Shortly after midnight on 1 November 1986, fire broke out in a chemical warehouse at the Sandoz site in Schweizerhalle, on the Rhine just upstream of Basel. Nobody died. The water used to fight the fire, however, washed insecticides, fungicides containing mercury compounds and other products into the river. That runoff killed fish, eels above all, for hundreds of kilometers downstream and forced waterworks in several countries to close their intakes. This guide covers the fire, the disputed release figures, the slow warning downstream, the Rhine Action Programme and the European rules that followed, and why firewater containment belongs in a local emergency plan.

1 November 1986: a night fire in Warehouse 956 at Schweizerhalle

Schweizerhalle is an industrial area on the south bank of the Rhine in the municipality of Muttenz, in the canton of Basel-Landschaft, a few kilometers upstream of the city of Basel. In 1986 the chemical company Sandoz operated a site there. One of its buildings, identified in most accounts as Warehouse 956, was used to store products, a large share of them agricultural chemicals. Fire was discovered in that building shortly after midnight on Saturday, 1 November 1986. It spread through the stored material quickly enough that the building was fully involved within a short time, and accounts describe drums and containers bursting as the heat reached them.

The Sandoz works fire brigade was joined by public fire services from Basel and the surrounding communities, and accounts describe a large commitment of personnel and apparatus through the rest of the night. Their main effort, as it is usually described, was to apply enough water to stop the fire from spreading to neighboring buildings on a site that held other chemical stores. By morning the fire was out or under control, depending on which account you read. I have not seen a reliable minute-by-minute timeline in English, and readers who need one should go to the Swiss cantonal and federal documentation rather than to retellings, including this one.

The record shows no deaths from the fire, the smoke or the release. That fact needs stating plainly, because Schweizerhalle is often listed beside Bhopal in summaries of industrial disasters, and the pairing can imply a human toll it did not have. During the early morning hours sirens sounded, and people in the Basel region were told to stay indoors and keep their windows closed while a strong, unpleasant smell from sulfur-containing combustion products spread over the city. Accounts describe people seeking medical attention for irritation of the eyes and airways, headaches and nausea, but I have not seen a figure for the number treated that I would repeat with confidence.

The harm that made Schweizerhalle a turning point in European chemical safety was done to the river. The fire destroyed a warehouse, and the water used to fight it carried part of the warehouse contents into the Rhine. From Basel the river flows north along the French and German border, then through Germany and the Netherlands to the North Sea. That mechanism is the subject of this guide. It is the part of the story that applies to a fire department anywhere with a pesticide distributor, a pool chemical warehouse or a farm supply store on ground that drains to a creek.

What the building held and the cause investigators reported

The quantity stored in the warehouse is most commonly given as about 1,350 tonnes of chemicals. The inventory included organophosphate insecticides, a class of compounds that is highly toxic to fish and aquatic invertebrates at low concentrations, along with herbicides, fungicides, solvents and dyes. Some of the fungicides contained organic mercury compounds, which is why mercury appears in every account of the release. Most accounts attribute the red color that observers reported in the river the next day to the dyes. Detailed product lists appear in the scientific literature published after the fire, and readers who need specific compounds should go to that literature and to the documentation of the International Commission for the Protection of the Rhine rather than rely on a summary.

The building was a general warehouse. The accounts I have read describe it as lacking the automatic sprinkler protection and firewater retention that a modern store of agrochemicals in Switzerland or the European Union would be required to have. That describes the standards of a later period. It is not a finding that the building broke the rules in force in 1986, which I am not in a position to make. What matters for planning is that a large inventory of substances dangerous to water sat in a building whose drainage led to the river.

The investigation into the cause concluded that the fire most likely began in pallets of Prussian blue, a pigment. Those pallets had been shrink-wrapped in plastic film with a heat tool the day before, and the investigators concluded that the material smoldered unnoticed for hours before breaking into open flame. That is the finding most commonly reported, and I give it as the investigators’ conclusion rather than as a certainty. Claims of deliberate sabotage have circulated in the years since, including allegations naming foreign intelligence services. I have not seen any of them substantiated by an official finding, so I treat them as claims.

For a fire officer, the cause matters less than what the inventory meant once the fire started. A warehouse holding packaged products in mixed lots changes its contents with orders and seasons, so the hazard written into a preplan two years ago may not match what is on the floor. A distributor’s spring stock of insecticides presents a different runoff problem from the same building in December. A preplan for such a building is therefore only as good as the latest inventory the operator has reported. In the United States that means the facility’s annual hazardous chemical inventory filing, which is a snapshot that may be many months old by the night of a fire.

