Jeffrey LewisClose Calls

I have a special Halloween column up at Foreign Policy on various nuclear crisis, accidents and other near-catastrophes.

As I said in the opening piece, I was struck at how much I remember correctly – but also how many things I didn’t know.  I am sure there will be a little error here or there, as well as some big omissions.  So, consider this an open invitation to swap nuclear horror stories.



  1. russiannavyblog (History)

    “Khey boyskis. Soviet Union ees no more. Time to go home and split bottle three ways. Just leaf dat thing wkhere eet ees. Will be okay!”

    In May 1995, in Tunnel 108, the last “nuclear device” on Kazakhstan territory was destroyed by specialists from Russia’s VNIITF with the assistance of Kazakhstani experts from the Ministry of Ecology and Bioresources. The operation was conducted under an agreement between Russia and Kazakhstan.[1] Boris Lebedev was in charge of the destruction. The explosion of the 0.3-0.4 kT undetonated device took place 425 feet below ground. (For more information on this explosive device see the 5/94 and 8/19/94 entries below.)[2] While preserving the classified nature of the technical specifications of the device, Kazakhstani scientists explained at a conference in 1997 that the device was not a bomb and not a warhead. It was a unique nuclear device constructed and placed in the Tunnel 108 in 1991. It was meant to be used within a few months after its emplacement. However, after the disintegration of the Soviet Union, political considerations prevented the detonation of the device. Experts determined that if left unexploded, the device could eventually contaminate the water table.[1] As shipping the device to Chelyabinsk-70 for disposal would have been exceedingly dangerous, it was decided to destroy it in place using chemical explosives.[3]

    [1] Emily Ewell, NISNP trip report”International Conference on Nonproliferation Problems,” September 1997, KAZ970900, p.7.
    [2] Bruce Pannier, “Kazakhstan Nuclear-Free,” OMRI Daily Digest, 1 June 1995; “Kazakhstan,” Central Asia Monitor, No. 4, 1995, p. 6.
    [3] Bruce Pannier, “Kazakhstan To Explode a Nuclear Device,” OMRI Daily Digest, No. 100, Part 1, 24 May 1995.

    • Jeffrey (History)

      Oooh. That’s a good one. Gaukhar will kill me when I see her next for forgetting that one.

    • Gaukhar (History)

      I’m not quite that violent. I think…

      Degelen project documents must be a fascinating read. There was a lone report in early 1995 that there might have been more charges left at the Semipalatinsk test site, but no follow-up reporting as best I can tell.

      Great article, Jeffrey!

      According to some reports, there are in fact three nuclear devices at the Semipalatinsk test site in addition to the well-known 0.3-0.4 kiloton nuclear explosive device that has been there since 1991. One 150 kiloton device is supposedly located in a horizontal tunnel, and the other two, of undisclosed yield, are in vertical shafts at a depth of 500 meters. (No other sources have reported the existence of any additional devices.)

      [PPNN Newsbrief, First Quarter 1995, p. 3.]”

  2. K L Mills (History)

    Jeffrey; you’ve got a typo: Baneberry’s yield was 10 kt. kilotons.

    • Jeffrey (History)

      Thanks. Didn’t even see that.

  3. Cthippo (History)

    PEPCON? Really? How does a spectacular industrial accident in a non-nuclear facility rate? If you want big booms then how about the Texas City blast in April 1947? PEPCON was only about 1 kT, but Texas City is rated at 2.7 kT AND took place in a heavily populated city. Failing that, the N-1 launch explosion at least involves a rocket.

    Likewise with the USS Belknap, an explosion in one point safe weapons aboard a ship at sea is probably going to make for a bad day, but is pretty minor in comparison to a lot of others on the list. Baneberry was spectacular, but compared to all the years we just set them off in the air and didn’t worry about the fallout… Castle Bravo going off at 2-3 times it’s anticipated yield seems scarier.

    How about the SL-1 blast? Pinning bodies to the ceiling with the control rod (correction, shield plug) from an exploding reactor is pretty ghoulish, especially when they were claiming for a while that it might have been intentional.

    • Jeffrey (History)

      I just love PEPCON and didn’t want to leave it out. Felt a little guilty about including it, but I figured everyone else would just enjoy it as a guilty pleasure.

      SL-1 is a good one. December.

  4. Pavel (History)

    The Kyshtym accident was in 1957, not in 1956.

