging or other causes.
Such problems may not be identified until long after the warhead enters
stockpile, but they were there to begin with.
Metals corrode, and organic materials such as plastics, adhesives, and
HE that are present in a nuclear warhead will deteriorate with ageSuch
aging effects degrade a warhead's reliability rather than its safety.
The sensitivity to impact or fire of the HE used in nuclear warheads
does not increase significantly with age.)
A severe case of aging was the deterioration of the HE in the W68 Poseidon
warhead, which produced a harmful, chemically reactive effluent. This
resulted in a potential loss of warhead reliability that necessitated
a complete rebuild of all W68 warheads in stockpile. The reliability,
but not the safety, of the warhead was affected. (37)
Dr. John Immele, Director of Los Alamos National Laboratory's (LANL's)
Nuclear Weapons Technologies Program, spoke to this same point on December
8, 1994 at a public hearing in New Mexico.
Audience: I have one more question ... in a deleterious
way, they may age or crack. What do you mean, is there a risk
to the public?
Immele: No, there's not a safety risk. There's a performance
problem ... because insensitive high explosive is so insensitive
that sometimes if it's cracked it won't light on the other side when
it's supposed to, so it's basically a performance problem.
Audience: A reliability problem?
Immele: That's right, it's a reliability issue. We have
not found aging problems that affect safety, that make the explosive
This testimony is very important, since the long-term behavior of IHE
is not understood as well as that of HE, and it is sometimes mistakenly
implied that this uncertainty extends to questions of safety, which it
clearly does not.
The impression should not be left that once weapons are put into the stockpile
they are forgotten. The stockpile surveillance program coordinated
by the laboratories, especially by Sandia, routinely inspects weapons
and their components and ensures, among other goals, that safety problems
do not develop. Thus safety throughout the aging process is ensured
by both the current weapons' inherent safety features and a coordinated
F. DOE's proposed new facilities have nothing to do with weapons
As part of its proposed SBSS program, DOE is planning to build a number
of new experimental weapons science facilities with a total cost running
into the billions of dollars. (39)
The largest of these is the National Ignition Facility
(NIF), a laser fusion machine with a currently-estimated capital cost
of about $1.2 billion and a lifetime cost of about $4.5 billion.
of NIF in the SBSS program is to simulate the implosion process that occurs
in the thermonuclear "secondaries" of nuclear weapons.
Throughout the 40-year history of the thermonuclear arsenal, no safety
problems have ever been identified with secondaries, which contain neither
plutonium nor high explosives. Nor are any expected. It is
therefore patently obvious that NIF has nothing to do with nuclear weapons
safety. Dr. Steven Younger, who was at the time Deputy Program Director
of the LANL Nuclear Weapons Program, was among others who made statements
to this effect at a DOE-sponsored workshop on NIF on September 8, 1994
in Washington. (41)
For the same reason, other facilities for simulating secondary implosion,
like ATLAS (LANL's proposed new pulsed-power implosion facility) have
nothing to do with nuclear weapons safety either. The only facilities
with potential relevance for nuclear weapons safety are those being built
or planned for the simulation of nuclear weapons primaries. These
are the hydrodynamic testing facilities: the $124 million Dual-Axis
Radiographic Hydrotest facility (DARHT), under construction at LANL but
halted for environmental review, and the $422 million Advanced Hydrotest
Facility (AHF), planned for the early years of the next century.
Hydrotest facilities cannot test either the high explosives or the plutonium
pits of stockpiled weapons. The former cannot be separated intact
from the pit they embrace, and the latter cannot be tested without a nuclear
explosion. Therefore, these facilities test mock
assemblies. There is very little point in conducting hydrodynamic
explorations of the safety of existing weapons. These weapons are
to be one-point safe and their plutonium-dispersal
properties are already clear -- either they have IHE and FRPs or
they do not.
The only purpose of these facilities, as far as safety is concerned, is
that they can be used to design new
primaries that have IHE and
FRPs, either primaries for new weapons with new military characteristics
or to retrofit into existing warheads and bombs. This can
be done either directly, by testing mockups of these new primaries, or
indirectly, by conducting precise hydrodynamic tests on existing designs
for which nuclear testing data is available ("benchmarking"). Benchmarking
allows the nuclear testing database to better inform the nuclear weapons
codes, which can then be used to design new weapons.
