Arms Control Wonk ArmsControlWonk


Trevor Findlay’s Canadian Centre for Treaty Compliance has just released a report on Nuclear Futures which makes for excellent reading. In the report, there is a brief paragraph on information driven safeguards.

The Agency claims it is pursuing what it describes as “information driven” (IAEA, 2007b: 16) safeguards, supported by a modern “knowledge management system” (including a database that records the experiences of all safeguards inspectors). However, there remain concerns that the IAEA inspectorate’s “culture” has still not entirely changed from one of examining a narrow range of information to one that considers each individual state’s activities holistically. Further concerns have been expressed about the lack of transparency and openness within the Agency that permits vital information about state compliance to be too tightly held within certain offices, thereby defeating the purpose of the holistic approach (ICNND, 2009: 91-92)

Beyond doubt, the last few decades have seen a transformation of the safeguards business. Until about 20 years ago, the International Atomic Energy Agency only corroborated states’ nuclear material on declared nuclear sites. They did this through measurements and observations. They quantified the nuclear material, they counted fuel rods, and inspectors even watched how materials flowed through the plants themselves.

The inspectors of the 1970s and 1980s realized the limits of the system they enforced. It doesn’t take much to figure out that the easiest way to defeat safeguards is to bypass them. All proliferative states have had to figure out how to deal with one form or another of safeguards on their territories.

Most states seeking to develop nuclear weapons have found that it is best to conduct forbidden activities well away from the prying eyes of the international inspectorate. Iraq’s strategists, for instance, only felt that they could divert its HEU at Tuwaitha after IAEA inspectors had been expelled. And Sweden’s cold warriors felt that they could only convert the R-3/Adam reactor to military use once it had been purged of safeguarded heavy water. Even the most primitive IAEA safeguards works as an ‘area denial’ tool and this is a benefit not to frown on.

In the last two decades, the IAEA has faced ever increasing expectations on what it is able to deliver. Many expect the organization to become nuclear detectives where it has once been nuclear accountants. This is by no means a bad idea, as some have suggested.

The IAEA has moved rapidly towards a system of state-level evaluations. This regime itself is indistinct, and descriptions of it are coached in management language such as “information driven” and “knowledge management system”.

In a nutshell, the IAEA has started to use the large information flow given to it by the Additional Protocol and other knowledge sources to its advantage by feeding it into analytical tools such as the physical model (see one description here). The IAEA has also, over the years, started to realize the power of open source information and satellite imagery analysis. One plan is to feed all this information into a central repository which, well, looks a bit like an internal Wikipedia. I have seen the demo, and it is impressive.

All of this may give the folks in Vienna the potential to describe and understand the entirety of the country’s fuel cycle. Once they are allowed to do that, it will become increasingly difficult to hide away a parallel fuel cycle. But reform does need a change in attitude within the IAEA. Some serving or former safeguards inspectors still like to point out how slowly the organization has changed, and how much work remains to be done.

Be that as it may, the most important thing is that work has started.


Yesterday, I attended a meeting with the International Law Association in Brighton, and the Swedish nuclear programme was raised again. It is interesting that so much international attention has been given recently to what, essentially, is a but a side note in the broader Cold War narrative. Perhaps it’s because Carl Bildt, the Swedish Foreign Minister, reportedly likes to talk about it at meetings. Or perhaps it’s because Jeffrey has given me leeway to write about arcane and, for the most of you, uninteresting topics?

A while back, a friend from a Swedish ministry also asked my why I had not, in my previous posts on the Swedish programme, had not mentioned the Swedish delivery vehicle, SAAB project 1300, or the A36 tactical bomber.

I’m an aviation enthusiast. My father was, for many years, an employee of the Swedish Fortifications Administration. I spent a fair share of my childhood around air-force bases, and got familiar from an early age with the wonderful machines that SAAB has produced over the years. Of course, the SAAB J-35 Draken (‘Kite” or “Dragon’) is a favorite, and so are the JA-37 Viggen (‘Thunderbolt’) and the JA-39 Gripen (‘Griffin’).

Some of you know that I also used to have a glider certificate. I’m not sure if that makes my a lapsed pilot. But my father once managed to get a friend of his to let me fly the JA-37 simulator at the F13 flotilla for one hour. As this counts as experience, I was allowed to log this as flight time in my logbook (I did manage to land the fighter safely. My father, however, crashed and burned).

The Swedish Nuclear Bomber

So I decided to find out more about the A36 bomber. This was a single seat, single engine, delta-wing design. The company planned to use the RR Olympus engine, used in the Vulcan and later used in the Concorde, to give the plane some speed. It was, after all, only supposed to make a quick dash over the Baltic, hit the Soviet embarkation ports, and then make a fast escape back to Sweden.

The aircraft was designed to carry one free-fall nuclear weapon (carried in an internal bay). The weight of the weapon was given as no more than 800 kilograms (or 1760 lb.). Some sources puts the weight of the payload to 600 kilograms (or 1320 lb.). The internal bay was only put into design due to concerns of accidental detonations caused by high air friction.

This was a fast plane, designed to hit Mach 2.2 at high altitudes and at least Mach 1.2 at lower runs. Urban Fredriksson, a Swedish X-Plane enthusiast, has modeled the aircraft and tried it out in the simulator. According to him, the plane “flies better than OK and very close in speed and range to what it should be and it has to be landed very nose high in the manner typical of deltas.” According to one of the designers of the aircraft, the main problem the SAAB engineers faced was the shape of the canopy, which had to be “narrow and pointy” to be feasible.

The project was submitted in 1952 but was cancelled in 1957, to allow for more resources to go into the JA-37 project.

