The final switch: Goldsboro, 1961

by Alex Wellerstein, published September 27th, 2013

The threat of nuclear weapons accidents isn't a new one. Even in 1945, Los Alamos physicists sweated when contemplating all that could possibly go wrong with their bombs, if they went off at the wrong place or the wrong time. Or didn't go off at all. That's the bind, really: a nuclear state wants a weapon that always goes off exactly when you tell it to, and never goes off any other time. That's a hard thing to guarantee, especially when the stakes are so high in both directions, and especially since these two requirements can be directly in tension.

Schlosser - Command and Control book

I recently heard Eric Schlosser give that elegant formulation at a talk he gave last week in support of the release of his new book, Command and Control: Nuclear Weapons, the Damascus Accident, and the Illusion of Safety. I haven't had a chance to read the book, yet (it's currently en route), but I'm looking forward to it. I read Schlosser's Fast Food Nation a decade (!) ago and found it completely eye-opening. But I went to his talk last week not sure what to expect. From McDonald's to nuclear weapons accidents? Stranger things have happened, but I worried that maybe he would take the "easy" route with regards to the accidents, not bothering to learn to nitty-gritty technical details that let one talk about such things sensibly, or, at the very least, sensationalize the findings. So I was pretty pleased to find that neither seemed to be the case. Schlosser has seriously done his homework, spending 6 years digging through records, FOIAing documents, and interviewing weapons designers. His discussion of the risks seemed right on the mark so far as I could tell — they don't need to be exaggerated one bit to be perfectly horrifying. He answered questions expertly, even a tough, devil's-advocate one from Hugh Gusterson. So I've been looking forward to reading the full book.

Last week, the Guardian released a new document, obtained by Schlosser through a FOIA request, regarding one particular accident, the 1961 crash of a B-52 near Goldsboro, North Carolina, which resulted in the jettisoning of two Mark-39 hydrogen bombs. The document in question is a government nuclear expert's evaluation of a popular account of the Goldsboro accident, in which he finds some major errors (like overstating the yield of the bomb), but ultimately concludes that at least one of the bombs was, in fact, pretty damned close to accidental detonation: "one simple, dynamo-technology, low voltage switch stood between the United States and a major catastrophe ... It would have been bad news – in spades."

The bomb in question, stuck in the mud.

The bomb in question, stuck in the mud.

I've been watching how the above document has been discussed by people on the web. The most interesting response has been people saying, "I thought that bomb lacked a nuclear core?" You know that there have been too many nuclear weapons accidents when people start getting them confused with one another. The missing-bomb-that-maybe-lacked-a-core is the 1958 Tybee bomb, where a Mark-15 hydrogen bomb was lost near Savannah, Georgia. Different bomb, different day.

The other response I commonly saw was one that assumed that any such fears of a bomb going off accidentally were exaggerated. Now this is kind of an interesting response. For the one thing, they're discounting a contemporary, internal, once-classified evaluation made by a relevant expert. In exchange, they're parroting either general skepticism at the idea that a nuclear weapon could technically be unsafe, or they are parroting a standard line about how hard it is to set off an implosion bomb accidentally, because all of the lenses need to detonate at exactly the same time. Which is sometimes the right approach (though not all American bomb designs were "one-point safe" — that is, there were designs that ran a real risk of producing a nuclear yield even if just one of the explosive lenses accidentally fired), but in this case, it's entirely irrelevant, for reasons I'll explain below.

I've been in touch with Schlosser since the talk, and he shared with me a video he had (somehow) gotten his hands on produced by Sandia National Laboratory (the weapons lab that specializes in making bombs go off at just the right moment) about the Goldsboro accident. He's put it up on YouTube for me to share with you. It is only a few minutes long and worth the watch.

I love the CGI — "all the sudden, now that weapon system is free." The bomb looks so... liberated. And the part at the end, where they talk about how they had plenty of opportunities for future data, because there were so many accidents, is wonderfully understated. But the stuff that really hits you in your gut is the description of exactly what happened:

"All of the sudden now that weapon system [the Mk-39] is free. As the weapon dropped, power was now coming on, and the arming rods were pulled, the baroswitches began to operate.1 The next thing on the timing sequence was for the parachute to deploy. When it hit the ground, it tried to fire." "There was still one safety device that had not operated. And that one safety device was the pre-arming switch which is operated by a 28 volt signal." "Some people could say, hey, the bomb worked exactly like designed. Others can say, all but one switch operated, and that one switch prevented the nuclear detonation." "Unfortunately there had been some 30-some incidents where the ready-safe switch was operated inadvertently. We're fortunate that the weapons involved at Goldsboro were not suffering from that same malady."

What's amazing about the above, in part, is that everything in quotation marks is coming from Sandia nuclear weapons safety engineers, not anti-nuclear activists on the Internet. This isn't a movie made for public consumption (and I've been assured that it is not classified, in case you were wondering). It's a film for internal consumption by a nuclear weapons laboratory. So it's hard to not take this as authoritative, along with the other aforementioned document. Anyone who brushes aside such concerns as "hysterical" is going to have to contend with the fact that this is what the nuclear weapons designers tell themselves about this accident. Which is pretty disconcerting.