Actually, DARHT may be inadequate either to design new weapons with IHE
and FRPs or to benchmark some stockpiled systems, with or without these
Recognizing the importance of continued research in radiography, the
Laboratory [LANL] cites DARHT as its top construction priority ... For
a number of stockpile systems, particularly those that are designed
with insensitive high explosives and fire-resistant pits, planned radiography
upgrades [i.e. DARHT] do not provide resolution adequate to observe
the gas cavity configuration of the primary stage late in the implosion
process. For effective monitoring of stockpile weapons [sic] of
this type, a next-generation hydrodynamic testing capability will need
to be developed. Such an Advanced Hydrotest Facility (AHF) will
include multiple beams that produce X-rays from four to six directions
at various times to characterize the physical state of the pit more
It is difficult to avoid the inference that DARHT is useful primarily
to design new weapons which lack IHE and FRPs
. This is not
at all unlikely from a policy point of view, given that these features
have already been intentionally omitted from significant portions of the
Any use of these new hydrotesting facilities to examine safety problems
relating to aging of weapons is moot, since logic, together with Drs.
Kidder and Immele, all agree that there are no such issues.
6. Reducing the nuclear danger
The risk to the public from a nuclear weapons accident
is only one among many interrelated risks associated with nuclear weapons.
Attempts to increase safety from nuclear weapons accidents through
redesigning warheads may easily create countervailing risks throughout
this interconnected system. For example, the replacement of hundreds
or thousands of warheads, and especially the plutonium pits of warheads,
will create risks to workers and the general public during the manufacturing
process and in the management of its wastes, not to mention during the
eventual decontamination of buildings and equipment that will later
be required. Any environmental contamination that occurs will
require cleanup. These considerable risks tend to be minimized
or forgotten entirely by the advocates of weapons redesign.
In fact, these manufacturing, waste management, decommissioning, and
cleanup risks are likely to be much greater than the risk reduction
that could be achieved by the addition of, for example, IHE to W88s
and W76s, to pick one modification currently under consideration. The
historical record suggests this. While there have been no
known or putative deaths due to accidents from explosions or plutonium
dispersal from completed weapons -- even during the Cold War when thousands
of weapons were shuttled all over the world -- numerous deaths have
occurred due to ordinary occupational causes in the nuclear weapons
complex as well as to chemical toxins and radioactivity. (43)
Meanwhile cleanup, which could be quite hazardous, has barely
Moreover, should the U.S. elect to upgrade its nuclear arsenal for the
sake of "improved" safety or for any other reason, it can be expected
that the other nuclear powers, particularly Russia and China but probably
also Britain and France, may do likewise. In that case, the morbidity,
mortality, and environmental damage in other countries, again notably
Russia and China, can be expected to be equal or greater than here.
Thus, the quest for greater nuclear weapons "safety," if allowed to
proceed within the compartmentalized thinking that characterizes the
bureaucratic "stovepipes" of the federal nuclear establishment, will
likely saddle current populations and future generations, here and abroad,
with increased, not decreased, risk. And, as has been the
case up to now, this risk will tend to fall most heavily on the vulnerable
members of society and the populations that are weakest politically.
A. The proliferation impacts of safety "improvements" must
Another component of the overall nuclear danger is nuclear proliferation.
A public opinion poll conducted by Sandia National Laboratory for DOE
suggests that the public considers the risk from nuclear proliferation
to be the number two danger facing us today, right behind world hunger
and ahead of AIDS, drug trafficking, and global warming. (45)
Thirty-two percent of the public thought nuclear proliferation was an
"extreme risk." This opinion is matched by a widespread concern
among experts that the proliferation problem, far from being under control,
is a very serious threat to the security of the United States. Assuming
this is true, we can conclude that if safety improvements in nuclear
weapons design incur even a small incremental risk to the world's nonproliferation
regime, such "improvements" are likely to increase, not decrease, the
nuclear danger. On its face, nuclear proliferation is simply a
much more serious problem than nuclear weapons safety.
It is beyond the scope of this paper to discuss in any detail the proliferation
dangers inherent in upgrading the U.S. arsenal to achieve "increased"
safety. Suffice it to say that even a heuristic analysis of comparative
risk must embrace the reality that such safety upgrades attempt to prevent
events whose probabilities are already very low. The probability
of proliferation under current policies, however, is not low at all,
and the probability of nuclear attack or threat of attack by proliferant
nations or groups approaches certainty in the long run if more effective
leadership on this issue is not forthcoming.
Any analysis of the nonproliferation impacts of contemplated weapons
upgrades must therefore consider every possible impact, even those which
are slight. Just as every credible nuclear weapons accident should
be examined and prevented, so should every credible potential proliferation
impact also be prevented. The probability of an accidental nuclear
detonation is kept at 10-9
per warhead lifetime, or roughly
for the current arsenal taken as whole over the next
few decades; any decision to upgrade U.S. weapons should likewise be
examined to see whether, through their impact on the nonproliferation
regime, such upgrades could cause an increase in the probability of
attack on the U.S. as small as 10-5
over the next few decades. Clearly, the expected mortality from
an intentional attack is much larger than that for any credible accident,
further emphasizing the relative importance of proliferation risks and
the efforts needed to combat it.