Effects testing

In 1956 and 1957, the Swedish military conducted a number of massive conventional explosions for research purposes at Nausta in Northern Sweden. The first test serious was given the code-name ‘Sirius” and involved three benyl charges (633, 6,040 and 61,000 kilograms). The military wanted to study intense pressures, and were, for some reason, also interested in the height of the mushroom cloud. According to some sources, they noted heights of between 350 and 1,020 meters.

The second series, code-named ‘Vega’, involved two benyl charges (5,000 and 36,000 kilograms). These tests aimed to explore weapons effects, and the military had therefore deployed a number of vehicles, airframes and other materials at the site.

More images from the test series are available here.

I am aware that there might be a number of new publications on the programme coming out in English sometime in the future. I listed a number of primary sources in post here but for some reason all the links are broken. They must have been moved to another part of the site.

However, if you Google, you shall find.


I was going to revisit Natanz in my 14th post on the Wonk, but something more interesting came my way. I could not help noticing that Dan Joyner, a member of the powerful International Law Association, has written a response of sorts to James Acton’s article Iran Violated International Obligations on Qom Facility (Proliferation Analysis, 25 September 2009).

In a nutshell, James argues that subsidiary arrangements in INFCIRC/153-type safeguards agreements are legally binding instruments, in essence contracts between the IAEA and the state. Dan, on the other hand, concludes that subsidiary arrangements may not carry any legal force, that they have a ‘non-binding legal character’.

To be clear, both James and Dan reaches the same conclusion, that Iran’s behaviour is a cause for concern. Dan, however, looks at the problem from a structural perspective, and readily agrees there are worrying consequences for the safeguards system if his interpretation is correct.

I would probably subscribe to the view that subsidiary arrangements are part and parcel of the safeguards agreement itself. Without them, the safeguards agreement would not be meaningful. Their legal force is not explicitly stated, but the safeguards agreement is littered with functional references to its subsidiary arrangements.

For instance, paragraph 32 requires the state to set up certain measures (such as procedures for taking a physical inventory), as specified in the subsidiary arrangements. Paragraph 39 states that “provision should be made for the possibility of an extension or change of the Subsidiary Arrangements by agreement between the Agency and the State without amendment of the Agreement”. In other words, that they can be changed without having to go through the ratification procedure again. In addition, like any other contract, subsidiary agreements enter into force (see paragraph 40).

As an intergovernmental organization, the IAEA has what we lawyers call “legal personality”. This means that it has right to enter into agreement with states or, for that matter, non-state actors. And it has the right, as any sovereign state has, to expect that agreements are kept. The question is whether Iran has kept its agreement with the Agency.

And that is something that is debatable.


Many things remain unknown about Sweden’s nuclear programme, especially why it was suddenly discontinued. It is interesting to note, of course, that the US intelligence community estimated that Sweden was on its way to the weapon up until 1964. Later intelligence assessments are considerably more cautious. That’s probably because our American cousins knew that Sweden wasn’t seeking it any more. I say cousins as about 1 in 20 of the US population is said to have Scandinavian heritage. My favourites of those are probably Uma Thurman, Charles Lindbergh, Kim Basinger and Buzz Aldrin: beautiful women and daredevil pilots.

Cultural similarities aside, little is known about the deal that was made between the Swedish and U.S. governments at that time. Details seem to be heavily classified on both sides of the Atlantic. Swedish historian Wilhelm Agrell, however, points to the mysterious enlargement of Swedish Air Force bases in the latter half of the 1960s, where several runways were extended to be able to receive strategic bombers. Of course, the Swedish Air Force had none of those. This week, I spoke with Ove Bring, a mentor of mine and author of a 2008 book on Swedish Neutrality. He also told me of the Air Force’s decision to equip Swedish tankers with NATO specification nozzles. It would seem like the Air Force started to work closely with its NATO counterparts in the mid 1960s, but few primary sources exist to confirm this.

While it will take some time to get confirmation on why Sweden abandoned its weapons plans, it is possible to piece together the programme itself. Sweden’s fuel cycle activities in the 1950s and 1960s are fairly well documented. When reading about many of these assets, their relationship with the military programme is implied, but never stated and sometimes denied. Most fuel cycle assets except a reprocessing facility were in place in 1969. However, by that time Sweden had joined the 1968 Nuclear Non-Proliferation Treaty, and its weapons experts had become disarmament experts. In the 25 years that followed, most assets were shut down or decommissioned.

This post is based on a number of sources, including some primary documentation, papers written by FOA researchers, as well as, surprisingly, local community history websites. There are a number of good reports written on the programme, but some of them seem speculative or vague in parts. Of course, the writings of Wilhelm Agrell and Thomas Jonter is highly recommended.

Uranium extraction

The designers of Sweden’s nuclear programme realized, as so many other states seeking nuclear weapons, that the key to the bomb is easy access to uranium ore. If you do not have access to domestic ore, its little point engaging on a full-scale weapons programme. In this respect, Sweden found itself in a favourable position. The country was, and still is, very rich in natural uranium. However, the ore grade is quite low (mostly shale), and therefore requires extensive mining and milling. According to a relatively recent survey, Sweden has something between 4 and 32 million tonnes of extractable ore – which was enough for the surveyor to call the country “the Saudi Arabia of uranium” (Continental Precious Minerals Inc., 2005: Results of NI 43-101 Geological Report 7/29/05).

Be that as it may, the ore grade in Sweden varies between 200 and 300 parts per million, which makes it mostly uneconomic to extract. The uranium is hidden in Cambrian period seabed, formed some 400 to 600 million years ago. The Swedish weapons programme focussed on two areas.

Kvarntorp (59° 7’32.11“N 15°16’23.32“E). Uranium extraction started in 1953 by a Swedish government public venture (Svenska Skifferoljebolaget), which had been conducting oil exploration in the region since 1941. The ore in the area is extremely low grade (only 200 parts per million). By 1956, the company had only managed to extract about five metric tonnes of uranium. When mining stopped in 1963, the company had managed to extract about 50 metric tonnes of uranium.