There are further details in another document sent to me by Schlosser, a previously-classified review of nuclear weapons accidents from 1987 that clarifies that one of the reasons the Goldsboro bomb in particular almost detonated was because of the way it was tossed from the aircraft, which removed a horizontally-positioned arming pin. That is, an arming pin was supposed to be in a position that it couldn't be removed accidentally, but the particulars of how violently the aircraft broke up as it crashed were what armed the bomb in question. The other bomb, the one whose parachute didn't fire, just had its HE detonate while it was in the mud. From the 1987 review:

Before the accident, the manual arming pin in each of the bombs was in place. Although the pins required horizontal movement for extraction, they were both on a lanyard to allow the crew to pull them from the cockpit. During the breakup, the aircraft experienced structural distortion and torsion in the weapons bay sufficient to pull the pin from one of the bombs, thus arming the Bisch generator.2 The Bisch generator then provided internal power to the bomb when the pullout cable was extracted by the bomb falling from the weapons bay. The operation of the baroswitch arming system,3 parachute deployment, timer operation,4 low and high voltage thermal batteries activation, and delivery of the fire signal at the impact by the crush switch all followed as a natural consequence of the bombing falling free with an armed Bisch generator. The nonoperation of the cockpit-controlled ready-safe switch prevented nuclear detonation of the bomb. The other bomb, which free-fell, experienced HE detonation upon impact. One of the secondary subassemblies was not recovered.5

The secondary subassembly is the fusion component of the hydrogen bomb. Normally I would not be too concerned with a lost secondary in and of itself, because bad folks can't do a whole lot with them, except that in this particular bomb, the secondary contained a significant amount of high-enriched uranium, and lost HEU is never a good thing. The government's approach to this loss was to get an easement on the land in question that would stop anyone from digging there. Great...

Mk-39 ready-safe switch

From the video, I was also struck by the picture of the ready-safe switch then employed. I'd never seen one of these before. Presumably "S" means "safe" and "A" means "armed." It looks ridiculously crude by modern standards, one little twirl away from being armed. This little electronic gizmo was all that stood between us and a four megaton detonation? That's a wonderful thing to contemplate first thing in the morning. Even the later switches which they show look more crude than I'd prefer — but then again, probably all 1950s and 1960s technology is going to look crude to a modern denizen. And again, just to reiterate, we're not talking about "merely" accidentally igniting the explosives on the primary bomb — we're talking about the bomb actually sending a little electrical charge through the firing circuit saying "Fire!" and starting the regular, full-yield firing sequence, stopped only by this little gizmo. A little gizmo prone to accidentally firing, in some of the bombs.

Lest you think that perhaps Sandia overstates it (which seems rather unlikely), take also the testimony of Secretary of Defense Robert McNamara into account. In January of 1963, McNamara explained at a meeting between the Defense and State Departments that he was opposed to Presidential pre-delegation of nuclear weapons in part because of the danger of accidental detonation — either ours or the Soviets'. In the meeting notes, posted some time back by the National Security Archive and forwarded to me by Schlosser, McNamara's participation is listed as follows:

Mr. McNamara went on to describe the possibilities which existed for an accidental launch of a missile against the USSR. He pointed out that we were spending millions of dollars to reduce this problem to a minimum, but that we could not assure ourselves completely against such a contingency. Moreover he suggested that it was unlikely that the Soviets were spending as much as we were in attempting to narrow the limits of possible accidental launch. He went on to describe crashes of US aircraft[,] one in North Carolina and one in Texas, where, by the slightest margin of chance, literally the failure of two wires to cross, a nuclear explosion was averted.

This one's interesting because it embeds these accidents in a context as well — the possibility of either us, or the Soviets, accidentally launching a nuke and wondering if a full-scale nuclear exchange has to follow. It's not quite Strangelovian, since that would require a rogue commander, but it is very Fail-Safe.

As to what the Goldsboro blast would look like, the only time we tested this warhead at full yield was the shot "Cherokee" at Operation Redwing, in 1958. It was a pretty big boom, far more impressive than some of the Hiroshima shots that have been posted along with the Goldsboro story:


And, of course, you can use the NUKEMAP to chart the damage. I've added the W-39 warhead to the list of presets in NUKEMAP2, using 4 megatons as the yield (the tested yield was 3.8 megatons, though the W-39 is often stated as an even 4. I rounded up, just because quibbling over 200 kilotons seemed pointless), and a fission fraction of 55%.6 It's a pretty big explosion, with a fallout plume capable of covering tens of thousands of square miles with hazardous levels of contamination (and nearly a thousand square miles with fatal levels). Note that the Cherokee test was a true airburst (the fireball didn't touch the ground), and so didn't generate any significant fallout. The Goldsboro bomb, however, was meant to operate on impact, as a surface burst, and would have created significant fallout.