This spring, the Nuclear Nonproliferation Treaty (NPT) was extended
indefinitely. Despite some media claims to the contrary, this
was not done by consensus but by an uncontested acknowledgement that
a majority of states favored extension. (46)
This extension was won only after what the Washington Post called a
"global full-court press" by the U.S.;(47)
late as January of this year, roughly one hundred countries did not
support indefinite extension.(48)
The reason for the vocal and sometimes bitter debate that went on in
New York was, above all else, deep international discontent with the
failure of the nuclear weapon states to dismantle their nuclear arsenals
pursuant to Article VI of the NPT. Because of this failure, substantive
further strengthening of the world's nonproliferation regime now appears
to be temporarily out of reach.
When even maintenance
of arsenals can threaten to unravel the
fragile fabric of the world's nonproliferation efforts, as it did this
spring, how can upgrading
that arsenal, whether for so-called
"safety" improvements or any other purpose, not damage U.S. nonproliferation
efforts? The conclusion of Drell and Peurifoy is sound: safety
improvements to the nuclear components of weapons, which cannot be undertaken
without underground testing, run counter to U.S. nonproliferation objectives.
Other nations may be concerned that "safety upgrades" can mask the development
of entirely new weapons. There is considerable justification for that
concern, as is discussed in detail by William Arkin, Greenpeace, and
Foreign "safety upgrades" may also be used to mask new weapon
development, an outcome with negative security implications for the
B. Fewer deployments and further retirements are virtually
the only way greater overall nuclear weapons safety can be achieved
Given that upgrading and replacing nuclear weapons is likely to create
serious countervailing risks -- risks which, on their face, are considerably
greater than those gained from any purported safety "improvements" --
the search for greater nuclear safety must be directed elsewhere. Clearly,
all other factors being equal, the probability of a serious weapons
accident is proportional to the overall size of the arsenal. A
minimum deterrent force -- however one may define "minimum" -- is also,
therefore, an optimum safety arsenal. It is a minimum cost arsenal
as well. A smaller arsenal would assist U.S. nonproliferation
objectives, and would cause fewer dangers to the environment, to worker
safety, and to public health. A smaller arsenal, to the extent
that it corresponds to smaller arsenals abroad and especially in Russia,
reduces the number of missiles which could be targeted at the United
Indeed, given comparable reductions in other nations' stockpiles, it
can be persuasively argued that the optimum safety arsenal is one that
is extremely, if not vanishingly, small. Many military and senior
civilian defense leaders, past and present, have come to adopt this
In addition to reducing the size of the arsenal, the movement of additional
weapon systems away from active deployment and into safe bunkers would
also reduce risk from accidents, as would the movement of nuclear weapons
away from bases located in populous areas. Further restrictions
on the airborne transport of nuclear weapons should also be considered.
C. Increasing safety will decrease reliability -- and hence
could generate calls for nuclear testing
As we have seen, the quest for "increased" safety has a very marked
cost in terms of weapons reliability. That is, if weapons are
changed to make them "safer," they will be less reliable -- unless we
are able to proof-test the new designs. Resuming nuclear testing, however,
would conflict strongly with nonproliferation goals and our treaty commitments.
Thus the quest for "safer" weapons, if accepted at face value,
could keep the weapons labs in a booming business for a long time by
eroding the reliability of the weapons.(51)
7. Public safety, not nuclear safety, is the goal
So far we have examined nuclear weapons safety issues
from a narrow technical perspective and in the broader context of reducing
the nuclear danger as a whole. From the still broader vantage
point of public health and safety as a whole, further investments in
safer designs for nuclear weapons have vanishingly small returns.
A. Further investments in nuclear weapons
safety have a very low benefit/cost ratio compared to other public safety
Would upgrading to IHE and FRPs be worth the expense? Analysis
-- not to mention common sense -- shows that investments in other government
programs (e.g. highway improvements, cancer screening) yield orders
greater safety benefits to the general public.
Steve Fetter and Frank von Hippel estimate that a worst-case accident
involving explosion of the HE in ten W88 warheads at the Bangor Trident
base directly upwind from Seattle would involve on the order of 1,000
plutonium-induced cancer fatalities in the long run. (52)
They suggest, for the sake of argument, that the risk of this accident
can be assumed to be on the order of 0.1% per year, in which case the
expected fatality rate from this type of nuclear weapon accident is
1 death per year. The prevention of this accident by the spending
of about $1 billion (Drell and Peurifoy's more recent article cites
a cost of $1.6 to 1.8 billion) to equip some 3000 submarine missile
warheads with IHE would represent a cost on the order of $100 million
per fatality avoided. (53)
This accident has, subsequent to their paper, been made
very unlikely by loading the missiles and the warheads separately, lowering
the expected fatality rate by probably at least one, if not two or more,
orders of magnitude and correspondingly raising the cost per fatality
Fetter and von Hippel cite cost estimates in the range of $20,000 to
$140,000 per life saved by cancer screening, $400,000 per life saved
by kidney dialysis, and $30,000 to $300,000 per life saved for various
highway safety improvements. Thus the IHE warhead upgrade program,
even by their highly conservative calculations, would cost on the order
of 250 to 3,000 times more than these other prevention programs per
life saved -- or, given the operational changes already put in place
by the Navy, at least 10 to 100 times as much as this. This great
disparity of benefit -- at least 3 if not more than 5 orders of magnitude
-- signals that the overall sense of their conclusion is robust with
respect to large changes in their assumed accident rate.
