Eight minutes
El Al Flight 1862 was a Boeing 747-258F, registration 4X-AXG, built in 1979 and powered by four Pratt & Whitney JT9D-7J engines. On 4 October 1992 it was flying cargo from John F. Kennedy International in New York to Ben Gurion in Tel Aviv, with a stop at Schiphol for a crew change and cargo processing. It landed at Amsterdam at 13:40, was due out at 16:30, and got a slot time of 17:20.
It began its take-off roll on runway 01L at 17:21, at a gross weight of 338.3 metric tons, with four people aboard: captain, copilot, flight engineer, and one non-revenue passenger.
At 17:27:30, climbing at flight level 65, the number 3 engine and its pylon tore away from the right wing. The separating engine struck the number 4 engine, which left the wing as well. The leading edge between them was extensively damaged and several aircraft systems went with it.
The crew declared an emergency and turned back for Schiphol. The last transmission from the aircraft carries the sound of the landing gear, then two further sounds that were afterwards identified as the stick shaker and the ground proximity warning system — the aeroplane telling its crew it was about to stall, and then that the ground was coming.
At 17:35:42 it struck the apex of two connected eleven-storey apartment blocks, Groeneveen and Klein-Kruitberg, in the Bijlmermeer in south-east Amsterdam, about thirteen kilometres from the runway it was trying to reach. Fourteen minutes and twenty-one seconds had passed since it left the ground. Fragments of the aircraft and the buildings were scattered across an area roughly 400 metres wide and 600 metres long, and the fuel explosion scorched the facades over a wide area.
What the Board found
The Netherlands Aviation Safety Board — the Raad voor de Luchtvaart — investigated, held a public hearing in The Hague on 14 and 15 October 1993, and published Aircraft Accident Report 92-11. Its statement of probable cause is worth quoting as written:
The design and certification of the B 747 pylon was found to be inadequate to provide the required level of safety. Furthermore the system to ensure structural integrity by inspection failed. This ultimately caused — probably initiated by fatigue in the inboard midspar fuse-pin — the no. 3 pylon and engine to separate from the wing in such a way that the no. 4 pylon and engine were torn off, part of the leading edge of the wing was damaged and the use of several systems was lost or limited.
There is a hard irony in the component that failed. The fuse pins are designed to break. They are hollow shear pins at the four attachment fittings between pylon and wing, and the wing structure around them is deliberately made stronger, so that an engine coming off takes the pins with it and leaves the fuel cell in the wing intact — what the report calls "a clean breakaway." The safety feature is the thing that let go, and it let go early, under load it was meant to survive, probably because of fatigue in the innermost pin. The pylon's midspar fuse pins had last been inspected on 17 June 1992, 257 flight cycles earlier.
Having lost both right-hand engines the crew had, in the Board's assessment, "very limited control of the airplane," and "because of the marginal controllability a safe landing became highly improbable, if not virtually impossible." The Board's first recommendation was to redesign the 747 pylon structure outright.
Sabotage was examined and ruled out on the evidence. A detailed investigation looked at the type of cargo, the dispatching of the aircraft, security and maintenance; the engines and pylons were inspected for signs of a high-energy explosion; structural parts and foreign objects from the wreckage were put through chemical analysis. The results were negative, and the report records that no evidence of sabotage was found. Engines 3 and 4 were recovered from water about half a kilometre from the entrance to Naarden Harbour, roughly 200 metres apart.
How many died
The report's table of injuries lists three crew as fatal, and 43 others. The figure used in the Netherlands, and on the memorial, is 43 in total — four on the aircraft and thirty-nine in the buildings.
It has been doubted since the week it happened. The Bijlmermeer housed a large immigrant population, including residents from Suriname, Aruba and Ghana, some of them undocumented and therefore on no list that a disaster roll could be checked against. What Dutch reporting generally says is that 43 people are known to have died. Nobody has ever been able to close the gap between that and the number who were actually in the two blocks at half past five on a Sunday evening.
