Here be dragons

by Alex Wellerstein, published November 20th, 2015

The most famous experiment conducted by Los Alamos during the Manhattan Project, after the Trinity test itself, is the one with the most evocative name. “Tickling the Dragon’s Tail,” also known internally as just “Dragon,” is straightforward about its meaning, compared to the enigma of “Trinity.” Dragons don’t like to have their tails tickled — so watch out for the fire.

On the latest episode of Manhattan (204), protagonist Frank Winter encounters the "dragon" — and pushes it a little further than he ought to have.

On the latest episode of Manhattan (206), protagonist Frank Winter encounters the "dragon" — and pushes it a little further than he ought to have.

The “dragon” moniker was coined by Richard Feynman (who else?) after he heard about it from fellow scientist Otto Frisch. It was one of a category of criticality experiments that Frisch (nephew of Lise Meitner, co-author of the famous Frisch-Peierls report) was working on at Los Alamos. Criticality experiments were dangerous by design: they were attempts to experimentally determine the critical condition of different quantities, types, and geometries of fissile material. Because of the unknowns involved, all of these experiments involved pushing very close to the boundary of an uncontrolled fission chain reaction, an embryonic atomic bomb (or reactor) that, while probably not very explosive (it would likely destroy itself before too much energy was released), would create enough radioactivity to cause serious hazard to those working around the site.1

The experiment Feynman dubbed “dragon” was what Frisch had called the “guillotine,” and was one of the more ambitious and dangerous of Frisch’s many criticality experiments. It involved dropping a slug of enriched uranium hydride through an almost-critical assembly of the same substance. Gravity alone would cause the two pieces to briefly form a critical mass — and then to briefly un-form, before too many fission reactions had occurred. If all worked as planned, the slug would release a burst of neutrons and then stop reacting. But if the slug got stuck in the critical figuration, it would release impressive amounts of radioactivity and potentially cause a (very small) explosion.2

Otto Frisch's original "dragon" reactor — the uranium "guillotine." Source: R.E. Malenfant, "Experiments with the Dragon Machine" (LA-14241-H, August 2005).

Otto Frisch's original "dragon" reactor — the uranium "guillotine." Source: R.E. Malenfant, "Experiments with the Dragon Machine" (LA-14241-H, August 2005).

The experiments could produce upwards of 20 million watts worth of energy, increasing the temperature of the fuel by 2 degrees C per millisecond. At their most daring, one burst of the experiment released 1015 neutrons. These experiments were, as the official, secret Manhattan District History notes, “of historical importance,” as they constituted “the first controlled nuclear reaction which was supercritical with prompt neutrons alone.” As far as I can tell, this particular “guillotine” was the original experiment that earned the nickname “dragon,” but the name has been applied to other, similarly close-to-critical experiments as well.3

Criticality experiments were inherently dangerous. They didn’t have to kill you immediately to be a threat: it had been known since the days of the “Radium Girls” that radiation exposure could be cumulatively crippling. The experimental physicists by the 1940s had lost a bit of the “devil may care” air that they had in the early years of radioactivity, when you could spot an X-ray operator by his mangled hands. The Health Group at Los Alamos attempted to keep external radiation exposures within the national radiation standards at the time (0.1 roentgens per day), and optimistically hoped they could aim for zero internal exposures per day. For the time, this was considered conservative, though by the late 1950s the standards for exposure had dropped by a factor of seven.4

Los Alamos scientists keep their distance from a 1,000 ci radiation source used in the RaLa experiments.

The first criticality accident at Los Alamos wasn’t a fatal one, but it did cause some trouble. The experiment was (ironically, or appropriately?) made in the name of safety: it was a question of what would happen if certain geometries and enrichments of uranium were submerged in water. For a weapon that was going to be deployed to the Pacific Ocean, this was not an idle danger — sink Little Boy in the ocean and it becomes a nuclear reactor, because, for enriched materials, regular “light” water acts as a neutron moderator, lowering the effective critical mass. The Manhattan District History outlines the experiment and its outcome:

A large amount of enriched uranium, surrounded by polythene, had been placed in a container in which water was being slowly admitted. The critical condition was reached sooner than expected, and before the water level could be sufficiently lowered the reaction became quite intense. No ill effects were felt by the men involved, although one lost a little of the hair on his head. The material was so radioactive for several days that experiments planned for those days had to be postponed. [emphasis added]5

“Although one lost a little of the hair on his head” — one of those sentences one rarely runs across, especially without any further elaboration, that really sounds disturbing to the modern ear. There were other “minor” exposures too, noted briefly (and anonymously) in the Manhattan District History. Not all were related to criticality; some were related to other experiments, such as the “water boiler” and “power boiler” reactors (more on those in a second), and the RaLa (Radiolanthanum) implosion experiments:

Operation of the power boiler resulted in several instances of mild overexposure to radiation caused by leaks in the exhaust gas line and one serious exposure of several chemists during decontamination of active material. The implosion studies of the RaLa Group which used large amounts of radioactive barium and lanthanum brought a serious situation which the health group monitored closely. A series of accidents and equipment failures caused considerable overexposure of chemists in this group. This condition persisted about six months until the system of remote control operation was finally perfected.6

Interestingly, the Health Group had “no responsibility” over the criticality experiments, “except that of being sure that the men were aware of the dangers involved.” The Manhattan District History notes that the criticality experiments were “especially dangerous” because “there is no absolute way of anticipating the dangers of any particular experiment, and the experiments seem so safe when properly carried out that they lead to a feeling of overconfidence on the part of the experimenter.” The author of this section of the History attributes this overconfidence to the death of Harry Daghlian, who died after accidentally creating a critical mass with a plutonium core. It also notes another accident where “four individuals” received an “acute exposure… to a large amount of radiation” during a similar experiment. The same core would lead to the death of another scientist, Louis Slotin (known for his nonchalance regarding the hazards), less than a year later.7

Harry K. Daghlian's blistered and burnt hand after he received his fatal radiation dose from his own dragon-tickling experiment gone wrong.

Harry K. Daghlian's blistered and burnt hand after he received his fatal radiation dose from his own dragon-tickling experiment gone wrong.

Reading through the various exposures and radiation hazards in the Manhattan District History can be a bit spine-tingling, even if one tries to have a measured view of the threats of radiation. Radiation risks, of course, are more exciting to most of us than the dozens of other ways to die at Los Alamos during the war. Radiation is relatively exotic and mysterious — simultaneously invisible to our basic senses while very easy to track and follow with the right instruments. You can’t see it until you start looking for it, and then you can find it everywhere.

But even with that caveat, some of these reports are still pretty eyebrow raising. One example: The “water boiler” reactor was a small assembly of enriched uranium used as a neutron source at the laboratory. The scientists knew it presented radiation risks: the fuel inside the reactor would get fiendishly radioactive during and after operation, and if there was a small, inadvertent explosion, it could be a real contamination problem. So they (sensibly) isolated it from the rest of the laboratory, along with the criticality experiments.8

But later study showed that they hadn’t quite solved the problem. Gaseous materials, including fission products, were being discharged “near the ground level at the tip of the mesa just to the south of Los Alamos Canyon.” This, the Manhattan District History notes, was “most unsatisfactory and represented a potential and serious health hazard.” They had warning signs, but they were “inadequate and the area was accessible to any casual visitor.” Radiation intensities “in excess of 50 r/hr were repeatedly measured near the discharge point when the boiler was in operation.” Just to put that into perspective, even by the relatively lax standards of the Manhattan Project, you would hit your yearly limit of acceptable radiation exposure if you spent about 45 minutes near the discharge point when the reactor was running. By the standards from the late 1950s onward, you would hit your yearly limit after only six minutes. (The committee recommended to put a fence around the area, and looking into building a large smoke stack. Later work determined that the larger smoke stack improved things a bit, but did not ultimately solve the problem.)9

The "Water Boiler" reactor at Los Alamos — a neat scientific experiment, but watch where you put the exhaust port. Source: Los Alamos Archives (12784), via Galison 1998.

The "Water Boiler" reactor at Los Alamos — a neat scientific experiment, but watch where you put the exhaust port. Source: Los Alamos Archives (12784), via Galison 1998.