What is more, the government and private programs cited by Fetter and
von Hippel are almost certainly not the most effective ones offered
by government or private sources, either in terms of average cost or
marginal cost per life saved. Programs targeted at populations
at risk like the Women, Infants, and Children (WIC) program, for example,
are arguably at least one order of magnitude more effective per dollar
spent than highway improvements in preventing deaths. And, on
average, WIC and other perinatal programs address a younger population--one
young adult and one infant -- than do the programs Fetter and von Hippel
cite, giving more years of life saved per person in these cases.
It would be interesting to compare Fetter and von Hippel's numbers to
the benefits of investments in strictly military
health and safety.
The military environment is a dangerous place, and large numbers of
military accidents occur annually, sometimes with accompanying civilian
deaths. It is highly likely that it would be more cost-effective
to use a billion defense dollars to prevent unnecessary military mortality
-- through, for example, more complete training -- than it would be
to use this sum to upgrade Trident missiles and their warheads, for
which purpose a billion dollars would probably not be adequate in any
This, of course, is a peacetime comparison. Since, realistically,
nuclear weapons are useless in actual fighting, any investment in them
deprives the soldier, sailor, or airman of just that much supporting
materiel or training when he or she needs it most.
B. The quest for nuclear weapons safety is inconsistent
with other federal and DOE positions, past and present
In the United States, 32.5 persons per 100,000, or approximately 81,250
people, died from "accidents and adverse effects" in 1990. (54)
This is a very large fatality rate, much more even than a major conventional
war. In the same year (as in every other year since the beginning
of the nuclear age) not one person died from the accidental explosion
of a nuclear weapon or, as far as is known
, from exposure to
nuclear materials from a nuclear weapons accident. (55)
Actuarially speaking, nuclear weapons accidents don't even appear on
But what about the future? A heuristic analysis, which can only
be very crudely approximate, suggests that an estimate of risk of death
due to a nuclear weapons accident is likely to be, on its face, two
to three orders of magnitude below the risk factors typically used as
a basis for federal environmental health standards, namely 10-5
per lifetime of exposure. (56)
This comparison is made, for all its inevitable flaws, because
the weapons laboratories from time to time engage in struggles to weaken
these environmental standards, notably in regard to ground and surface
water quality, saying that the safety risks involved are "negligible."
Irony aside, this comparison is a prima facie indication that
the public health cost of weapons safety "improvements" could easily
exceed the risk reductions attainable through design changes -- even
if those "improvements" caused only those population exposures which
were considered safe and no accidents did not occur in production, waste
management, decommissioning, and cleanup associated with the design
More ironic still is the fact that the same laboratories who are even
now clamoring for money to develop "safer" nuclear weapons -- this after
the weight of evidence presented in the past four years and the military's
lack of interest in the subject -- are exactly the ones who were saying
that above-ground nuclear testing was not dangerous just a few decades
ago. Federal agencies, such as the Veterans Administration and the Department
of Justice, as well as many individuals at the laboratories, still deny
the legacy of this testing in specific cases and in general.
When it was convenient to do so, these institutions systematically lied
to the public and to the servicemen who were intentionally exposed about
the actual risks of nuclear testing. Even today they continue
to withhold extensive information about measured fallout distributions.
nuclear weapons accidents, which, as far as we know, have been the cause
of no casualties, the casualties of this intentional program were immense.
A special team convened by the International Physicians for the
Prevention of Nuclear War calculated the expected mortality from fallout
worldwide to be 430,000 deaths by the turn of the century. (58)
Frank von Hippel's update of Sakharov's 1958 estimates of mortality
from above-ground testing suggest that the long-term death toll will
be in the neighborhood of 5 million persons. (59)
Against this background, the call for "safer" nuclear weapons rings
very hollow indeed.