Again, one doesn't have to exaggerate the risks to find it unsettling. The bomb didn't go off, that final switch thankfully did work as intended. But that's cold comfort, the more you learn about the accident. Our current nuclear weapons are much safer than the Mk-39 was, back in 1961, though Schlosser thinks (following the testimony of experts) there are still some unsettling aspects about several of our weapons systems. If we are going to have nukes, he reasons, we should be willing to spend whatever it costs to make sure that they'll be safe. That seems to me like an argument guaranteed to appeal to nobody in today's current political climate, with the left-sorts wanting no nukes and no modernization, and the right-sorts not really wanting to talk about safety issues. But I'll get to that more another day, once I've read the book.

If that bomb had gone off, we'd speak of "Goldsboro" as a grim mnemonic, in the same way that we do "Chernobyl" today. One wonders how that would have changed our approach to nuclear weapons, had the final switch not held strong.

  1. The "arming rods" were pull-out switches that would activate when the weapon left the bomb bay. The baro(meter) switches were pressure sensitive switches that would make sure the bomb was nearing the appropriate height before starting the next sequence of arming. In the World War II bombs, the next stage in the sequence would be to consult radar altimeters to check the precise distance from the ground. The Goldsboro bombs were set to go off on ground impact. []
  2. A Bisch generator, as the context implies, is an electrical pulse generator. []
  3. Again, a pressure-sensitive switch that tried to guarantee that the bomb was roughly where it was supposed to be. []
  4. Timer switches were often used to make sure that the bomb cleared the aircraft before seriously starting to arm. []
  5. R.N. Brodie, "A Review of the US Nuclear Weapon Safety Program - 1945 to 1986," SAND86-2955 [Extract] (February 1987). []
  6. Chuck Hansen, in his Swords of Armageddon, estimates that shots Cherokee and Apache of Operation Redwing had an average fission fraction of 55%, but isn't able to get it any more precise than that. Given what I've read about the bomb — that it used an HEU secondary, for example — I would expect it to be at least 55%, if not more. It seems like a pretty "dirty" weapon, emphasizing a big yield in a relatively small package over any other features. See Chuck Hansen, Swords of Armageddon, V-224 (footnote 325). []

Tags: , , , , ,

36 Responses to “The final switch: Goldsboro, 1961”

  1. Paul Guinnessy says:

    One also wonders whether people would really believe it was an accident, or a first strike instead.

    • It would be an odd “first strike,” though — why North Carolina with a single 4 megaton gravity bomb, without any indication of an approaching bomber? I mean, I’m sure that wasn’t some kind of standard war plan.

      • Puncheex says:

        Ever read Clancy’s Sum of All Fears? Sometimes what exactly has happened isn’t important, but rather what perceptions he has, particularly in a time-restricted panic scenario.

  2. […] *Historian of science Alex Wellerstein discusses the vulnerability of nuclear weapons in his recent blog post, “The final switch: Goldsboro, 1961.” […]

  3. Peter says:

    As to the fallout spread shown on Nukemap, am I correct in interpreting the darkest, 1,000 rads/hour shading as the lethal zone, while the 100 rads/hour shading as being the area in which the fallout would be dangerous, requiring sheltering, but not necessarily deadly?

    • A single 1,000 rad exposure will kill you dead. A single 100 rad exposure is not good for you but not fatal. But exposures in the 200-1,000 rad range are still very dangerous, and can make you sick at the very least. We’re talking about 100 rads an hour, so the potential for higher exposures than 100 rads are possible in that region. (One future addition to the NUKEMAP would be something that would let you calculate a total exposure based on a given point within that curve, but I haven’t written it up yet.) So I would consider that to be a pretty dangerous region as well (especially since these curves are idealized) — you’d definitely want to be sheltered, or moving out of it very fast.

      • Peter says:

        Thanks. If one more question is okay, what about the 10 rad/hour exposure in Washington? Would that have required any sheltering?

        By the way, I started reading Command and Control a couple days ago. So far it’s very good. I found the account of the 1980 Titan blast in Arkansas quite telling in one respect: the Air Force had all sorts of extremely detailed procedures to deal with every aspect of missile maintenance and control, but when things went very wrong they were largely clueless.

        • 10 rad/hour is worth staying indoors for when it arrives, just to avoid unnecessary exposure, but it is not a very high dose and would decay pretty quickly anyway. This FEMA report has a good discussion of what to do with various exposures.

          • Peter Solem says:

            I’ve always wondered a bit about the ‘rad’ measurement for radiation exposure, given (1) that alpha-, beta- and gamma-type radiations have different penetrating power and effects and (2) so much depends on whether or not the radioactive elements/molecules (and which ones out of the dozens of fission products) are ingested into the body or not.