Did these cavalier radiation exposures have long-term consequences for the scientists? (Other, of course, than the two who actually died, or the few people whose acute radiation exposures were so high that they produced obvious physical damage.) Remarkably, very little follow-up seems to have been made. It takes work to know whether there are hazards, and it takes even more work (longitudinal studies, epidemiological work, etc.) to see whether there have been health effects. Radiation-based cancers are probabilistic; exposures to radiation just increases the chance of a cancer, it doesn’t guarantee it. Epidemiological studies, like the ones done on the Japanese who survived the attacks on Hiroshima and Nagasaki, look for the statistical excesses, the cancers beyond what you would expect to naturally occur in a given population. This apparently was never done for Manhattan Project employees. There are many anecdotes about exposed employees developing debilitating health effects, but little hard science — not because the exposures or consequences didn’t happen, but because apparently nobody did the studies necessary to establish their existence.10

Why wouldn’t the Manhattan Project or Atomic Energy Commission officials follow up on this question? Two interrelated and non-exclusive hypotheses immediately spring to mind. One is that they were genuinely rather sanguine about the effects of radiation in low exposures. Their standards for “low exposures” were considerably higher than ours are today, and the requirements of war didn’t encourage them to adopt the precautionary principle, to say the least. The second is that there were legal stakes involved. They were eager, especially in the postwar, to avoid claims of radiation damage from former employees. Partially one can see in this the attitude of the bureaucrat who believes they are protecting the government’s interests (at the expense of labor’s), partially this is another reflection of the aforementioned sanguinity regarding radiation exposure (they legitimately believed the claims were probably false, or at least not provable). Following the community of scientists, technicians, and laborers after they had left the laboratory would have been difficult. And what if they had found higher-than-normal rates of injury and death? Better not to look at all, from that standpoint.11

  1. One of the key factors in designing an actual atomic bomb is holding together the reacting mass as long as possible. Without that, once enough energy has been released to separate the reacting material, the reaction will stop. So a chain-reacting critical assembly ought not release more than a few pounds of TNT worth of explosive power — but it would release an awful lot of radiation in the immediate area. []
  2. On Feynman and Frisch, and Frisch’s earlier experiments, see Richard Rhodes, Making of the Atomic Bomb (Simon and Schuster, 1986): 610-611. The description of “dragon” and its dangers in this paragraph comes from Manhattan District History, Book VIII (Los Alamos Project), Volume 2 (Technical), 15.7. For an example of the size of the explosion, consider the effect of the accidental criticality excursion on another such device, “Godiva.” []
  3. Manhattan District History, Book VIII (Los Alamos Project), Volume 2 (Technical), 15.8. The “dragon” experiment had one criticality “excursion” of note, when towards the end of a series of experiments of increasing power, a burst of 6 x 1015 fission reactions occurred, blistering and swelling the cubes that composed the assembly. No one was exposed and there was no contamination, but it got put into a criticality accident report. United States Atomic Energy Commission, Operational accidents and radiation exposure experience within the United States Atomic Energy Commission (Washington, DC: Atomic Energy Commission, Division of Operational Safety, 1975), 38. []
  4. The 0.1 roentgens per day (so around 37 r per year) standard for whole-body exposure was adopted by the United States in 1934. By 1946, the US had dropped the standard by half that amount. By the late 1950s, the standard for permissible amount of radiation exposure had dropped to around 5 r per year, where it remains for people who work in nuclear settings (the standard for the general public is lower). Note that in the 1940s the roentgen unit changed to the rem, and is now measured in sieverts, but they are pretty easy to convert (~1 r = 1 rem = 0.01 Sv). See George T. Mazuzan and J. Samuel Walker, Controlling the Atom: The Beginnings of Nuclear Regulation 1946-1962 (Washington, DC: Nuclear Regulatory Commission, 1997), 35, 39, and 54. On Manhattan Project standards, see Vincent C. Jones, Manhattan: The Army and the Atomic Bomb (Washington, DC: Center of Military History, United States Army, 1985), 419, and Barton C. Hacker, The Dragon's Tail: Radiation Safety in the Manhattan Project, 1942-1946 (Berkeley: University of California Press, 1987). Separately, it is of interest that the “Radium Girls” was not just an oblique connection: scientists from Los Alamos, Chicago, and Oak Ridge visited a luminous (radium) paint company in Boston to learn how they dealt with radiation hazards in industry, and adapted their techniques to the problems of dealing with plutonium. Manhattan District History, Book VIII (Los Alamos Project), Volume 2 (Technical), 3.95. []
  5. Manhattan District History, Book VIII (Los Alamos Project), Volume 2 (Technical), 15.10-15.11. The accident in question took place in June 1945, involved 35.4 kg of 83% enriched uranium cubes. United States Atomic Energy Commission, Operational accidents and radiation exposure experience within the United States Atomic Energy Commission (Washington, DC: Atomic Energy Commission, Division of Operational Safety, 1975), 37-38. []
  6. Manhattan District History, Book VIII (Los Alamos Project), Volume 2 (Technical), 9.34. []
  7. Manhattan District History, Book VIII (Los Alamos Project), Volume 2 (Technical), 9.34. []
  8. Manhattan District History, Book VIII (Los Alamos Project), Volume 2 (Technical), 6.60. []
  9. Manhattan District History, Book VIII (Los Alamos Project), Volume 2 (Technical), Supplement, 2.85. []
  10. There have been some very small-sample studies of very specific cohorts from this period, but nothing of the sort one might imagine might exist. []
  11. Gabrielle Hecht’s Being Nuclear: Africans and the Global Uranium Trade (Cambridge, Mass.: MIT Press, 2012) emphasizes, in the case of exposures from uranium mining in Africa, that the easiest way to avoid worrying about radiation exposures is not to measure them, not to do the work that makes them “exist” as observable scientific facts. []