Ten thousand cubic meters of water and no basin to hold it

Accounts commonly put the volume of water used on the fire at between 10,000 and 15,000 cubic meters, and some secondary sources give higher figures. Ten thousand cubic meters is ten million liters, or roughly 2.6 million US gallons, which fits what a large defensive operation flowing master streams for several hours would produce. The site had no retention basin that could hold anything close to that volume. The water ran off the fire area, picked up dissolved and suspended material from the burning and ruptured containers, and drained to the Rhine.

Not all of the contaminated water reached the river. Part of it soaked into the ground beneath and around the warehouse and left the soil heavily contaminated, and cleanup of the site continued for years. The water that went into the soil became a long-term groundwater concern, while the water that went into the river killed fish within hours. Both consequences followed from applying water on a site without containment. The fire officers made that choice for sound reasons, with the information and the drainage they had in front of them.

That choice is the core of the lesson, and it is easy to judge unfairly in hindsight. An incident commander facing a warehouse fire on a chemical site has adjacent buildings, other stores and people downwind to protect, and water is the tool that protects them. Every liter applied has to go somewhere, though, and on a site without containment it goes wherever the grades and the drains send it. Firewater guidance developed in Europe since then, including guidance from UK environmental regulators, discusses the option of a controlled burn. In a controlled burn, the fire is allowed to consume the stock under supervision while exposures are protected and water on the fire itself is limited, because combustion destroys much of a pesticide that runoff would carry away intact. The cost is a heavier smoke plume. The decision belongs to the incident commander, taken with environmental advice and with the plume weighed against the runoff, and it can be made well only if someone has worked through it in the preplan.

How much reached the Rhine: the figures and why they disagree

The quantity of chemicals that reached the Rhine is the most often misquoted figure in the story. General accounts most often repeat a figure of about 30 tonnes entering the river with the firewater, and that is the number many readers will have seen. Lower estimates also appear in published sources. I have not been able to tie each figure to a single authoritative table that I could cite by title and page, so I give the spread rather than one number. The official assessments are held by the International Commission for the Protection of the Rhine and the Swiss federal environmental authorities, and anyone citing a figure professionally should take it from them and name the document.

Mercury is a similar case. Organic mercury fungicides were part of the stock, and published estimates of how much mercury reached the river differ from source to source. Mercury mattered out of proportion to its mass because it persists in river sediment and can enter the food chain long after the acute plume has passed, so the sediment below the site remained a concern for years. I am not going to pick one of the mercury figures and present it as settled.

The figures disagree for good reasons, and none of them requires anyone to have been careless. Part of the stock burned, part was carried off in the smoke, part stayed in the debris, part went into the soil and part went into the river, and nobody could weigh what went down each path. There are two ways to estimate the river load. One measures concentrations in river samples as the plume passed and integrates them over the flow. The other works backward from the inventory and from what the fire destroyed. The two methods will not produce the same number. Estimates published in the first days came out before either method had been carried through, and some of them were repeated for years afterward.

For a planner, the useful habit is the one this site applies to casualty figures. A release quantity from a chemical fire is an estimate with a method behind it, and the method should travel with the number. When an after-action report or a state environmental agency gives a figure, record who gave it, when they gave it and how it was derived, and expect a later revision.

A red plume, dead eels and closed water intakes downstream

The firewater entered the Rhine during the night. By the next day, observers downstream of Basel reported the red discoloration in the water. The plume moved downstream at the speed of the river, and dead fish appeared along the banks over the following days. Eels were hit hardest. Most accounts describe the eel population as wiped out along a stretch of river measured in hundreds of kilometers. Other fish species, and the aquatic invertebrates that fish feed on, suffered heavy losses over shorter reaches. Published estimates of the total number of fish killed vary, and I would treat any precise count as uncertain.

The Rhine supplies drinking water to millions of people downstream. Much of that water is drawn through bank filtration wells along the river and through surface intakes in Germany and the Netherlands. As the plume approached, waterworks closed intakes or switched to other sources. Timing those closures depended on knowing what was in the water and when it would arrive. Farmers drawing irrigation water and operators of facilities on the river needed the same two facts, and in the first days they were hard to get, which is the subject of the next section.