            In modern usage, ‘rads’ have been replaced by ‘grays’, defined as “the absorption of one joule of energy, in the form of ionizing radiation, per kilogram of matter.” Notice that absorption of one joule of energy in the form of non-ionizing radiation (say, sitting around a fireplace absorbing infrared radiation) is an equivalent amount of energy, but has a very different effect. Likewise, do alpha/beta/gamma forms of ionizing radiation have differing effects?

            It’s also worth remembering that the structure and function of DNA was basically unknown throughout the early period of nuclear weapons development, and the role of ionizing radiation in instigating base pair mutations was also unknown. If such mutations take place in critical regions of the genome involved in cell cycle control, cancer or other genetic malfuctions are the long-term result.

  4. McNamara’s reference in 1963 to an aircraft crash involving another nuclear close call is puzzling because there are otherwise no public records (as of yet) describing such an accident. Douglas Keeney, who has spent a lot of time investigating these accidents, wrote about this a few years ago –

    • Schlosser says it is a B-47 carrying another Mk-39 that released it near Dyess AFB, near Abilene, TX:

      The B-52 crash in North Carolina wasn’t the only accident that involved fully assembled, sealed-pit weapons—and McNamara soon learned about others. A B-47 carrying a Mark 39 bomb had caught fire while taking off at Dyess Air Force Base, near Abilene, Texas. At an altitude of about two hundred feet, the pilot realized the plane was on fire, banked to avoid a populated area, and ordered the crew to bail out. Three of the four crew members got out in time. The high explosives of the hydrogen bomb detonated but didn’t produce a nuclear yield.

    • Kelly M says:

      The Abilene Broken Arrow is not that secret. There over two dozen Broken Arrows just up to that point in 1963.

      I’ll copy paste my Broken Arrow reference list here, as you can see most are aircraft related:
      February 13, 1950: A B-36 jettisons its Mark 4 off the west coast of Canada before crashing. Weapon HE detonates in midair, no core.

      April 11, 1950: A B-29 crashes into a mountain after takeoff from Kirtland AFB, New Mexico with a Mark 4 on board, and the area it crashed into was a nuclear weapons storage area as well. Weapon burns, no core.

      July 13, 1950: A B-50 crashes near Lebanon, Ohio with a Mark 4 on board. Weapon HE detonates, no core.

      August 5, 1950: A B-29 crashes and burns on takeoff while carrying a Mark 4 with no core at Suisun AFB California, weapon HE detonates in the fire killing 19, including General Travis. Base now named Travis AFB.

      November 10, 1950: A B-50 jettisons a Mark 4 over the St. Lawrence River in Canada after aircraft trouble, weapon HE detonates in midair, no core.

      March 10, 1956: A B-47 carrying two nuclear cores in birdcages disappears over the Mediterranean Sea near Lagos, Portugal. Wreckage is never found. ((This one is probably the most alarming incident I’ve encountered))

      July 27, 1956: A B-47 crashes into a weapon storage igloo at RAF Lakenheath, England containing three Mark 6 bombs with no cores. Weapons were damaged but did not detonate.

      May 27, 1957: A Mark 17 falls through the bomb bay doors of a landing B-36 just outside the fence line of Kirtland AFB New Mexico and HE detonates on impact. Core was onboard aircraft, not in weapon.

      July 28, 1957: A C-124 jettisons two (probable) Mark 5s after loss of engine power off the coast of New Jersey. The weapons did not detonate and were not recovered.

      October 11, 1957: A B-47 crashes on takeoff at Homestead AFB Florida while carrying a Mark 15 in ferry configuration and a core in a birdcage. The bomb was partially destroyed in the ensuing fire, the birdcage was recovered with minimal damage.

      January 31, 1958: A B-47 carrying a Mark 36 in strike configuration crashes on takeoff at on overseas base, purportedly Sidi Slimane AB in French Morocco. The plane burned for 7 hours, no detonation took place. Localized contamination was cleaned up.

      February 5, 1958: A B-47 collides with a F-86 near Savannah, Georgia. The B-47 jettisons its Mark 15 over Wassaw Sound before landing at Hunter AFB. The weapon does not detonate and was never recovered. No core.

      March 11, 1958: A B-47 accidentally jettisons a Mark 6 near Florence, South Carolina. The weapon HE detonates on the property of Walter Gregg, damaging his house and causing several injuries. No core. The crater can still be seen today.

      November 4, 1958: A B-47 caught fire on takeoff and crashed near Dyess AFB Texas. The Mark 39 bomb HE detonated on impact in a field near Butterfield School south of Abilene.

      November 26, 1958: A B-47 caught fire on the ground at Chennault AFB Louisiana. The nuclear weapon, undisclosed but a sealed pit variety possibly a Mark 15 or Mark 39, is destroyed in the fire. Local contamination was cleaned up.

      January 18, 1958: A F-100 parked with a nuclear weapon, likely a Mark 7 with no core, caught fire and burned after an explosion in an external fuel tank. Undisclosed location, likely Osan AB in Korea. No detonation.

      July 6, 1959: A C-124 crashed on takeoff from Barksdale AFB Louisiana with three Mark 29s on board. One weapon is destroyed in the fire, no detonations occur.