We have seen how fears about the "safety" of nuclear
weapons have been rhetorically advanced by the weapons laboratories
and the DOE with little regard to these simple facts:
- current U.S. nuclear weapons have the benefit of fifty years
of technical improvements in safety, and further design improvements
can bring only marginal and diminishing returns in actual risk reduction
at a very large dollar cost;
- upgrading the arsenal for the sake of safety will create countervailing
risks throughout the complex and the world;
- upgrading the arsenal could have enormous nonproliferation impacts;
- much more cost-effective and elegant non-technical solutions
to decreasing risk are available -- such as retirements, further
reductions in arsenal size, and changes in deployment and transportation.
It is as if the Department were simply pandering to images of doom in
order to generate political capital for its science-based stewardship
program, which actually has very little to do with safety. The
Department as a whole is unreasonably tolerant of the stark contrast
between its own very public promotional rhetoric and that of its contractors
regarding the "safety" issue, on the one hand, and its joint testimony
with the DOD to Congress, on the other, which essentially lays the issue
From a broader perspective, the nuclear weapons "safety" debate has
lost its sense of proportion because it has focused on the safety of
, rather than on the safety of people
this all-too convenient process of linguistic contraction, weapons are
made the primary reference reality, not the public. These distortions
have occurred because the pitch from the labs and the DOE is actually
not motivated by safety, but by a desire for a less-conflicted weapons
ideology, especially for the younger scientists, and for perpetual funding.
It is this funding
that would be made safer. Nothing
else can explain the irrationality of the open-ended quest for so-called
"safer" nuclear weapons.
1. Testimony of Deputy
Secretary of Defense John Deutch before the House Foreign Affairs Committee,
October 5, 1994: "...advanced permissive action devices, so-called
coded control devices, will be introduced into the B-52, the Minuteman
IIIs, and eventually into the Trident force as well."
The draft "Stockpile Stewardship Program Plan for Fiscal Years 1995
through 1997," prepared for the Interagency Working Group on February
27, 1995, makes it clear that permissive action links are the subject
of active research, development, and replacement in several, if not
many or all, weapons in the stockpile.
2. Sidney Drell, John Foster, and Charles Townes,
"Report of the Panel on Nuclear Weapons Safety," House Armed Services
Committee, December 1990.
3. Testimony before the House Appropriations
Committee, Subcommittee on Energy and Water Development, Energy and
Water Development Appropriations for 1995, Part 6
, pp. 413-414.
, p. 419.
5. Quoted in Tom Zamora-Collina, "New Jobs
for Old Labs," Bull. Atomic Scientists
, November 1992, p. 16.
6. Quoted in Frank von Hippel and Tom Zamora-Collina,
"Testing, Testing, 1, 2, 3 -- Forever," Bulletin of the Atomic Scientists
July/August 1993, p. 28-32. Deutch went on to say, "...there are
operational steps that one can take to...ameliorate the safety problems
when you mate the warheads to the missiles that are being looked into,
but I would not think this an immediate problem..."
7. Testimony before the Senate Subcommittee
on Nuclear Deterrence, Arms Control, and Defense Intelligence, cited
by von Hippel and Zamora-Collina.
8. Sidney Drell, John Foster, and Charles Townes,
Report of the Panel on Nuclear Weapons Safety
, House Armed Services
Committee, December 1990.
9. The most authoritative and detailed review
of these issues in the open literature is still that of senior Livermore
weapons physicist Ray Kidder, who was requested to provide an independent
analysis of the safety question by several members of the U.S. Senate.
See Kidder, "Report to Congress: Assessment of the Safety
of U.S. Nuclear Weapons and Related Nuclear Test Requirements," July
26, 1991, UCRL-107454, and also his "Post- Bush Initiative Update" of
the same paper in December of 1991, UCRL-109503. His thorough
investigation followed, and built upon, that of the Drell Panel and
incorporates relevant portions of their work in summary fashion.
10. Public safety is only one of the goals
of government, and so this hierarchy of goals is really just one component
of a larger picture. While this outline is an oversimplification,
the primary roles and risks of nuclear weapons are all included here.
Some believe that nuclear weapons have a role in extended deterrence,
i.e. in protecting against conventional attack against U.S. forces on
foreign soil or against non-nuclear weapons of mass destruction. Others,
such as the present writer, believe that nuclear deterrence in any form
is counterproductive and illegal in the long run, if not also in the
short run. These issues are beyond the scope of this paper.
11. This slogan has been in frequent use in
the nuclear weapons community since at least 1993. It is the integrating
principle of the draft DOE National Security Strategic Plan, dated November
3, 1993. Reducing the Nuclear Danger: The Road Away From the
is the title of a book by McGeorge Bundy, William Crowe, and
Sidney Drell (Council on Foreign Relations, 1993).
12. These safety concerns are reflected in
the military characteristics (MCs) required by the Pentagon during the
weapon design process. Those requirements include, in order of
priority, nuclear safety, size and weight, plutonium dispersal safety,
operational reliability, yield, conservative use of nuclear materials,
and operational simplicity. In the event of a conflict between
design priorities, nuclear safety is the highest
George H. Miller, Paul S. Brown, and Carol T. Alonso, "Report to Congress
on Stockpile Reliability, Weapon Remanufacture, and the Role of Nuclear
Testing," 1987, UCRL-53822.