Ranstad (58°16’19.09“N13°42’44.36“E). The government instead decided to go for the mine in Ranstad. AB Atomenergi got permission to excavate the area in 1959. Ranstadsverket was inaugurated in 1965, but was only running for about four years. In 1969 the mine was shut. It had extracted about 200 metric tonnes of uranium. This production is still accounted for in the OECD Red Book.

Norwegian heavy water
(59°52’43.59“N 8°33’27.03“E)

Another asset that the Swedes needed was heavy water to moderate their reactors. US supplied heavy water came with a troublesome string attached – namely the right to inspect the facilities where the water was used. However, neighbouring Norway had no qualms exporting the precious commodity without strings. Under secrecy, Sweden purchased five tonnes of heavy water from its neighbour.

The Wallenberg family might have facilitated the deal. Marcus Wallenberg, a Swedish lawyer and banker, sat on the Norsk Hydro board for 37 years. Norsk Hydro owned and operated the Norwegian heavy water plant in Rjukan. Wallenberg was also a founder of ASEA (which were later in charge of developing Sweden’s nuclear power plants) and the family controlled SAAB, the makers of most fighter-bombers of the Swedish Air Force.

The Swedes also had a good ally in Jens Christian Hauge, the Norwegian defence minister, an eccentric former resistance fighter. The Social Democratic leadership in Sweden despised him. Prime Minister Erlander, for instance, once described him as reckless. But Hauge had a good relationship with the Swedish military, and that relationship is sometimes described as one contributing factor to Sweden’s closeness to NATO. The Swedes had noted that Mr. Hauge had already helped orchestrate an export of safeguards-free heavy water to Israel. If Norway could export to Israel, why would it not export to its good neighbour?

FOA Grindsjoen
(59° 5’8.08“N 17°52’16.51“E)

The Swedish weaponization effort was located in a research area south of Stockholm, called Grindsjoen. The research facility was established in 1941 through a personal donation of Olof Arrhenius, son of the Nobel-prize winner Svante Arrhenius (a Swedish chemist and physicist). It was established to promote ‘natural sciences for defence needs’. It was initially never a formal authority of the state, but rather a collaboration of the physics faculties of all Swedish universities. The military found the location perfect. Since it was so remote, it was easy to keep it secret. The remote location was also appealing to the young scientists employed there. The nature was beautiful, and the local waterways were perfect for fishing and other activities.

At the end of the 1950s, it employed about a hundred scientists and engineers, all tasked to figure out how to build a nuclear weapon. According to some who worked there, they never actually received any instruction to build the weapon. They were conducting basic theoretical studies: such as implosion technology, material studies (especially on UK supplied plutonium), and other basic calculations. The design team early opted for a plutonium-fuelled implosion device. Calculations on the optimum configuration were made on a state-of-the-art IBM 7090 computer installed by FOA in 1961 (the IBM 7090 was also used by NASA in the Mercury Programme). They validated the calculations by conducting cold tests with suitable surrogate materials. The conclusion was that it was not a simple feat to achieve ideal geometry. However, within a couple of years, the team felt that they had good knowledge of what needed to be done.

The Grindsjoen team also needed to know how plutonium behaved under intense pressure. They knew that if they could increase the metal’s density (under pressure), the critical mass would decrease. For that reason alone, high explosive research got priority. The FOA team therefore conducted compression tests where small samples of plutonium were compressed by high explosives within a steel container (which in turn was placed in a facility glove box). The team studied, albeit not very extensively, casting, shearing, and stabilization of plutonium metal. And they also developed a prototype neutron initiator.

The R1 reactor
(59°21’0.33” N18° 4’0.91“E)

The R1 was Sweden’s first reactor. It was fuelled by three metric tonnes of uranium metal provided by France, and moderated by five tonnes of heavy water supplied by Norway. The reactor is located 27 meters under the buildings of the prestigious Royal Institute of Technology. It was not a reactor designed to produce significant quantities of weapons grade plutonium (its effect was only 1 MWth). Rather it was built to give the Swedes reactor operation experience and to supply knowledge for how to build bigger, more powerful, reactors down the line. The CEO of AB Atomenergi, none other that Sigvard Eklund, inaugurated the reactor in 1954. One of the scientists who worked there was Professor Rolf Maximilian Sievert. AB Atomenergi made no secret of the reactor. Indeed, the Swedish King was present at the inauguration, and there were even postcards produced with the reactor as a motif.

The R2 and R2-0 reactor at Studsvik

The second reactor to be built was the R2, a pool type reactor with a 50 MW thermal effect. The R2-0 reactor was much smaller, with a 1 MW thermal effect. Both reactors were commissioned in 1960. The idea was to transform the sleepy little town of Studsvik into a nuclear research area. The institute was initially called “Atomic City”. Both reactors were nominally civilian, and was brought on-line for material testing and irradiation. The reactors were shut down in 2005. Remarkably, the US Atomic Energy Commission contributed 350,000 dollars to its construction. The United States also supplied the reactor with its fuel, 93.5 per cent enriched uranium. A typical core started with a load of 12-13 kilograms of highly enriched uranium, but prolonged burn-up often reduced the quantity to about 160 grams.

The R3 reactor at Agesta

One of the most fascinating reactors associated with Sweden’s efforts to seek nuclear weapons is R3/Adam. This was a small nuclear power plant, which was not really part of the weapons programme. It was seen as a test reactor on a very large scale. It had an 80MWth effect (about 12 MWe). The heavy water was supplied by the United States. The reactors burn-up was declared to be 3000 MWd/t. The total fuel load was 18 metric tonnes of natural uranium. If you simulate the reactor with the IAEA’s INFCIS system, you’ll find that the spent fuel if off-loaded within the year, would have contained nearly 27 kilograms of Plutonium-239 and only about a kilogram of Plutonium-240. In other words, near perfect weapons grade plutonium. Some of this plutonium is still stored in Sweden (not enough for a weapon). The vast majority has been shipped off to the United Kingdom.