10 Responses to “Here be dragons”

  1. John Coster-Mullen says:

    Some of the MP scientists I interviewed who had done criticality experiments told me whenever they got too many “guest” visitors they would demonstrate one of the tests to them and then subtly move their hand closer to the assembly. When the neutron counter clicks got to a loud hum they would exclaim “Oh Oh, something might be going wrong here, you guys might want to leave NOW!” They would politely thank them and run out of the room to the quiet laughter of the scientists. This meant, of course, their own bodies actually became part of that critical assembly and those gamma rays, neutrons, and x-rays were coursing through their bodies. The fact they were telling me this when they were in their 70’s and 80’s obviously meant it did them no harm whatsoever.

    The late Don Mastick gave me a call in 1995 and explained in detail how that vial full of plutonium (in gaseous form) burst in his hand while he was holding it. “My natural reaction was to gasp thus drawing most of it into my gullet and lungs. I still have a lot of that inside my body more than 50 years later and there is nothing wrong with me medically. Humans have very strong bodies!”

    • This strikes me as no different than the old chain-smokers who say, “I had a pack a day my entire life, obviously smoking is safe because it didn’t get me.” Or someone who survives at Russian roulette, deriding the possibility of danger. Anecdotal, individual cases do not establish safety baselines, especially with probabilistic effects like cancers. You need epidemiological studies to establish risk factors. Separately, there is a methodological problem here — a selection bias: you didn’t talk to anybody who didn’t treat radiation exposures with respect and didn’t live to their 70s or 80s.

  2. I’m curious about a quotation in your fourth paragraph from the Manhattan District History: “the first controlled nuclear reaction which was supercritical with prompt neutrons alone.” Though I know little enough about all of this, I’m acquainted with the concept of fast neutrons and wonder whether the phrase “prompt neutrons” is only an elegant variation or has a more precise meaning.

    • My understanding is that they are fast neutrons, to be sure, but the distinction they are making is between using delayed neutrons as part of the reaction. The water boiler reactor, for example, used prompt neutrons (the neutrons released immediate from the fission reactions) to get it to near critical, and then would edge into critical when you took into account the neutrons being released from fission products. In a reactor this is what gives you a sense of control, being able to keep the level of operation mostly constant once you’ve gotten the system into a critical configuration. By contrast, they are saying that in this case the reaction is entirely based on prompt neutrons alone — much more like a bomb works than a reactor, and yet it is a controlled reaction (i.e. not a bomb).

      (The above is based upon how they talk about the role of both prompt and delayed neutrons in the same report, with regards to the operation of the “water boiler” reactor.)