The damage to both buildings was beyond repair and they were demolished. Hundreds of people were made homeless. The soil was heavily contaminated with fuel, oil and the combustion products of the aircraft and its freight.
The recorder that was never found
The flight data recorder was recovered, damaged by heat and water, its tape broken in four places and cracked, discoloured and contaminated — worst in the section holding the last two and a half minutes. A full readout was achieved anyway.
The cockpit voice recorder was a different matter, and the report's two sentences on it are the most quietly consequential in the document:
Despite intensive search activities to recover the CVR from the wreckage area, the recorder was not found. El Al personnel stated that a CVR had been installed in the aircraft.
So there is no record of what was said in that cockpit for eight minutes, and the Board's basis for believing a recorder was fitted at all is the airline's statement that one was. The 1999 parliamentary inquiry would later find that a criminal investigation into the missing recorder should have been opened in the first months after the crash, and was not.
The cargo
The accident report is specific about the load: 114.7 metric tons of cargo, "of which 6.5 metric tons were considered low grade dangerous goods." It does not itemise them.
Six years passed before the public learned what some of that was. In 1998 El Al confirmed that the aircraft had been carrying 190 litres of dimethyl methylphosphonate, a chemical used in the synthesis of the nerve agents sarin and soman. It was travelling from a plant in the United States, under a US Department of Commerce licence, to the Israel Institute for Biological Research at Nes Ziona. Israel's account was that the material had been entered on the cargo manifest as the regulations required, that it is not itself toxic, and that its purpose was testing the filters in chemical-weapons detectors.
Two points are worth keeping straight, because they usually get collapsed into each other. DMMP is today a Schedule 2 precursor under the Chemical Weapons Convention, CAS 756-79-6 — but that convention entered into force in April 1997, five years after the crash, so it was not a controlled substance under it at the time. And on the available record the shipment was licensed and declared. What took six years was not the paperwork. It was anyone in authority saying out loud what the paperwork contained, to people who had been breathing the smoke.
The uranium
A Boeing 747 of that generation carried depleted uranium as balance weights in the tail. The Board addressed this directly: two of the originally installed weights had been replaced with tungsten, so "not more than 400 kg of depleted uranium was involved in the crash and the fire after impact." A radiation expert from the Netherlands Energy Research Foundation told the Board that depleted uranium's low radioactivity posed no real danger, that as a heavy metal it is poisonous but can only enter the body as a soluble oxide, dust or vapour, and that its boiling point of 3,813 °C is far above anything a kerosene fire reaches.
After the site was cleared, about 150 kilograms of that uranium could not be accounted for.
In 2000 a team at RIVM, the Dutch national institute for public health and the environment, published a full risk analysis in the Journal of Hazardous Materials. They modelled the uranium source term, ran several dispersion models against the actual local geometry and the weather that evening, and estimated what a bystander could have inhaled. Uranium oxide formed in a fire has low solubility, which makes radiotoxicity rather than chemical toxicity the thing to measure. Their best estimate was a dose below 1 microsievert; their worst case put an upper bound below 1 millisievert. Both are far beneath the level at which acute effects appear, and they concluded it improbable that the missing uranium had caused the reported health complaints.
The paper is candid about its own position, and the candour is the valuable part: the analysis was run six years late, and "virtually no data were available to validate any calculated result."
A loaded flight
Residents and rescue workers went on reporting illness, and the questions about the cargo would not go away. In October 1998 the Tweede Kamer opened a full parliamentary inquiry under Theo Meijer. It held public hearings from 27 January to 12 March 1999, questioned ninety witnesses, and presented a 414-page report on 22 April 1999 titled Een beladen vlucht — "A Loaded Flight," the pun intact in Dutch.