The idea was not to use Agesta as the plutonium producer. For that purpose, AB Atomenergi was building a huge heavy water moderated reactor near Marviken. However, as the Marviken project was running into difficulties in 1966 and 1967, the military started to turn its attention to Agesta again. However, the heavy water was supplied from the United States, which meant that there was no easy way to bypass safeguards. Some preliminary plans for a crash program were drawn up. This would involve emptying Agesta of all US supplied heavy water and fill it up with Norwegian supplied stocks. However, these plans never materialized.

Agesta is remarkably well preserved today. According to visitors, one gets the feeling that the operators are out on a coffee break and may return any moment. However, the Swedish Nuclear Inspectorate has given assurances that the reactor itself may not be run without significant investment. At present, there is a battle going on between those who want to keep the reactor as a cultural artifact, and those who want to tear it down.

The R4 reactor at Marviken
(58°33’10.65“N 16°49’55.65“E)

R4/Eva was to be the biggest reactor produced. It was finished in 1968, but never got permission to load. Moderated by 185 metric tonnes of heavy water, the reactor effect was about 100MWe. The reactor was designed to be able to be loaded and unloaded while in operation. This is obviously useful when producing plutonium for weapons, since it allows the reactor operator to control the burn-up without having to shut the unit down. However, the link to the military programme was never made explicit. The reactor was planned to produce about 80 kilograms of weapons grade plutonium per year.

In his recollections, Peter Margen, the AB Atomenergi manager responsible for reactor projects wrote “… our design team at Atomenergi had introduced the requirement that it should be possible to refuel at full reactor pressure, and thereafter even at full load for purely economic reasons, initially as part of the studies of future large scale plants, and then in our suggested design for R4/Eve. In our discussion with Atomenergi management, the fact that this could also be of value if a situation would arise in future where the production of military plutonium became desirable was mentioned, but never as a directive for our design work.”

However, the design had a flaw, which necessitated the use of slightly enriched (1.2 per cent) uranium and a higher burn-up. Initially, the designers envisioned a burn-up of 5,000 MWd/t, but this was later increased to about 13,000 MWd/t. This fuel could only be supplied by UKAEA. Large quantities of heavy water had been promised from Savannah River. The changes needed delayed the project by a year and a half, and would have considerably increased cost. The plans to bring the reactor into operation were abandoned.

Lessons learned from Marviken were later applied in the construction of the O1 light water reactor in Oskarshamn. By then, all nuclear weapons plans had been abandoned completely.

Sannas Reprocessing Plant
(58°44’29.57“N 11°14’50.93“E)

In the late 1950s, money was allocated to buy land near Sannäsfjorden. It was here, as far away from the Baltic Sea as possible, where the government planned to build a reprocessing facility. AB Atomenergi bought the land in 1963, and later expanded the area to 2.3 square kilometres. Advanced studies were made to build the facility underground, straight into the cliffs by the coast.

The project was discontinued in 1970. Formally, it was due to opposition from the local council. However, it seems more likely that Sannäs was abandoned at the same time as the Marviken project collapsed.

Nuclear testing

The Supreme Commander’s Nuclear Weapons Group (Kärnladdningsgruppen) very briefly touched on nuclear testing in a classified 1962 memorandum. They concluded that Sweden had the capacity to carry out underground nuclear tests in their northern and western mountain ranges. Interestingly, however, they also held that nuclear testing, albeit desired, wasn’t necessary for the production of a nuclear device. This to me, at least, indicates high confidence in the design concepts worked out at FOA Grindsjoen as early as the beginning of the 1960s (Fst/Forskn 21/9 1962 nr KH 0800).

The Ahasverus system

Reportedly, the Swedish Air Force considered deployment and basing of their new weapons. According to some sources, about 100 weapons were ordered. Some of these weapons should have been stored in underground storage facilities, and the operational bombs should have been rotated around on active bases. The Air Force called this the “Ahasverus System”, after the lore of the Jew who taunted Jesus on the way to the Crucifixion and was then cursed to walk the earth until the Second Coming.

So what remained to be done?

According to former FOA researchers, Sweden still had some way to go. Amongst other things, it needed:

√ To accumulate a sufficient stockpile of safeguards-free weapons grade plutonium. This would have involved running the reactors for a while before committing to the weapon.

√ To finish the reprocessing facility.

√ To finish work on the weapons design.

√ To build a metallurgical laboratory capable of treating the plutonium, shaping the pit, and assembling the physics package.

If Sweden had not signed up and ratified the NPT, a blue and yellow bomb could have been reality in the first few years of the 1970s. Without doubt, all preparations for the bomb would have been carried out under the guise of its “peaceful nuclear programme” and under the concept of “expanded defensive research”.


Jeffrey and I have something in common. I’m Swedish. In his youth, Jeffrey went to some Swedish school (Augustana College in Rock Island, Illinois). And we’ve both developed an interest in nuclear weapons issues.

But what do Madonna and nuclear weapons have in common? Nothing except that Nothing Really Matters was shot on the site of the R1 reactor in Stockholm, Sweden.

This reactor, buried about 30 meters under the city, played a small but important role in Sweden’s nuclear weapons programme. The programme is an interesting piece of non-proliferation and disarmament history, of which relatively little is known, at least outside Sweden.

This post is very long, so I’ve divided it into two parts. The first will go through some of the political context. Later, I will post on some of the fuel cycle facilities that have been identified.