      September 25, 1959: A P-5M ditched in Puget Sound off Whidbey Island Washington while carrying an unarmed nuclear antisubmarine weapon. Aircraft and weapon were not recovered.

      October 15, 1959: A B-52 collides with a KC-135 refueling tanker near Hardinsberg, Kentucky. Two Mark 15 weapons were recovered from the crash site mostly intact.

      June 7, 1960: A BOMARC air defense missile explodes at McGuire AFB, causing high local contamination as the core burned and melted. The base is eventually closed and remains fenced off today, the contamination making it one of the best preserved BOMARC sites remaining.

      January 24, 1961: A B-52 loses its right wing and crashes near Faro, North Carolina, deploying two Mark 39 weapons. One disintegrates hitting a swamp at high speed, all but the secondary is recovered. The second lands via parachute and is recovered intact.

      March 14, 1961 A B-52 experienced failure of crew compartment pressurization and defended in altitude. Increased fuel consumption caused fuel exhaustion before rendezvous with a tanker. The crew bailed out and the B-52 crashed, tearing loose two Mark 39s. No detonation.

      June 3, 1962: A Thor ICBM with a W-50 warhead, launched from Johnston Island for the Bluegill event was destroyed 22 miles downrange after range radar failed. Debris fell to the ocean.

      June 20, 1962: A Thor ICBM with a W-49 warhead, launched from Johnston Island for the Starfish event was destroyed between 30k-35k feet after rocket motor failure. Plutonium contaminated debris rained on the Island and just offshore, some recovered by divers.

      July 25, 1962: A Thor ICBM with a W-50 warhead, launched from Johnston Island for the Bluegill Prime event was destroyed on the launchpad after an engine malfunction. Extensive damage and alpha contamination to the launchpad and surrounding areas.

      October 15, 1962: A Thor ICBM with a W-50 warhead, launched from Johnston Island for the Bluegill Double Prime event was destroyed 156 seconds after launch due to engine malfunction. Some radioactive debris fell back onto the Island.

      November 13, 1963: A storage igloo containing HE components of obsolete nuclear weapons detonates at Medina Base, Texas.

      January 13, 1964: A B-52 comes apart in a heavy winter storm over Cumberland, Maryland. Two Mark 53’s were recovered intact from the aircraft wreckage. Crew members are scattered as they eject in the storm, two died of exposure, two survived finding shelter in the blizzard, one did not eject in time and died on impact.

      December 5, 1964: A LGM-30B Minuteman I missile’s retrorocket fires, separating the reentry vehicle from the missile and sending it crashing to the bottom of the silo. No arming or detonation took place. Ellsworth AFB South Dakota, Launch Facility L-02.

      December 8, 1964: A B-58 loses control on taxi at Bunker Hill (now Grissom) AFB Indiana and catches fire as it strikes light fixtures and electrical boxes at the edge of the runway. Four Mark 43s and a BA-53 were onboard, portions of all burned. No detonation, localized depleted uranium contamination.

      October 11, 1965: A C-124 catches fire while fueling on the ground at Wright-Patterson AFB Ohio. Packaged tritium bottles, explosive generators and a training Mark 53 were destroyed in the fire. Minor local contamination.

      December 5, 1965: An A-4 loaded with a Mark 43 rolls off the USS Ticonderoga 80 miles off the Japanese Ryuku Islands, 200 miles east of Okinawa. Plane, pilot, and weapon are lost.

      January 17, 1966: A B-52 collides with a KC-135 while refueling over the Mediterranean while carrying four Mark 28FI weapons. Two HE detonated on impact near Palomares, Spain, one was recovered intact on land, one was recovered after a lengthy search from the sea, utilizing submersibles Alvin, Aluminaut, And CURV-III.

      January 21, 1968: A B-52 crashed on sea ice 7 miles from Thule AFB in Greenland carrying four Mark 28FIs, which HE detonated in the fire. Weapons components were recovered from the burned wreckage during the decontamination operation.

      Spring 1968: Sinking and loss of the USS Scorpion in the Atlantic with two Mark 45 ASTOR torpedoes armed with W-34 nuclear warheads.

      September 19, 1980: A Titan II ICBM located near Damascus, Arkansas had its fuel tank was ruptured by a dropped heavy wrench socket, causing a fuel leak. Vapors ignited and exploded 8 hours later, destroying the missile and its silo. The damaged but mostly intact 8 megaton W-53 warhead and re-entry vehicle were recovered 500 feet from the silo.