Interestingly, this official paper, in its exhaustive review of all
the reasons to continue nuclear testing, does not mention any
unresolved safety issues with then-current warheads or bombs. It
was only in the early 1990s that safety issues gained prominence, first
during the nuclear testing debate, and now as a challenge for science-based
13. Drell, Foster, and Townes, pp. 25-26.
14. These weapons are the 9-megaton B53-1
gravity bomb (which has only one electrical safety system rather than
two independent ones) and the W62 Minuteman III warhead; see Christopher
Paine, "CTB Negotiating Issues with Implications for Nuclear Nonproliferation,"
Natural Resources Defense Council, 1994, and Stan Norris and Bill Arkin,
"Nuclear Notebook," Bulletin of Atomic Scientists
1995, pp. 69-71.
15. Drell, Foster, and Townes, p. 13.
16. More recently, "multi-point" safety has
also been made an objective. The discussion which follows includes
both one- and multi-point safety concerns.
17. Mechanical safing, which has been available
and in successful use for more than two decades, can virtually eliminate
the possibility of an unintended nuclear explosion, even if many points
of the HE or IHE are detonated at once. According to Frank von
Hippel, mechanical safing can be added to a nuclear warhead without
nuclear testing. See von Hippel, testimony to the Senate Foreign Relations
Committee, July 23, 1992.
18. Drell, Foster, and Townes, p. 25. The
only place these analyses could have been done, given the classification
barriers, the specialized codes, and the data requirements, was the
weapons labs. It is quite likely that at least preliminary analyses
of this type had already been done when the Drell Panel began its work.
Indeed, at least part of the impetus for the Drell Panel was testimony
offered by the three weapons lab directors before the Senate Armed Services
Committee in May, 1990, which all but said that the Short-Range Attack
Missile-A (SRAM-A) was too dangerous to deploy in peacetime (see Drell,
19. "It is not sufficient to pull such weapons
off the alert ALFA force but retain them in the war reserve stockpile
in view of the hazards they will present under conditions of great stress..."
, p. 32.
, p. 29.
21. The Drell Panel estimates that a Pu fire
or deflagration would contaminate about one square kilometer, while
an explosion "could" contaminate an area of "roughly" one hundred square
, p. 30.
22. Frank von Hippel and Steve Fetter, "Worse
than Chernobyl?" Arms Control Today
, September 1992, p. 13. Descriptions
of these accidents can be found in Chuck Hansen, U.S. Nuclear Weapons:
The Secret History
, Orion Books, 1988 or in the Congressional
, August 3, 1992, pp. S11172-5.
23. A third such accident may have occurred
at Dyess AFB in Texas in 1958. See note 54.
24. Vanadium has been used; see Miller, Brown,
and Alonso, op. cit.
25. von Hippel and Fetter, op. cit.
26. J. Carson Mark, "Do We Need Nuclear Testing?",
Arms Control Today
, November 1990, pp. 12-17.
27. See Kidder's December 1991 update, cited
above, for the text of the President's announcement. Kidder notes
this announcement did not go quite as far as to completely eliminate
the air transport of nuclear weapons in peacetime, but points out that
such transport "could be reasonably terminated after the mandated return
of overseas nuclear weapons has been completed..." (p. 3).
28. Cited by Ray Kidder in "How Much More
Nuclear Testing Do We Need?," Arms Control Today
, pp. 11-14.
Claytor spoke before the House Armed Services Committee, DOE Defense
Nuclear Facilities Panel, February 20, 1991.
29. Air Force Response to the Drell Panel
Nuclear Weapons Council Standing Committee briefing, Lieutenant Colonel
John R. Curry, Secretary of the Air Force/Assistant Secretary, Acquisition,
August 1, 1991; cited by Kidder, 1992.
30. Norris and Arkin, op. cit.
Paine, op. cit.
32. Kidder, July 1991, pp. 8-9.
33. Sidney Drell and Bob Peurifoy, "Technical
Issues of a Nuclear Test Ban," Annual Review of Nuclear and Particle
Science, 1994, pp. 285-327.
, p. 313.
, pp. 321, 325, and 326.
, both quotes are from p. 326.
, p. 6.
38. Colloquy at public environmental impact
scoping hearing regarding the Dual-Axis Radiographic Hydrotest Facility
(DARHT), December 8, 1994, in Santa Fe; transcript available from DOE
Albuquerque or LANL.
39. Tom Zamora-Collina and Ray E. Kidder,
"Shopping Spree Softens Test Ban Sorrows," Bulletin of the Atomic
, July/August 1994, pp. 23-29.