The river recovered faster than many people feared at the time. Fish populations in the affected reaches rebuilt over the following years, helped by improvements in wastewater treatment that were already under way before the fire. The contaminated sediment and the soil under the site took much longer to deal with. In Basel and downstream, the company’s name became shorthand for industrial pollution of the Rhine. Sandoz settled compensation claims and funded environmental work afterward, and in 1996 it merged with Ciba-Geigy to form Novartis. The ICPR’s own histories present the accident as the trigger for the Rhine Action Programme, which the river’s recovery has to be read alongside.

Sirens in Basel and a slow message to the countries downstream

Two separate warnings were needed that night. One had to reach the people of the Basel region about the smoke, and the other had to reach the countries downstream about the river. Both drew criticism afterward. In Basel, sirens sounded in the early morning and the instruction to stay indoors was carried by radio. Accounts from the time describe residents who heard the sirens without knowing what they meant or what to do. The episode is often cited as an example of how little information a siren carries without a channel that tells people what action to take.

Downstream, the complaint was about speed and completeness. Governments and water authorities in Germany, France and the Netherlands said word of the release had reached them late and that early information about the substances and quantities was incomplete. Waterworks were left to make intake decisions with partial information. The Swiss authorities and the company were criticized in the weeks that followed, and the matter became a diplomatic problem as well as an environmental one. I have not seen a reliable account that fixes the exact time each downstream authority was first notified, and I would not repeat a specific delay figure without one.

The doctrine that came out of this is simple to state. A transboundary warning about a river release has to carry the substance names, an estimate of the quantity, the time and place the release entered the water, and a forecast of when the leading edge will reach each downstream intake. It also has to go to a contact point that the downstream authority has agreed in advance will receive it at any hour. At Schweizerhalle those elements arrived piecemeal and through improvised channels, so each downstream operator had to work out the hazard for itself while the plume moved toward its intake.

The ICPR, the alarm plan and the Rhine Action Programme of 1987

The International Commission for the Protection of the Rhine, usually shortened to ICPR, had existed since 1950 as the forum through which the Rhine states worked on pollution. Before Schweizerhalle its progress was slow and concentrated on particular pollutants such as chlorides. After the fire, ministers from the Rhine states met over the following months. In 1987 they adopted the Rhine Action Programme, which the ICPR’s histories present as the direct political response to the accident.

The programme set ecological goals as well as pollution targets, which was unusual at the time. Its best-known aim was that species that had vanished from the river, with salmon as the symbol, should be able to return by the year 2000, and the salmon goal gave the programme the name many people remember it by. It also set a goal commonly summarized as halving the discharges of priority hazardous substances within about a decade, and it addressed drinking water supply and the condition of the river’s sediment. Salmon have since been recorded returning to parts of the Rhine system, largely supported by stocking. A self-sustaining population across the river had not been established by 2000, and the ICPR continues to report on that work.

The parts of the programme closest to emergency management were the safety of industrial installations and the international warning arrangements. The ICPR’s warning and alarm plan for the Rhine links principal warning centers along the river, so that a release reported at one point is relayed to authorities and water operators downstream, and the plan was strengthened after the fire. The ICPR also produced recommendations on preventing accidental releases from industrial plants. Readers who want those recommendations, including anything specific on firewater retention, should take them from the ICPR’s publications, because I would otherwise be summarizing from memory.

In 1999 the Rhine states signed a new Convention on the Protection of the Rhine, which replaced the earlier treaty arrangements and broadened the commission’s remit to cover the river as an ecosystem. For a responder anywhere, the most transferable piece is the standing, tested chain of contacts along a shared river that every party knows will be used when something goes into the water. In November 1986 that chain was improvised, and afterward it was written down and exercised.

Seveso amendments, Seveso II, the Swiss ordinance and the Helsinki convention

Schweizerhalle came four years after the European Community adopted the original Seveso Directive in 1982, which the site’s article on Seveso 1976 covers. That directive was amended twice in the late 1980s. The amendments, which are commonly linked to both Bhopal and Schweizerhalle, widened its treatment of the storage of dangerous substances, the kind of gap a warehouse fire on a chemical site had exposed. Switzerland was not a member of the Community and acted separately. Its federal Major Accidents Ordinance of 1991 set up a national regime for facilities holding dangerous substances above threshold quantities, with risk assessment by the operator and review by the cantonal authorities.