I’ve drawn on the writings of Wilhelm Agrell and journalist Christer Larsson. The former has written a book on the subject in 2002, and the latter is the author of the 1985 article in Ny Teknik that first attempted to chronicle what really happened during the 1950s and 1960s.

A majority of the post is directly sourced from Peter Hansson’s excellent documentary on Swedish National Radio, which was broadcast in 2008. I’ve also had some brief conversations with people both in Sweden and Norway on the topic. Norway’s role, in particular, is spellbinding. Sweden’s western neighbour was the first non-nuclear weapon state to acquire a nuclear reactor. And it also had large quantities of another important asset: safeguards free heavy water.

Some of the secret documentation on Sweden’s nuclear programme was declassified in the mid-1990s. But the qualified secret material, which is under a 70 year classification rule, is not likely to be released until the beginning of the 2020s. Moreover, some believe that the really important documentation was never archived – it may not even be written down. This, of course, fits nicely with a well-known exception to the Swedish constitutional principle of government transparency. Certain memoranda, written simply as a support for your memory, are considered private and may not be given out to the public. We used this loophole frequently at the Swedish Court I once worked in, in order to protect internal assessments and investigations into sensitive cases.

The Swedish nuclear weapons programme was also heavily sectored. No person had access to the complete picture. This has resulted in an increasingly fragmented recollection of what actually happened. Some argue that a nuclear weapons programme never can develop secretly in a free and open society. Yet, the true extent of democratic Sweden’s programme is still unknown. And I suspect its true nature will remain opaque for many more years to come.


The Swedish military reacted slowly to the news of Hiroshima and Nagasaki. Few military strategists within the military high command saw the atom bomb as a truly revolutionary weapon. Instead, they saw it as a powerful asset to be used only sparingly and at very important targets. Gradually, however, the military started to realize that the atomic bomb was a true game-changer. As tension between the Soviet Union and the United States rose, the military started to believe that the next war would be fought with nuclear weapons, and felt that Sweden needed to prepare itself for that. British and U.S. ideas heavily influenced the Swedish Defence Forces’ doctrinal thinking at that time. And some may recall that influential strategists, such as Air Marshall Sir Arthur Harris, advocated the use of nuclear weapons to deliver a decisive and pre-emptive blow to the perceived Soviet threat. This must have resonated strongly within the Swedish High Command.

The Air Force, under the direction of General Nils Per Robert Swedlund, was one of the driving forces behind the nuclear weapons programme. It realized that Sweden was in a favourable position. The country’s industry was untouched by the Second World War, it had excellent scientific expertise, and the country was well connected. Amongst the people that the country reportedly depended on for support was Nils Bohr, a world-leading physicist based in Demark, and of course Glenn Seaborg, an American 1951 Nobel-prize winner, generally considered to be the father of the U.S. plutonium production programme. Dr. Seaborg’s parents were Swedish, and he himself spoke the language fluently. After the end of the programme, Seaborg was elected a foreign member of the Royal Swedish Academy of Sciences in 1972. It is sometimes said that Seaborg played a role in dissuading the Swedish government from seeking weapons, but the exact details of what happened in the first half of the 1960s are still classified.

One interesting aspect of the Swedish programme is its close relationship to private enterprise. In many other aspiring states, the weapons programme has been an exclusively state run effort. In Sweden, however, private industry was deeply involved. In fact, the programme started with the founding of a joint government-business venture. In 1947, the government established AB Atomenergi (the Atomic Energy Company). The company was owned by 4/7 by the Government. The other 3/7 was owned by a number of private companies active in the mining, steel and manufacturing industries. The company’s task was to establish the fuel cycle assets necessary for the weapons programme. The military would work out the bomb design. Therefore, AB Atomenergi had a close relationship with the Defence Research Institute (FOA) from the start, through a co-operation agreement signed in 1948. The Defence Research institute had already established a research area south of Stockholm (FOA Grindsjoen) that became the epicentre for military R&D (more on this area in the next post).

As in other proliferative states, the public was never informed about what the reactors were built for. The nuclear programme was portrayed as civilian. Atomic power was seen as the key to a new and better type of society, where all energy needs will be easily satisfied. The establishment of fuel cycle assets were seen as indicators of Swedish industrial progress, and was, I believe, a source of national pride.

In the meanwhile, the military made no secret that they were working on nuclear weapons related questions, but argued that all research was defensive. In 1954, Prime Minister Tage Erlander delivered a speech that argued that the atomic bomb had put all nations in a ‘state of fear’, and held that in order to protect itself from its effects, one would need to know how the weapon worked. This was spinned as a common-sense justification for the research throughout the 1950s, and would take absurd proportions with the concept of ‘expanded defensive research’, introduced in 1959.

He’s working for the Atomic Energy Company

Private enterprise

Sweden, despite having a reputation of being a socialist country, has always had a very strong business culture. The nuclear programme was no exception. Fuel cycle research was conducted under the umbrella of AB Atomenergi, but other interests were pushing for the bomb. The power company ASEA wanted the weapon, since that meant that it would get more orders for nuclear power plants (ASEA later became ASEA-Atom and is now part of Westinghouse). Moreover, the powerful arms manufacturer Bofors reportedly made several internal studies on its capability to assemble the weapon, and later lobbied government to get that role (Bofors is now part of BAE).

Of course, aircraft manufacturer SAAB would also have been involved in the effort. Although the design team at FOA Grindsjoen had some early design ideas about missile delivery, the Air Force strongly advocated the concept of using the SAAB 32 Lansen aircraft for weapons delivery. Studies, although ‘not terribly detailed’, were made on how the weapon would need to be designed to be hung under the fuselage of the platform.