40. See DOE FY 1996 Congressional Budget
Request, Project Data Sheets, Vol. 1
, p. 332. The total cost
cited is the sum of capital and annual operating costs.
41. Personal communication at LANL site-wide
environmental impact statement hearing, Santa Fe, fall 1994.
42. Los Alamos National Laboratory, Institutional
, p. 43.
43. A 900-page report on this subject is due
from the International Physicians for the Prevention of Nuclear War in
July of 1995, entitled A Global Guidebook to Nuclear Weapons Production
and Its Health and Environmental Effects
, MIT Press, Cambridge.
44. See Office of Technology Assessment, Hazards
Ahead: Managing Cleanup Worker Health and Safety at the Nuclear Weapons
, OTA-BP-O-85, February 1993.
45. Released by Sandia on June 24, 1994.
46. "Beyond the NPT: `Abolition 2000!'"
A special report of the Western States Legal Foundation, June 14, 1995,
47. Washington Post
, May 14, 1995.
48. "Atom Arms Pact Runs Into A Snag," New
, January 26, 1995, p. 1.
49. See "Changing Targets: Nuclear Doctrine
from the Cold War to the Third World," Hans Kristensen and Joshua Handler,
Greenpeace International, January 1995; "Nuclear Agnosticism When Real
Values Are Needed: Nuclear Policy in the Clinton Administration," William
Arkin, Federation of American Scientists. The potential proliferation
impacts of the DOE's "science-based stockpile stewardship" program, and
DOE's design program for new nuclear weapons, will be analyzed in forthcoming
papers from Tri-Valley CAREs.
50. General Horner, former head of the Air Force
Space Command and leader of the air war against Iraq, is one ("U.S. Should
Trash Nukes, Top Air Force General Says," Albuquerque Journal
July 16, 1994). General Andrew Goodpaster, former NATO commander,
is another ("Tighter Limits on Nuclear Arms: Issues and Opportunities
for a New Era," and "Further Reins on Nuclear Arms: Next Steps for
the Major Nuclear Powers," The Atlantic Council of the United States,
1992 and 1993 respectively). The views of Les Aspin, recent Secretary
of Defense, are likewise well-known.
Even some weapon designers are beginning to agree. Tom Thompson,
dean of current designers at Livermore, admits, "I can't think of any
target for anything in our stockpile" ("Science Comes in from the Cold,"
Los Angeles Times
51. Dr. Steven Younger, Deputy Program Director
for Nuclear Weapons Technology at Los Alamos, admitted to the writer that
these arguments were "defensible." He concluded, for these reasons
as well as reliability concerns, that "we should not open up existing
weapons [to changes] unless it is absolutely necessary" (personal communication
after Los Alamos Study Group panel discussion in Los Alamos, July 18,
52. Steve Fetter and Frank von Hippel, The Hazard
from Plutonium Dispersal by Nuclear-Warhead Accidents, Science &
, Volume 2, 1990.
53. Norris and Arkin estimate that the total
number of W76s which will remain after START II is 1328, and that of W88s
400, making 1728 submarine-launched warheads without IHE after START II,
slightly more than half the number von Hippel and Fetter used in 1990.
54. Statistical Abstract of the United States
U.S. Governmental Printing Office. The accidental death rate has
declined some 40 percent over the last two decades, reflecting the effectiveness
of societal investments in safety like the highway improvements cited
by Fetter and von Hippel. These investments, which compete for funding
with nuclear weapons, have apparently prevented tens of thousands of deaths
55. The unclassified record summarized by Chuck
Hansen in his U.S. Nuclear Weapons: The Secret History
what appear to be nine nuclear weapon accidents involving release of nuclear
materials from weapons. The official summaries of each event do
not mention which, or how much, nuclear materials were released.
A tenth such accident, cited by Drell and Peurifoy in their Table 1, involved
the detonation of a bomb at Kirtland AFB in 1957; that bomb did not have
its plutonium pit installed when it was inadvertently dropped from the
airplane. Declassification of nuclear weapons accidents is not complete,
however, and it is possible that further accidents have occurred.
It is likely that cleanup crews were exposed to plutonium at some of these
accident sites. The accidents at Palomares, Spain (January 17, 1966),
and at Thule, Greenland (January 21, 1968), as well as possibly at Dyess
AFB, Texas (November 4, 1958) involved detonations of HE with resultant
nuclear contamination; these accidents appear to be the ones generating
the most difficult cleanup problems and the greatest likelihood of long-term
residual contamination. I am not aware of any study of morbidity
or mortality of the workers who cleaned up after these or any other accidents.
The cleanup crews, together with any undocumented population exposures
(at the time of the accident or in the long run), represent the primary
uncertainties in the claim that no one is known to have been injured from
the specifically nuclear aspect of U.S. nuclear weapons accidents.