Its central finding was about information rather than chemistry: that the Tweede Kamer had "too often" been informed incorrectly, unclearly, untimely or incompletely, and that important information had not always reached ministers either. It criticised the Aviation Safety Board for serious failures in its duties, found the recovery of wreckage careless, and said complaints should have been filed over the parts that went missing. It faulted the former justice minister, Winnie Sorgdrager, for not telling parliament about the men in white suits whom witnesses had described at the crash site, on the view that she had not found the accounts serious enough to pass on.
On the cargo itself the commission did the unglamorous thing and found the missing documents. The roughly twenty tonnes whose contents had never been established turned out to consist mainly of computer parts and electronics. And it did not establish a causal link between the cargo and the health complaints.
What the health research found
Two large medical investigations followed, and their results are more interesting than either side of the public argument.
The epidemiological study, published in the Archives of Internal Medicine in 2005, compared 334 firefighters, 834 police officers and 241 hangar workers who had sorted and examined the wreckage against colleagues who had done no disaster-related work. The data were collected a mean of 8.5 years afterwards. The exposed workers reported significantly more autoimmune-like symptoms — tingling, muscle pain, loss of strength, fatigue, a feeling of sand in the eyes across all three groups, and others specific to particular groups. In the same people, testing for antinuclear antibodies, rheumatoid factor and antineutrophil cytoplasmic and anticardiolipin antibodies found no difference at all from the unexposed. Blood, urine and saliva distinguished nobody.
A later study, in 2008, examined what taking part in that research had done to the participants. Among 1,019 rescue workers and 453 residents, health anxiety and bodily sensitivity were higher twelve weeks after the examination than before it, and reassurance from a physician worked less well. Whether someone had actually been exposed to the disaster did not predict that change; how many symptoms they already had did.
The clinician who reviewed the whole programme for the Dutch medical journal in 2005 drew the conclusion that sits least comfortably with everyone. The investigations did not reassure anybody, and complaints rose slightly after them. The findings may have been shaped by the anxiety and attention the investigations themselves created. And it is technically impossible, at this distance, to state that chemical exposure — depleted uranium included — could not have caused adverse health effects. Residents who are ill still attribute it to what was in the air.
2024
The archive of the Aviation Safety Board's investigation was classified. The secrecy was due to lapse in 2022 and was extended shortly beforehand without a clear reason given.
In April 2023 the member of parliament Pieter Omtzigt put down a motion to lift it in part, and carried almost the whole chamber. An advisory body on openness reported that July on what could be released, and on 10 September 2024 part of the archive was opened — 27 of its 81 dossiers. The remainder stays shut, or has to be separated from material that stays shut. Under the Convention on International Civil Aviation, accident investigation documents can be held for seventy-five years, which for this crash runs to 2069. Omtzigt's objection was that the sealed files include the Boeing 747 manual, and that there is no sense in classifying a manual.
What is settled and what is not
The mechanism is settled, and settled unusually well. A fatigue crack in a fuse pin designed to fail cleanly caused it to fail early; one engine took another with it; the aircraft could not be flown. The Board said so in 1994, said the design and certification had been inadequate and the inspection regime had failed, looked hard for sabotage and found none, and told Boeing to redesign the pylon.
Almost everything else about the aftermath is a record of things not said. A cargo manifest that took six years to produce a chemical its owners insist was declared all along. Twenty tonnes of freight nobody could account for until a parliamentary commission went looking for the paperwork and found computer components. A hundred and fifty kilograms of depleted uranium unaccounted for, and a risk analysis run too late to validate against anything. A cockpit voice recorder that was never recovered, and no criminal investigation into its absence when one might still have found something. A parliament formally told it had been misinformed too often. Men in white suits a minister decided not to mention. An archive partly sealed until 2069.
None of that is evidence of the thing it is usually offered as evidence for. The health research is reasonably clear: the symptoms are real and more common among the people who were there, and the antibodies and the dose calculations do not support the exposure that was supposed to have produced them. What the record does establish is narrower and harder to argue with. Thirty-nine people were killed in their homes by an aeroplane, an unknown number of others were killed who were never on a list, and for the better part of a decade the people who survived it could not get a straight answer about what had burned.