The Air Force remained the key driver behind the programme. After all, it was the Air Force’s fighter-bombers that were supposed to deliver the new weapon to target. But the programme had no public face. The various agencies and companies that worked on the programme realized that they could do a lot of progress without involving the Parliament or the general public. It was only after a fall-out between the Defence Minister (Sven Andersson) and the Supreme Commander of the Swedish Armed Forces (Swedlund) that the research became public. Surprisingly, the Defence Research Institute (FOA) made a public request for funds to develop nuclear weapons. A divisive and bitter debate ensued, which almost threatened to break the ruling Social Democratic Party apart.

Enter Olof Palme. This young man, which later was to become one of Sweden’s best-known Prime Ministers, was called in by Prime Minister Erlander to unify the party and give some top cover for the weapons effort. Palme was the secretary in a Social Democratic Party working-group on the nuclear weapons question. The group released its report in 1959, and the official position became ‘we’re not seeking it’ but unofficially the government wanted ‘freedom of action’. The FOA request was denied, but the budget on defensive research was significantly increased. Prime Minister Erlander, in consultations with the Supreme Commander, reportedly made it clear that ‘defensive research’ also included work on the weapon itself. The term used was ‘utvidgad skyddsforskning’ (i.e. expanded defensive research). In fact, the defensive programme remained a systematic offensive programme.


From 1962 and onwards, the programme slows down. There were likely many factors in play. Public opinion turned sharply against nuclear weapons, and the Swedish government played an important role in the negotiation of the 1968 Nuclear Non-Proliferation Treaty. In 1966, Karl Frithiofsson, a Ministry of Defence official, holds a speech at the Royal Defence Collage where the concept of a nuclear armed Sweden was formally dismissed. According to some, few in the audience had any idea what Frithioffson was talking about . However, historians hold that Sweden reached an understanding with the United States at some point during this period. The deal was that the US nuclear umbrella would protect Sweden, and so there was no need for any nuclear arms. Documents from this time are not likely to be released until the 2030s.

Sweden made good use of their expertise after this. Her nuclear weapons experts become ‘disarmament experts’, and made significant contributions to the debate in Geneva and Vienna. And one of the programme’s first directors and driving forces, Dr. Sigvard Eklund, went to Vienna to head up the International Atomic Energy Agency, a post he held for 20 years.

And so, in the years that followed, the programme, and its significant fuel cycle assets, would simply fade from the collective Swedish memory. As the veterans from the programme now quite old, some fear that the true depth of the programme will never be uncovered. In the meanwhile, Sweden’s programme emphasises how easy it is to hide a weapons effort under the guise of a civilian project. It also shows how simple it is to obfuscate a country’s intentions in the name of ‘defensive military research’.

I’m off to India this week. But when I return, I’ll post a list of some of the facilities associated with the programme. I’ll include a description of the well preserved R3/Adam reactor. And, of course, details of R4/Eva, the finished HWR that was never fuelled.


Today’s The letter from Iran to the IAEA is indeed a worrying development. As Jeffrey points out, our worst fears have been realized. Iran has – using the fact that IAEA inspectors cannot venture beyond declared strategic points – constructed a new enrichment facility. National technical means combined with human intelligence, however, seems to have averted the worst outcome, the establishment of a parallel fuel cycle.

However, two serious questions remain:

First, the unclassified US talking points state that the facility would be capable of producing about a weapons worth of material per year. A 3,000 centrifuge facility using Iran’s antiquated IR-1 centrifuges would be able to produce about one and a half weapons worth of high enriched uranium per year (39 kilograms with tails set to 0.4 per cent and 34 kilograms with tails set to 0.3 per cent). If equipped with more advanced centrifuges, the facility becomes quite lethal. The last generation of SNOR designs, for instance, if installed in Qom, could easily produce up to 80 kilograms worth of weapons grade uranium per year. The centrifuges would require little room, about 30 meters square, and draw very little power.

In practice, the facility would probably have three product areas: one for low enrichment, one for intermediate enrichment, and one for enrichment up to weapons grade. In a later briefing, US White House officials said the size of a facility would be about right “for a bomb or two a year”. They also said that the facility was “very heavily protected, very heavily disguised”.

Second, the facility would need to be supplied between 18 and 40 metric tons of natural uranium hexafluoride gas per year. An attempt to divert this from Esfahan, Iran’s only declared conversion facility, would entail a diversion of about seven to 16 per cent of its total capacity. I don’t have to do MUF calculations on that. A diversion that big would, with an extremely high probability, be detected by the IAEA. Indeed, I almost dare to say that detection is assured. This means that Iran would need to set up a clandestine conversion facility somewhere in order to bypass safeguards. There is nothing in the US speaking notes on that. But it seems like that the US intelligence community is keeping an eye on this.

Finally, we are definitely looking at a safeguards violation. It’s worth recalling that Iran did upgrade its subsidiary arrangements to oblige them to report facilities to the IAEA when they were at the design stage. They did this in 2003. They unilaterally pulled out of this arrangement in 2007. As James Acton correctly points out, the arrangement entered into force through simple exchange of letters. As in any contract, the principle pacta sunt servanda prevails (just put that term in Google).

A state can no more pull out of a contract than you can get out of, say, a mobile phone contract before it expires. It takes two parties to terminate an agreement. And the IAEA never accepted Iran’s withdrawal.

Not that it matters. It would seem like construction started at some time before March 2007. That is, at a time when even Iran itself considered itself bound by Code 3.1.

This is going to be very difficult to explain away, even by Iran’s highly talented spin-doctors.


More data is becoming available on Monday’s nuclear test. NORSAR has published the waveform data from two of their stations. The primary wave is very noticeable and sharp, which indicates a man-made explosive event (earthquakes tend to brew a while before really making noise).