      Events not listed in the Official DOE Summary of Nuclear Weapons Accidents but meet the criterial for broken arrows:
      October 15, 1964 W31 Nike Hercules
      1949-1953 Mark 4 in a Beta Crate fell to the ground, severely damaged
      1951-1962 Mark 6 Beta Crate
      1951-1962 Mark 6 dropped from C-124 during loading, severely damaged
      1951-1962 Mark 6 dropped from C-124 during loading, severely damaged
      Unknown date, unknown weapon accidentally salvoed from parked aircraft
      1952 unknown weapon accidentally salvoed
      March 1956 Loring AFB ME Mark 17 dropped through a B-36’s bomb bay doors while stopped on runway
      1956-1963 Mark 1 BOAR W7 impact damage from jet blast hurling a crash kit into the weapon, unknown location
      1956-1966 Mark 39 accidentally salvoed during ground sunup, weapon damaged
      1957-1964 T-4 Atomic Demolition Munition fire damage in storage building
      July 19, 1957 Mark 15 jettisoned from A3D Skywarrior before aircraft ditched off Jacksonville Florida, not recovered
      February 8, 1958 Mark 15 B-52 weapon salvoed during unloading operation at Ellsworth AFB, after body and fins damaged
      May 1958 Mark 21 dropped from B-52 or B-47 during loading, internal and external damage
      April 1960 W-31 Honest John warhead dropped from crane, possible tritium leak internally
      July 31, 1958 Possibly W7/AF at dropped during handling operations Missile Test Center
      1958-1967 Multiple Mark 28s damaged as weapons space onboard and aircraft carrier flooded due to ruptured fire main
      Dec 13, 1960 Unknown weapon, location. Possibly accidental ignition of rocket motor
      October 27, 1961 Italy W49 Jupiter, tritium leak, Giola del Colle, Italy
      Nov 6, 1961 Plutonium release from ruptured pit tube of W47 warhead, Pantex Plant
      July 19, 1962 W49 Jupiter, lightning strike damaged ablative material, electrical systems, overseas, potentially Giola del Colle, Italy
      Dec 15, 1962 Mark 25 warhead in an AIR-2A Genie missile was attached to a Herman Nelson Heater during checkout on a F-106. Temperature went unregulated and melted the high explosives in the missile
      Mar 28, 1964 Mark 101 Depth Bombs with Mark 34 warhead and Mark 57s were flooded at a weapon storage magazine, probably in NAS Kodiak Alaska due to tsunami
      November 17, 1964 W-28 Hound Dog on a B-52G received frag damage from an engine starter that disintegrated, possibly in Clinton-Sherman AFB, Oklahoma
      July 15, 1965 W28 Mace Missile installation suffered repeated lightning strikes, internal damage. Unknown location
      January 19, 1966 W45 Terrier missile dropped from mount due to failure of holding pin, warhead dented and cracked. Probable location NAS Mayport Florida
      February 7, 1966 Mark 28FI weapon raised in a hoist for painting, left unattended for a lunch break wherein the hoist failed and the weapon fell on the floor. Fuse and fins damaged. Unknown location, probably a SAC base, AVDS depot, or AEC facility
      March 25, 1967 Aboard ship at sea, potentially off coast of Vietnam DMZ aboard USS Ozbourn. Enemy mortar shell exploded in ASROC compartment, damaging the missiles
      May 17, 1989 A tritium reservoir activated, contaminating an unknown weapon and 9 pits at Pantex plant

      • Cthippo says:

        One correction to your otherwise excellent list…

        the 9-25-59 incident involved a navy P-5M which was based at NAS Whidbey Island, but actually crashed about 100 miles west off the Columbia River. The crew was rescued by a Coast Guard cutter out of Astoria OR.

      • The additional Broken Arrow list by Kelly M is derived from “Broken Arrow, Volume II” by myself and James C. Oskins (ISBN 13:9780557655939), our book is available from and other booksellers . The W31/Nike Hercules incident did not involve a war reserve weapon, it was an operational suitability test unit with no nuclear material. We were going to print as we received the new info, and didn’t have time to alter the book content.
        We have uncovered two additional US nuclear weapons accidents, more information forthcoming in our next book.

      • Rolando Santos says:

        One question regarding this Broken Arrow list: the B-47 crash on March 10, 1956 is situated near Lagos, Portugal, but every other site or document that I read regarding this case says that the crash occurred in the Med, off Algeria. Can you sort this out?

        • Mike Maggelet says:

          Jim Oskins, my co-author, received the aircraft accident report several years ago. It shows that the flight had circled in a racetrack pattern awaiting their refueling aircraft (KC-97’s); the B-47 in question crashed a few miles into the Med, just SE of Gibralter, and NE of Cueta in several thousand feet of water (in any case the plutonium capsules have dissolved in place inside their birdcages due to the corrosive action of seawater). The location is in the general vicinity of where the HMS Ark Royal was sunk in 1941.

  5. John Cloud says:

    My father was an Army Air Force bomber pilot in WW II so good he was made a key instructor to train others, and never saw combat. Through the late 50s until, I think, 1961 he was in the Air Force Reserves, not in SAC but associated with it through bomber pilot training. He and his coterie of pilots were openly contemptuous of every claim of nuclear weapons safety they ever heard expressed, based on their direct experiences and stories from close associates they trusted. They wouldn’t/couldn’t go in great detail, which was very frustrating to their sons and daughters, but their insights were pretty apparent. “Just a matter of time” was a phrase they threw around a lot.