The broader rewrite came with Council Directive 96/82/EC, adopted in December 1996 and known as Seveso II, which replaced the original directive. Seveso II required operators of covered establishments to have a major accident prevention policy and a safety management system. For the higher-tier sites, it required internal emergency plans from operators and external emergency plans from the authorities. It also brought land-use planning around hazardous sites into the framework, addressed domino effects between neighboring establishments and strengthened the duty to inform the public. Directive 2012/18/EU, known as Seveso III, replaced it in 2012. As far as I have seen, the directive text does not spell out firewater retention as a specific requirement. That requirement arrives through national rules and guidance and through the competent authority’s review of the operator’s assessment. Germany, for example, adopted a guideline in the early 1990s on firewater retention at sites storing substances hazardous to water.

At the international level, the UN Economic Commission for Europe adopted the Convention on the Transboundary Effects of Industrial Accidents in Helsinki in 1992. The convention covers prevention of, preparedness for and response to industrial accidents that can affect another country. It requires parties to set up notification systems so that an accident with possible transboundary effects is reported promptly to the affected party, and it provides for mutual assistance. The problem it addresses is the one Schweizerhalle exposed, and the UNECE secretariat publishes the text and the guidance that accompanies it.

The United States has no direct equivalent to Seveso II. The Emergency Planning and Community Right-to-Know Act, signed in October 1986 a few weeks before the fire and prompted by Bhopal, created local emergency planning committees and the annual hazardous chemical inventory reports that tell planners what a facility holds. The site’s article on Bhopal covers that law. Where firewater containment at hazardous materials storage is addressed at all, it is through the fire code each jurisdiction adopts. The International Fire Code includes spill control and secondary containment provisions for hazardous materials, some of which account for sprinkler discharge. Which edition is in force, and how it applies to an existing building, is a question for the local fire marshal or code official.

What to do at your agency

• The fire marshal or prevention bureau chief obtains the current hazardous chemical inventory filings for the jurisdiction from the local emergency planning committee or state emergency response commission. From them, the marshal produces a written list of facilities storing pesticides, fertilizers or other substances dangerous to water whose storm drainage reaches a creek, river, lake or reservoir.
• The preplanning coordinator updates the preplan for each facility on that list with the storm drain locations, the receiving water, any containment and its capacity, and the location of any storm system shutoff valve. The coordinator confirms each entry on a walk-through with the facility operator.
• The hazmat team leader confirms that the apparatus carries enough drain covers, booms and absorbent to block the storm inlets around a large commercial fire. The team leader also records the date crews last practiced placing them and schedules a drill if that date is more than a year old.
• The communications center manager adds two notifications to the run card for a working fire at any listed facility: the state environmental spill reporting line and the operator of the nearest downstream drinking water intake. The manager verifies both around-the-clock numbers with a test call this month.
• The emergency management coordinator puts firewater runoff on the agenda of the next scheduled LEPC meeting and asks each facility representative present to state how many minutes of firefighting water their containment holds and where it overflows.
• The training officer writes one paragraph into the existing hazardous materials guideline naming who advises the incident commander on runoff and on the controlled burn option at a pesticide or fertilizer storage fire, and how that adviser is reached at night.
• Where the jurisdiction shares a river with a downstream county or state, the emergency management coordinator confirms the downstream counterpart’s contact point for a release and writes it into the hazardous materials annex of the emergency operations plan.

Takeaways

• The Sandoz warehouse fire at Schweizerhalle, near Basel, was discovered shortly after midnight on 1 November 1986, and the record shows no human deaths from the fire or the release.
• The warehouse held about 1,350 tonnes of chemicals by the most common account, including organophosphate insecticides and fungicides containing organic mercury compounds.
• Accounts put the firefighting water at between 10,000 and 15,000 cubic meters, and with no retention on site it ran into the Rhine and soaked into the soil beneath the warehouse.
• About 30 tonnes is the release figure most often repeated, lower estimates also appear, and the official figures should be taken from the ICPR and the Swiss federal environmental authorities with the source named.
• Eels were killed over hundreds of kilometers of river, downstream waterworks closed their intakes as the plume passed, and the river recovered faster than many people feared at the time.
• Downstream governments complained that notification was slow and incomplete, and the response included a strengthened ICPR warning and alarm plan and the Rhine Action Programme of 1987.
• The Seveso amendments of the late 1980s, Seveso II in 1996, the Swiss Major Accidents Ordinance of 1991 and the UNECE Helsinki convention of 1992 all address what Schweizerhalle exposed, while firewater retention itself arrives through national guidance and local fire codes.
• Firewater from a fire involving chemicals dangerous to water is part of the release, so the preplan needs the drains, the receiving water, the containment volume and the downstream intake operator’s number.