The shear-wave seems to come in shortly thereafter, giving a distinct peak. I like to think of the difference between p and s-waves as flash and thunder. The p-wave comes in fast (in air it travels at the speed of sound) whereas the s-wave rumbles in afterwards. Now, I only have these two datasets, and no other forms, but the signal from ARCES looks peculiar to me. About 10 seconds in there is a sharp fluctuation in the peak-to-peak amplitude, which isn’t visible in the NOA readout. I’ve heard that the seismologists at the IMS division is confused about some of the data as well, but I’m not sure if it’s that peculiarity that they’ve focussed on.

When the s-waves hit, you’ll notice some refraction.

If you want to check that these forms match the time of the test, please consult the p-wave travel time image below.

Updated: those interested in antipodal seismometry (see Geoff’s post below) might find the image interesting for other reasons.

Most interestingly, the CTBTO has published the error ellipses and visualised them in Google Earth. As you can see, both error ellipses define a search area well within the 1,000 sq. km. maximum search grid stipulated by the treaty. In other words, if the CTBTO hypothetically were to conduct an OSI, they would have a pretty good idea where to start.

But these are only some of the goodies to come. I got an e-mail from Sean O’Connor this morning, who wrote that he’s found the likely location of at least two additional sites in the area. He’s going to publish his findings on IMINT & Analysis quite soon.

The yield estimate is still highly uncertain, but is likely to be below 4-8 kT range that has been reported in mainstream media so far. But given the discussion in my last thread, I’m attaching this nice Mb – Yield graph for the community to have a look at.

Note that using the NTS hard rock formula puts the yield at 1.6 kT, which fits the assessment of the United States.

Finally, the word is that meteorological conditions at the moment are favourable for a first noble gas hit in South Korea, maybe also in Ussurijsk and later in Japan. But that’s for another post.


As always when the DPRK tests, I’m in a seminar somewhere else. This time, I was discussing FMCT verification with the good people of SIPRI and the diplomats of the Conference on Disarmament.

The plot is based on the Mb to yield estimates for dry-rock, close coupled, underground nuclear explosions. I’ve used the USGS estimate for the yield calculation. The graph can only be used as a general indication because the exact geological conditions of the area is not known. It is known to have a shallow water table, which could explain why the Russians always get their yield calculations on the high end. We also don’t know how well this test was coupled to the rock.

However, as indicated, today’s North Korean test seems to be significantly bigger than its previous test. What is also interesting is that the centre of today’s test, according to the USGS, is about 5.5 kilometers away from the old test site. The IMS puts the test closer to the first site, but within the USGS margin of error. The USGS sets the error to +/- 3.8 kilometers, which strongly suggest that we’re looking at a second test site in close proximity to the first.

Interestingly, a couple of years ago I learned that the South Koreans were looking for test site preparations on the “northern side of the mountain” relative from the first site. This, to me, means that the second site has been known to the South for some time.

I’m now off to have some dinner.


Cross-posted from Verification, Implementation and Compliance.

The monitors have been switched off. The cameras are being removed from their mounts, and the seals are broken. The guesthouse just off the main site no longer houses the IAEA three-person team or the four-person US experts group. By now, the equipment is probably being packed into boxes by the former North Korean hosts, after which it will be carefully catalogued and transferred to storage in some building on the sprawling Yongbyon site. There, it will gather dust until the next time inspectors visit the facility. That is, if there will be a next time.

North Korea threatens to restart the facility, and there have been some educated guesses as to how fast this could be done. These guesses range from a couple of weeks, to six months, to possibly longer. Undeniably, it will take a year to get the entire facility back in order again, but some critical processes, such as the reprocessing of spent fuel, might get up and running by the summer of 2009. And this is possibly why the Russian Foreign Minister is about to visit Pyongyang quite soon, and why the Chinese are placing frantic phone calls to Washington DC.

But what are the North Korean’s required to do to get the plant up and running again? Despite wishes to the contrary, the agreed minute on disablement was never released to the wider arms control community. However, some details were nevertheless leaked, quite possibly since some involved principals on the US side felt that the disablement steps were wholly inadequate.

The first disablement action was to unload the 5MWe reactor, and transfer spent fuel to the cooling pond. This action does not appear to have been completed. The North Korean’s would now speed up their unloading operations, and transfer the remaining spent fuel rods to the cooling pond. It is possible that they would then ask the director of the Fuel Manufacturing plant to transfer the fresh load of fuel (pictured) to the GCR for reloading.

However, a number of immediate tasks would need to be completed before then. First, the reactor’s director would need to instruct his people to repipe the secondary cooling system and, obviously, rebuild the cooling tower, or jury-rig the system somehow. This is not likely to be completed before summer, so do not expect to see steam rising over Yongbyon until autumn. Naturally, the construction of the tower can be tracked by satellite. The reactor also needs to have its control rod mechanism reconnected.

At the reprocessing facility, work may progress slightly faster. The drive mechanism between the spent fuel receiving building and the hot cells need to be reconnected, and two steam lines would need to be re-attached and pressure-tested. Moreover, the drive mechanism for fuel cask transfers needs to be replaced, as well as some hot-cell doors. After these tasks are completed, the reprocessing facility is mostly ready for action. This can be done fairly soon, possible before July. The start of a reprocessing campaign can be detected through the release of radionuclides into the atmosphere.

The Fuel Fabrication Plant has also undergone some ‘disablement’. In order to get the plant back in operation, the site director needs to reinstall all three uranium ore concentrate dissolver tanks, all seven uranium conversion furnaces, metal casting furnaces and the vacuum system, and eight machining lathes. Again, this is something that can be done in a matter of months.

The pressing question is, of course, what happens next? The ejection of IAEA monitors and US experts will lead to a substantial degradation in knowledge of ground truth. While the North is unlikely to substantially add to its fissile material stockpile in 2009, larger scale production may be likely in the coming year. Of course, a new nuclear test cannot be ruled out. It’s very likely, even, that the test site director has already received instructions to elevate his level of readiness.