    Early into the Cuban missile crisis, we were living on the farm/ranch near Avondale, 19 miles out into the Colorado plains from Pueblo, where my father was a banker. Straight north of there were then somewhat nameless Air Force sites, now Schreiver AFB and some other sites. It was a crisp October morning, really early after a clear night. There was a great field of pumpkins covered with frost, very beautiful. My father was driving just me in the Jeep, as my sisters were sick with the flu. He was to drop me off at parochial school and go to the bank. We were listening to the news on the car radio, my father’s face gaunt. At one point he looked over at me, and said: “Take a good look at those pumpkins; we might not be coming back tonight”. Something strange entered me right then: the world was so very beautiful, and it could all disappear in a moment. Any sight of a field of pumpkins brings it back. Makes Halloween interesting…

  6. Cory Newman says:

    Alex, I’m sure you are aware of this, but I wanted to give the history channel link anyway concerning the Goldsboro Bomb and the Tybee Bomb. I saw this episode about the search for the Tybee Bomb on the History Channela few weeks past. I was not that suprized by this revelation about the Goldsboro bomb as I am a regular follower of this blog.

  7. Bill says:

    Great post. That McNamara quotation is fascinating not only because of his acknowledgement of the Goldsboro event and the one in Texas. What is also striking is his rejection of predelegation; yet this is the same person who the following year duly informed President Johnson that the Defense Department had updated the earlier predelegation arrangements and here is the latest iteration (which is still largely classified). So he knew it was a bad idea but didn’t want to go against it probably for political and instiitutional reasons.

  8. Bradley Laing says:

    1.) Was “the hot line” set up yet, to inform the Soviets that an *accidental* explosion took place.

    2.) Would the always suspicous Soviets refuse to believe what they were told, and reach other conclusions?

    3.) Would assorted people within the United States be convinced that a “Commie Kamiakzi Double-Agent” had set off the bomb. And which ones?

  9. Kelly M says:

    Its worth noting that they only gave up on recovering the secondary on the second bomb after two months spent excavating a 200′ diameter, 42′ deep hole looking for it, in a swamp that was constantly collapsing and flooding. They calculated that given the weight and shape of the secondary it was probably 180′ deep and improbable if not impossible to recover.

    Photos of the area can be seen here

    And I’m still not quite sure what was new in the documents released that caused all this recent media attention, everything that has been published online so far I had already read about in detail in The Goldsboro Broken Arrow by Joel Dobson, which has been sitting on my bookshelf for almost 2 years.

    • Peter says:

      It also helps that the site is right off a public road rather than in some remote location. If there were a bunch of Arabs operating construction equipment on the site, people would notice and questions would be asked.

    • Joel Dobson says:

      Thanks, Kelly M.
      In all the discussion about the Goldsboro Broken Arrow, it needs to be pointed out that none of the safety devices failed. In fact, they all worked exactly as designed. If the Radar/Nav’s DCU-9/A for Bomb No. 1 had, for some reason, been set to GRN or AIR instead of SAFE, we would have had the worst man-made disaster in history.
      But none of the controls were set to arm the bombs before the bail-out. So says, the surviving third pilot.
      On bomb No. 2, the one in the ground, the ARM/SAFE switch DID indicate ARM, so says the Tech Sgt EOD that found it. Luckily, it was the indicator that was distorted by ground contact, not the internal switch itself.
      Each bomb was 3.8 M/T, enough to change the coastline of NC, so says Dr. Jack Revelle, the EOD commander who ‘safed’ both weapons.
      “We thought the world had ended,” so says the women of Faro, NC, the small community north of Goldsboro.

  10. […] was itself error-prone. The bomb’s yield was probably in the two-to-four megaton range. Check out Alex Wellerstein’s post about Goldsboro and his NUKEMAP to see the weapon effects from Goldsboro to Richmond that were narrowly […]

  11. Mark Gubrud says:

    30 incidents in which the ready-safe switch was “operated inadvertently” (human error?), compared with all the unit-hours where it was not, suggests a low (no, not low enough, but low) probability that it would have been operated in this incident. Of greater concern is that the breakup of the plane might have resulted in a short-circuit operating the switch. That would be substantially likely if operating the switch was only a matter of energizing a single wire with no code sequence involved. Otherwise, it would be less likely if there were multiple wires or a required sequence. Anybody know about this?

    The claim that the bomb “tried to fire” is rather animistic.

    • I don’t think “tried to fire” is that animistic — it conveys the fact that all of the other firing aspects ran smoothly, when ideally they wouldn’t have. (Like the other one.)