    There were quite a few incidents with Soviet submarines. The most recent one was the Ekaterinburg submarine fire in December 2011 – there is a good chance that missiles and warheads were on board

    • Jeffrey (History)


  5. Kingston (History)

    Great piece! The brush Stanislav Petrov had with his early warning panel on September 26, 1983 was pretty scary:

  6. Bruce MacDonald (History)


    Your FP piece is excellent. Glad you included the Damascus, Arkansas event. I worked on nuclear issues at the State Dept at the time and that one kept us busy for a couple of weeks. A few more details: the explosion “launched” the Titan II warhead, reputedly with a yield of 9 megatons, about 300 yards into a nearby field. The airman working on the missile was using a huge socket wrench near the top of the missile, and it fell about 60-80 feet, bounced off the floor and punched a big hole in the missile. Fortunately for all of the crew, it leaked for about an hour before the missile exploded, so they had time to get away. All this less than two months before the 1980 election. The following year, Reagan tagged the Titan II fleet for dismantling. My take-away: one of the biggest nuclear threats we face is Murphy’s Law!

    • Jeffrey (History)

      Murphy is certified to handle nuclear weapons.

      I bet Senator Bumpers was not pleased.

  7. Chris (History)

    Great piece! Loved the dark humor as well. I was glad to see you inculded some of my “favorite” incidents– the Norwegian rocket incident, the various NORAD incidents… I would have to include the Suez crisis from November 1956 when NORAD “detected” a Soviet Strike, a British bomber had crashed, 100 MiGs were flying over Syria, a Soviet fleet was moving from the Black Sea to the Aegean Sea, and unidentified aircraft were in flight over Turkey (later to be determined to be a flock of geese.)

    I shiver to think how many other incidents have gone unreported– particularly in non-NPT signatory countries…

  8. George William Herbert (History)

    I also posted this off in the thread a FP, but…

    Mars Bluff (and a lot of other 1960s aircraft accidents) involved weapons with in-flight insertable fissile pits, which weren’t in the weapon when the accident happened.

    There were some bombs with pits in them lost or involved in accidents, some on earlier strategic bombers, the Navy accidents with loss (which were generally not IFI bombs).

    I personally would put three levels of severity. Increasing in worry:

    A nuclear system (including a bomb) but not an assembled nuclear weapon.

    A nuclear weapon (assembled).

    A nuclear command and control or early warning system.

    Yes, a training tape isn’t a nuclear bomb, but the people who see the screen outputs and respond to them have a whole bunch of them…

    • Jeffrey (History)

      That’s an important clarification, as in the case of the Tybee bomb the concern is the HEU secondary but not a missing plutonium pit.

    • George William Herbert (History)

      Right. The HEU in secondaries is a proliferation and environmental concern. The designs are a proliferation concern. The explosives are a environmental and explosive hazard. All of those are present if you lost a pit-less weapon. But it’s not going to detonate in a nuclear explosion.

      Fully assembled devices could detonate (however unlikely that is), which is a whole ‘nother magnitude of Bad Day, even if it’s in the middle of a desert or offshore somewhere. Obviously much worse if it’s on a ship in the middle of a battlegroup or in port or in a plane crash at a base or in inhabited areas…

    • Jeffrey (History)

      Although even if the pit is in place, the ESD should prevent nuclear yield.

    • George William Herbert (History)

      In $DAYJOB and $NIGHTJOB (IT consulting and aerospace engineering respectively) I study among other things reliability and risk factors.

      ESD, PALs, and modern strong link / weak link mechanisms and the like are supposed to prevent all sorts of things from going wrong.

      I simply categorically refuse to believe that inherent system reliability against all possible failures can be done better for those than for say aircraft, spacecraft, or reliable computing efforts. All of which are notable for interesting new failure modes, even for well established systems.

      I *think* – hopefully – that single-point safe weapons without valid arm codes and with suitably well built mechanisms should be remarkably hard to detonate by accident. But I can think of some pretty extreme failures (plane crashes in the rain, fireball engulfs weapon, rising fireball triggers lightning strike down through the weapon as the fire heats it up; someone accidentally perforates a weapon with a DU penetrator round from a tank gun, adding a heavy neutron reflector to the mix at 1,650 m/s…; two aircraft with weapons midair each other at supersonic speed, head on, with a nose-to-nose weapon contact bringing two pit assemblies into intimate contact at 800 m/s; meteorite impact; …; computer virus infects the embedded controller for the device…; main stage fire in a SLBM tube in a sub, ending in a massive electrical short through the warhead section from malfunctioning sub electrical systems because FIRE, dumping a lot of electrical energy into the firing leads simultaneously after the fire compromises the sealed safety barrier / hohlraum…).