Personally, I find it very difficult to see any easy way out of this predicament.


Cross-posted from Verification, Implementation and Compliance

At the present, I believe that the likelihood of an Iranian break-out is slim. The principal reason for this argument is that Iran’s installed capacity at the uranium enrichment plant in Natanz is still low, and that a break-out would entail significant political and security risks for the country. As long as Agency safeguards are in place at the Iranian sites, the international community is likely to get advance warning of any attempt to divert material or to use the existing facilities for nefarious purposes.

The problem is that not all of the nuclear fuel cycle is under safeguards. Processes downstream from the uranium conversion facility are generally covered. But uranium mining and milling as well as certain nuclear related activities (such as research centres or centrifuge assembly sites) are not monitored. Since this is the case, it is easy for a fairly technologically advanced state to construct a parallel nuclear fuel cycle, using indigenous uranium resources to fuel a clandestine weapons programme.

Figure 1: Safeguards under INFCIRC/153

The most appealing option for the cheater is to divert material where safeguards are not applied, in this case the nuclear ore. Let’s take Iran as an example. At present, Iran’s stockpile of uranium yellowcake is unknown. The only thing that is known, really, is that the country imported 600 metric tonnes from South Africa in the 1970s. If the Iranians have used that material up until January 2009, it would have about 188 tonnes of yellowcake left by now. But again, material accountancy is not carried out, and Iran is under no obligation to give answers if asked.

Status of Iran’s mines
The status of Iran’s two known mines is largely unknown, but the OECD publication Uranium 2007 at least sheds some light on the status of activities. In Saghand, the AEOI is presently engaged in sinking two cylindrical shafts, each having 4 meters in diameter and extending 350 meters in depth, as well as tunnelling (about 620 meters in total). All projects are scheduled to be implemented by the end of 2009. Ore will be excavated using the “room and pillar”, “cut and fill” and “sub-level stoping” methods.

Mining activities are on-going in the Gchine salt plug near Bandar-Abbas. This is an open-pit mine, and mining operations have been on-going since 2006. Its ore is being transported to Iran’s only operating uranium production centre (the BUP), which is capable of treating 48 tonnes of uranium ore per day. It has a production capacity of 21 tonnes of uranium per year. Iran’s second production facility lies near Ardakan, has a production capacity of 50 tonnes of uranium per year, and is scheduled to go on-stream later in 2009. Iran’s reasonably assured resources of uranium is very low, some 591 tonnes of uranium, and its inferred resources are not much higher, about 1,356 tonnes, most of it in metasomatite rock.

If the OECD’s figures are correct, it is possible to calculate how much ore would be left in the Gchine salt mine by the end of 2010 if the BUP operates as declared. This calculation is visualized in figure 2.

Figure 2: Mining

Consequently, by the end of 2010, about 60 per cent of the deposits at Gchine would be exploited. The mine would be more or less drained by 2014.

It is also possible to estimate the stockpile of domestically produced yellowcake, again if the BUP operates as declared and if it uses the acid leach solvent extraction process.

Figure 3: Milling

By the beginning of 2009, the stockpile would be some 42 metric tonnes of yellowcake. Probability statistics show that the actual stockpile in 2009 is somewhere between 9.2 and 33.8 metric tonnes (obviously the absence of data leads to an enormous uncertainty – and this is again assuming that the OECD has provided accurate information).

As indicated above, with only the comprehensive safeguards agreement in place, it is virtually impossible to keep track of this stockpile.

In order to be enriched, the yellowcake would obviously need to be processed further. And here is the catch if Iran would want to cheat. Safeguards at uranium conversion facilities are generally quite effective, especially if the throughput is low, and this more or less excludes using the uranium conversion facility at Esfahan for processing the yellowcake. Once it gets on the Agency’s books, the material is tracked downstream, and diversion becomes risky business.

Therefore, a state determined to cheat on its non-proliferation obligations would need not only to construct a clandestine uranium enrichment plant, but also a clandestine conversion facility. This facility would not need to be large; a capacity of 10 metric tonnes of uranium per year would be more than sufficient. However, it is an additional investment and it carries with it a risk of overhead or ground detection. The centrifuge facility could be minimal.

About 1,300 IR-1 centrifuges would be able to produce enough highly enriched material for one weapon per year. The cascade hall would require about 520 square meters of space (that’s 23 by 23 meters) so the entire operation could be comfortably hidden in a factory building somewhere (amusingly, old clock factories seem to be the preferred choice). It would not require more electricity than an average workshop, so it cannot be detected by a passive infrared survey.

If Gchine is operational, there is enough unsafeguarded yellowcake for 1-5 weapons stored somewhere in Iran. The potential of this material being used in a parallel fuel cycle is the real cause for concern and not a diversion or break-out scenario using declared and safeguarded facilities.

The importance of the additional protocol
This is why it is critically important that Iran reapplies the additional protocol. This instrument allows the Agency to ask for and receive information on Iran’s mines (as well as several other activities – such as the assembly of centrifuge rotors). This information can be followed up upon by means of complementary access. The scope of the Additional Protocol is best illustrated by figure 4, which I again have borrowed from a friend’s presentation.

Figure 4: Safeguards under INFCIRC/540

It is only through the additional protocol that the Agency can provide some assurance of the absence of undeclared nuclear activities on Iran’s territory. It can do so since it will be able to analyze a much broader array of information. Using this information, they can see whether the flows match up, if only approximately.

The application of the additional protocol will not by itself be able to answer many of the question-marks currently plaguing the Iranian file. For this, transparency measures going beyond the requirement of the additional protocol will be necessary. This is not something the Iranian government seems willing to implement at the moment.

This is unfortunate since for as long as this kind of transparency is not given, the file will never close.