      One of the nice bits from Schlosser’s talk was a discussion of the reliability numbers. They are very reassuring, when one thinks that even if this one had gone off, the nuclear safety program would have avoided accidental nuclear detonations something like 99.999…% of the time! (Schlosser had the exact number but I don’t have it on hand.) Of course, such statistics mask the unacceptable cost of even a single failure of that magnitude…

      • Mark Gubrud says:

        As far as I can tell, the only “safety” device that “failed” was the manual pin, pulled out mechanically by the breakup. The baroswitches worked as intended. The impact switch worked as intended. These do not seem to me to be “safeties” for the case of a mid-air breakup.

        However, the image of the electromechanical ready-safe switch shows shows clearly that it is constructed as a rotating drum with at least two “S” positions and probably no more than one “A” position. A motor of some kind must advance the drum from one position to the next. I see no reason why the motor would be designed to go in two directions so that it could advance to “A” from either of two redundant “S” positions. Therefore the most reasonable interpretation of this evidence seems to be that the switch was designed to advance through a series of “S” positions before getting to the “A” position.

        Since a continuous motor would not give positive control of the drum position, it is most likely that the motor was operated by pulses, with a required on-off cycle for each step.

        The switch may also have been wired to return a signal back to the cockpit to operate an indicator.

        The image shows a substantial number of wires, probably associated with multiple positions of a rotary switch, or as redundant safeties, or both. In any case, not just a single wire.

        The number of pulse cycles needed to operate the switch may have been as little as one, but in that case, the only reason I can think of to have two “S” positions is that a second pulse would return it to safe. It seems likely that there are actually more than two “S” positions, the others being obscured in the image. If so, the most reasonable interpretation is that a prescribed number of pulses (like, 9) was required to take the switch to the A position, and one additional pulse would return it to S.

        If this interpretation is correct, then the “short” that would have been required to arm the bomb would have needed to generate just the right number of pulses. This is also a ‘credible postulate’ but even less likely.

  12. Brenton says:

    I find it odd that you describe the device in question as looking crude when I find myself looking at most of the pictures in this blog and admiring how well made most of the ancillary equipment looks. I would shudder to think what would be the interpretation of some modern equipment that fulfills this same purpose. Modern relays and switches can be shorted or opened quite easily with mechanical force since the housings are made from plastic or ceramics.

    • Bradley Laing says:

      —Should I assume they thought of that, at Sandia National Laboratory, I should I assume they did not think of that at Sandia National Laborortory?

  13. Oldhippy says:

    Very much enjoyed your site however much was made of R.N. Brodie’s review of nuclear accidents sent to you by Mr Schlosser particularily as to the operation of the Bisch generator. Brodie however states the other bomb “experienced HE detonation upon impact.” and you indeed quote this. This contradicts Chuck Hanson’s account & all the wiki entries related to this accident and the recent press coverage generated by the FOIA story. This negatively impacts the veracity of the Brodie review generally & by association Sandia. Does that negatively impact the veracity of quoted opinion on the operation of the Bisch generator?

    My apologies if my ignorance of matters nuclear weapon related has allowed me to miss anything which would otherwise be noticeable. I would value your response.

  14. […] I usually expect a more pedestrian account to appear, explaining that the headline was overblown, but perhaps this is not so. Normally sober men seem […]

  15. Hunter S says:

    I wonder how the residents of Goldsboro and the surrounding towns would react to this if the bomb went off and what would happen to them, not to mention all the people down in the fallout zone… yeesh. Scary stuff.

  16. Schoolmarm says:

    I have been haunted for years by an experience I had in first grade in Raleigh, NC, and wonder if it can be explained. It would have been 1960 or 1961. I don’t know what month it was, but I don’t remember being particularly cold. I was walking by myself from school to daycare. I heard a whistling noise and looked up to my left. I saw what looked to me like a white passenger airplane with no wings or windows falling to the ground. It was falling nose-first and was not spiraling. I could see fin-like projections on the end, like tail, and a red, white, an blue emblem that may have been a US flag just above it. The nose was rounded. I lost sight and sound of it for a couple of seconds, then I heard a very loud crash. Even though I was only six, I knew it was from the US by the markings. (I was well aware at that tender age of the cold war and the Soviet threat.) I can’t remember discussing it with anyone, but a few years ago, I asked my mother about it, and she didn’t know anything and couldn’t remember me talking about it as a child. I have looked up pictures online of bombs and missiles of the day, but I haven’t seen anything quite like it. I hope this can be solved for me.

  17. Poose says:

    Just wanted to say thanks for the book recommendation. “Command and Control” was fantastic!

    Found your site via a cross-post from Pharyngula at Freethought Blogs regarding the silent footage of Fat Man being prepped.

    Still devouring your site (I’m a bit of an amateur nuke historian) and still think that Castle Bravo showed just how little we knew about our toys at the time…

  18. […] one fatality. It’s not one of the “standard” accidents one hears about, like the 1961 Goldsboro bomb, the 1958 Tybee bomb, the 1968 Thule crash, or the 1966 Palomares accident. But Schlosser’s […]

  19. […] don’t want it to be too hard to start nuclear war. So this is just another variation of the “always/never” problem: you want to be able to start nuclear war if you need to, and start it quickly and […]