      I can go on. Most people think I’m nuts for, this, but there are books and books full of computer/IT related failure stories and analysis, spacecraft accidents, military weapons accidents, etc. Most of the accidents fall into known categories, but many new things go wrong, with regularity.

      Just last year, Gulfstream found the hard way during testing a prototype G650 jet that in point of fact, some bizjet wings stall about 3 degrees lower angle of attack near the ground than in clear airstream, due to ground effect, vs a long-standing assumption that it was perhaps a degree worse but probably a degree better. Oops. All these years, and we’re still learning aerodynamics.

      A couple of days after the new Mars rover landed, a NASA experimental lander test rocket on its 31st test flight – the first disconnected from its safety tether – had its inertial measurement unit fail dead a second after liftoff, leaving it blind as to its attitude. After 30 completely good flights on a tether, hovering and moving around, with no sign of problems with the unit. As with Ariane 5 flight 1, this is usually a terminal failure for the vehicle.

      New York City went almost 400 years without a hurricane big enough to flood to the height of their 150-year-old subway system… Oops.

      I certainly don’t think the weapons labs are worse than industry averages in these other industries. But I don’t see evidence they’re orders of magnitude better, either. We have thousands of weapons.

    • Jeffrey (History)

      I share your perspective and am a big believer in “Normal Accidents” theory.

      The ValuJet crash is a perfect example of the limits to faced by even “high reliability” organizations.

  9. Bob K (History)

    My favorite is the Mark 17 that a B-36 inadvertently dropped near Albuquerque in May of 1957. The high explosives detonated. Luckily, there was no pit in the weapon.

  10. Jonah Speaks (History)

    In terms of sheer catastrophic potential, nothing beats a nuclear war with thousands of nuclear weapons aimed at real targets. Three events during the Cold War created the greatest risk:

    Cuban missile crisis – potential for accidental or unauthorized nuclear use, plus stupid and risky decisions by national leaders, any of which could have resulted in nuclear war.

    Able Archer military exercise – Soviet leaders suspect NATO simulation of nuclear war is merely cover for a nuclear sneak attack. American leaders did not have a clue that Soviet leaders were this paranoid.

    Stanislav Petrov refused to report satellite detection of five American missiles headed toward Soviet Union. Petrov claimed he was “just doing his job” by calling it false alarm. Ungrateful superiors transferred him to a different job.

    The above three incidents do not exhaust all the main possibilities for nuclear war, nor do they appear to fall within a clear pattern of risk, so that some simple remedy might cure or prevent most future nuclear wars.

  11. bradley laing (History)

    —When a refinery belonging to BP exploded in Texas City, Texas, the people involved were violating all of the written safety rules. Safety had been neglected, systematically, for years, by cost cutting.

    —How do we measure institutional decay of safety prodeedures, as a potential nuclear accident cause?

  12. JohnLopresti (History)

    This hopefully will add a gloss to the Cuban missle crisis subtext in the hilarious Foreign Policy Scare-Night eve of eve of All Hallows article. I thought the US President an adequately nimble and sufficiently advised amateur historian to manage the postbellum Г. Никита Сергейвич, and the barbudo Fidelito; though, true, there was a vibrant Кремль which provided context for Хрущёв. Each of the latter two heads of their respective states, were no match for the US State Department. Media coverage was more ‘reality’ programming than hype in that time. I was on home leave from a nuclear physics training program, of sorts, mid underclass in college.

    I highly recommend the exploratory thinking by newly appointed professor Mary Dudziak, Emory College Atlanta, in this regard, broadly speaking, for its new perspectives on human factoring in defining the concepts of war in legal and social terms.

    Draft chapter from prof. Dudziak’s recent book:

    Faculty page:

    Television was fairly nascent at the time.
    The US intell service also was in expansion mode then.
    Foreign policy was morphing. A post-Nagasaki order was developing.
    Communications were slow. It was pre-DARPA-internet This was 20 years prior to Iran-Contra, whose communications often occurred by closed-circuit telex.

    Although scary, the Cuban missle crisis was a mere blip, and was well managed in world diplomacy, much of which content remains classified secret because of impingements upon other, collateral world events then in the offing.

Pin It on Pinterest