  3. Mark M says:

    Correct that prompt neutrons come directly from fission, delayed neutrons come from fission products. A controllable reactor is not quite critical on just the prompt neutrons, it needs the delayed neutrons to “fill out” the population needed for criticality. Recall that an exactly critical reactor has a stable neutron population, it is at a fixed power level. If a reactor is increasing power, it is (very slightly) supercritical, if it’s losing power, it’s slightly subcritical. When a reactor operator “calls” criticalityh on startup, actually the reactor is slightly supercritical, since the count rate is increasing.

    These test rigs were apparently the first to go supercritical on prompt neutrons alone (most delayed neutrons would appear too late, the assembly would already be dis-assembling itself due to gravity).

    All neutrons are “born” fast – they become epithermal or thermal only after collisions with a “moderator” such as graphite or various hydrogen compounds (light water, heavy water, oils, etc.). In a previous post you can see a curve of U5 and P9 fission cross-section versus neutron energy. In general the cross-section is larger for thermal than fast neutrons.

  4. Joe Van Cleave says:

    So, did these criticality experiments also involve moderated (thermal) neutrons along with prompt and delayed? As I recall, there were experiments that surrounded a fissile sphere with various amounts of moderator/reflector material. Did the guillotine device have any such moderator material?

    Or perhaps I’m not understanding the terms correctly. Do all neutron reflector materials moderate (thermalize) the neutrons, or can they be reflected back into the core at energies closer to prompt?

    • A neutron reflector can moderate but does not have to. Some materials will result in elastic scattering (which is to say, the direction of the neutron will be modified but no energy will be imparted to the moderator), some will result in a little energy being lost. Some of the criticality experiments definitely used moderators of some sort (e.g. the ones with uranium hydride in water — the water is acting as a moderator, and the hydrogen present in the uranium is going to have a moderating effect as well). In a bomb, I don’t think there is ever a serious effort at moderating — there just isn’t enough time to worry about it (which is why the enrichment level must be so high, compared to a reactor).

      Guillotine/Dragon used uranium hydride (UH10 apparently) and also had carbon/graphite tampers. So there might have been some moderation on the whole, but what they were interested in was the prompt (unmoderated) neutrons.

      OK, fun fact that I just noticed: the button to drop the slug in Guillotine/Dragon was labeled “HWG” for “Here We Go.” I find that delightfully whimsical and somewhat perverse given the subject matter.

  5. The language that’s chosen to convey something, as in “HWG” for “Here We Go,” is part of what interested me in the distinction between “prompt” and “delayed” neutrons. My impression is that “prompt” is no longer commonly used in everyday situations; it calls to mind an earlier period (such as the 40s), in which students or employees would be reminded to “Be prompt” in their arrival, etc. It’s rather idle to speculate, but I’d guess that if we were only now developing the science and technology of fission, we might simply speak of fast and slow neutrons, which is in fact the way I first read the concepts discussed.

    • Mark M says:

      Prompt neutrons are released directly from fission, they are fast initially but may be moderated to slower speeds. In a reactor they are moderated, unless it’s a “fast” reactor, like one cooled by liquid metal. The Hanford reactors were moderated because it’s only possible to run a reactor on natural uranium with a moderator, since moderation increases the fission cross section of the small amount of U-235 available. In fact you have to have a very good moderator, such as graphite or deuterium, to run a reactor on natural uranium.

      Delayed neutrons come from the fission products, a considerable time by nuclear reaction standards after the fission event. They are initially fast as well.

      So we have “Prompt” and “Delayed” neutrons, depending on origin. We have “Fast”, “Thermal” and “Epithermal” neutrons, most of which are “Prompt” by birth, but a few of them are “Delayed” by birth.

      The term “Fast Breeder Reactor” – the term “Fast” refers to neutron speed, the term “Breeder” means it produces a little more fissile material than it consumes. To do this it does consume a slightly greater quantity of fissionable or fertile material (U-238 or Th-232, usually). When I was a kid I thought a “Fast Breeder” bred more material faster, that there was probably a design called a “Slow Breeder” out there, but, this being America, nobody cared about the slow design.