56. These order-of-magnitude estimates are included
here, for all their uncertainty, because it has been observed that rational
thought often breaks down when nuclear weapons safety is discussed. This
occurs even, or perhaps especially, at the top-most levels of the DOE.
With the image of the mushroom cloud foremost in our minds -- particularly
in the minds of those who are responsible for nuclear weapons -- no amount
of funding for safety improvements seems too much. That, of course,
is because they have not been asked to choose between their own programs
and other societal means of reducing morbidity and early mortality. It
may also be because safety as a goal is psychologically compensatory for
those who participate in threatening other nations with weapons of mass
destruction, which is, after all, what nuclear deterrence is all about.
These estimates suffer not only from attempting to quantify what cannot
be quantified, but also from the implicit error of assuming that the risks
of low-probability, catastrophic events are comparable to high-probability,
less-severe events. An accidental nuclear explosion, or even a plutonium
accident, would arguably have a qualitatively much more severe long-term
effect on a society than a comparable number of automobile fatalities.
From 1950-1980, the rate of "Broken Arrow" accidents was approximately
one per year. Current safety standards require that a warhead or bomb
be able to endure such accidents with a 10-6
or less chance
of nuclear detonation. Applying the 1950-1980 rate of serious accidents
to the deployed START II arsenal, the probability of roughly 4,000 total
weapons experiencing one accidental detonation per year with some nuclear
yield greater than 4 pounds of TNT-equivalent is less than 4 X 10-3
The current accident rate is much smaller than the historical rate, however,
since (a) there are to be fewer deployed weapons than the height of the
Cold War, (b) these weapons are now flown around on airplanes much less
frequently than in 1950-1980, (c) dangerous exercises like airborne refueling
are surely no longer conducted with live nuclear weapons, and (d) these
weapons are not kept on alert aircraft parked on runways. These
four factors, taken together, probably reduce the rate of "Broken Arrow"
accidents by a factor of about 100. Our order-of-magnitude estimate of
the current accident rate might therefore be about 10-2
year, and of accidental nuclear explosion about 4 X 10-5
Most nuclear weapons are stored and transported away from urban areas,
so the probability of an accidental explosion in a city is quite small.
The size of an accidental nuclear explosion could, by definition, be anywhere
between 4 pounds TNT-equivalent and about 1.2 megatons (see Table 1).
people is a reasonable upper bound for the
number that would be likely to die from an accidental explosion; that
is approximately the number of fatalities that occurred when the center
of a city largely made of light buildings -- Hiroshima -- was bombed.
In other words, we might expect, in very round numbers, a reasonable
upper bound of 4 deaths per year from a nuclear weapon explosion, with
an expected number of deaths per year perhaps two orders of magnitude
below this, or 0.04 deaths per year.
This compares with the worst-case scenario drawn up by Fetter and von
Hippel for a plutonium dispersal accident, for which the corresponding
number was 1 death per year. The probability of the particular accident
scenario which Fetter and von Hippel analyzed has been greatly reduced
if not virtually eliminated, however, by the simple expedient of separating
warheads from rocket motors during loading. The possibility of a
plutonium-dispersing accident elsewhere than the Trident docks has not
entirely vanished, of course, despite the fact that all
non-Trident weapons have IHE and some of them have FRPs as well (see Table
1). Assume, arguendo
, that these changes have reduced the
expected annual deaths by a factor of 10 to 100, i.e. to about 0.01 to
It seems therefore plausible that the a priori
of annual deaths from a nuclear weapons accident is on the order of 0.1
or fewer persons/year, or roughly one million times less than the expected
number of deaths from other accidents. The probability of
dying in any
accident is, in turn, about 50 times less than dying
of a disease, which is about 0.98. Our rough estimate of the probability
of any given person in the U.S. dying from a nuclear weapons accident
in a given year is thus about 4 X 10-10
or less, or less than
3 X 10-8
in a 75-year lifetime. These numbers are very
low, and errors in them of up to three orders of magnitude will not affect
the conclusions which follow in the text.
57. See Stewart Udall, The Myths of August:
A Personal Exploration of Our Tragic Cold War Affair with the Atom
Pantheon, 1994. Dr. Hugh DeWitt of Lawrence Livermore has recently
drawn attention to the continued classification of historic fallout data
and called for U.S./Russian bilateral release of this data.
58. Radioactive Heaven and Earth
Press, 1991, available from the Institute for Energy and Environmental
Research, Takoma Park, Md.
59. Andrei Sakharov, "Radioactive Carbon from
Nuclear Explosions and Nonthreshold Biological Effects," 1958, reprinted
with Appendix by Frank von Hippel in Science and Global Security
1990, V. 1, pp. 175-187.