A steel tank on Commercial Street in Boston’s North End failed at lunchtime on 15 January 1919 and killed 21 people with a wave of molasses, which is the kind of detail that turns a disaster into a punchline and keeps the useful part of it from being read. The useful part sits in the litigation record: a tank whose drawings were approved by a company treasurer with no technical training, leaks reported for three years and painted over, a corporate defense that blamed anarchist sabotage, and an auditor’s finding in 1925 that helped change who is allowed to sign a set of construction plans.

The tank, the schedule, and the man who approved the drawings

United States Industrial Alcohol built the tank in 1915 on the Commercial Street waterfront, operating the site through its subsidiary, the Purity Distilling Company, because molasses shipped up from the Caribbean could be pumped off a vessel at the wharf and held there until it moved on for conversion into industrial alcohol, which during the war was going into munitions manufacture. The company picked the location for the pier, and the neighborhood came attached to it, since the North End at that time was a dense Italian immigrant district with a freight yard, a city paving department yard, a fire station and the Boston Elevated Railway’s Atlantic Avenue structure all standing within reach of whatever that tank might do.

Published descriptions of the tank are consistent on the general dimensions, roughly 50 feet tall and about 90 feet in diameter, built of riveted steel plates that stepped down in thickness toward the top in the normal way for a cylindrical tank. Capacity is reported variously between about 2.3 and 2.5 million gallons depending on the source, and the quantity held at the time of failure is usually given as about 2.3 million gallons, which is something on the order of 13,000 tons of liquid. Anyone quoting a precise figure should take it from the case record rather than from a summary, because the numbers repeated in popular accounts do not all agree.

The part that matters for anyone in the inspection business came out in the hearings that followed. The construction was overseen for the company by Arthur P. Jell, its treasurer, who had no engineering or architectural training, and the drawings were not independently reviewed by a qualified engineer before the tank was built or before it was filled. Jell was working against the arrival of a molasses shipment, and instead of a full hydrostatic test, which means filling the tank with water and watching what it does under the load it will actually carry, the tank received only a few inches of water before it went into service. The Boston Building Department issued a permit, and the permitting practice of the day did not require what the disaster later made routine.

What happened on 15 January 1919

The weather that week had swung hard, with temperatures in the Boston area running near zero degrees Fahrenheit in the days before and rising to around 40 degrees on the 15th, and a fresh warm shipment had been pumped into the tank two days earlier on top of the older, colder contents. Shortly after noon the tank came apart. Witnesses described a sound like machine gun fire, which is what a line of rivets does when the plates tear away from them, and the shell separated so that the contents went out across Commercial Street in a single release rather than through a crack.

The wave height and speed that circulate in every retelling, commonly given as an initial surge in the range of 15 to 40 feet moving at around 35 miles per hour, are estimates reconstructed from witness accounts and physics rather than measurements anyone took, and they should be read that way. What is documented in the damage record is that a support column of the elevated railway structure was displaced and a section of track dropped, with an approaching train stopped short of the gap, and that buildings along the street were knocked off their foundations and carried. Twenty-one people died. The injured are usually counted at about 150, although the figure varies between accounts and depends on how the compiler treated people who were treated and released.

The dead included teamsters and city paving department workers who were at lunch, a firefighter in the station next door, residents of the houses along the street and two ten-year-old children who were gathering firewood near the wharf. Several victims were not recovered for days because the molasses cooled into a mass that had to be cut and dug through, and the last body was not found until months later, having been carried into the harbor. Those recovery timelines are worth holding onto, because they describe what happens when the medium sets up around the search area.

The response, including a fire station inside the debris field

The Boston Fire Department station on the wharf, home to Engine 31, was pushed off its foundation and partially collapsed. Firefighter George Layhe was trapped in the wreckage and died there while his own crew worked on him, which is the detail I would put in front of any chief who has never looked at a station location as a hazard exposure. The station was where it was because it protected the waterfront, and the thing it was protecting the waterfront from was standing across the street.

The people who arrived were whoever was close. Cadets from a nautical school training ship berthed nearby came ashore and worked in the wreckage, Boston police and fire units responded, and the Red Cross set up to handle the injured and the displaced. The extrication problem was unlike anything in the local experience, because the molasses filled voids, coated everything it touched, and thickened as it cooled, so that the ordinary business of finding a person by sound and by sight failed and crews were reduced to probing and cutting. Identification of the dead was complicated by the same coating.

Cleanup ran for months. Crews used salt water pumped from fireboats, which cuts molasses better than fresh water, along with sand, and the residue tracked through the city on shoes and streetcar floors for a long time afterward. In 2016, Nicole Sharp and collaborators presented fluid dynamics work at the American Physical Society’s Division of Fluid Dynamics meeting examining how the cold weather affected the event, reporting that the rising viscosity of molasses as it cooled both trapped victims and slowed the rescue, and that analysis is a reasonable thing to read if your jurisdiction stores any heavy liquid in bulk.

The leaks everyone saw and nobody recorded

The tank leaked from the time it was first filled in December 1915, and it kept leaking for the three years it stood. Neighborhood residents collected molasses from the seams in pails and cans, which is the sort of thing that reads as a quaint period detail until you notice that it is a continuous, publicly visible, three-year report of a structural defect delivered by dozens of witnesses to nobody who would act on it. The company’s response to the visible seepage, established in the later hearings, was to have the tank painted brown, which made the running molasses harder to see against the shell.

At least one employee pushed the issue. Isaac Gonzales, who worked at the facility, was sufficiently worried about the tank that he checked on it at odd hours and eventually raised his concerns with the company, and he left the job before the failure. Residents also described rumbling and groaning noises from the structure. None of that produced an engineering inspection, a load restriction, a fill limit or a report to the city, because there was no channel that connected a neighbor with a pail or a worried employee to anyone with the authority and the competence to order the tank emptied.

That gap is the part of this case I find most transferable, because the observation existed and the record did not. A modern agency has the same failure available to it whenever a citizen calls the non-emergency line about a smell, a stain, a hiss or a wet spot at the base of a tank, and the call is handled conversationally and leaves no retrievable record. If your dispatch system cannot produce, on demand, every complaint associated with a specific address over the past three years, then you have reproduced the 1919 arrangement with better telephones.

The sabotage defense and why the company ran it

United States Industrial Alcohol’s position in the litigation was that the tank had been destroyed by a bomb placed by anarchists. That defense was not invented from nothing, because 1919 was a year of genuine anarchist bombing activity in the United States, the Galleanist circle was active in the Boston area, and the company had a plausible story to tell about an alcohol facility connected to munitions production in a working-class immigrant neighborhood. Making the argument required the company to assert that its neighbors, who were also its dead and injured plaintiffs, contained the person who had done this.

The cost of that argument fell on a community that had already buried 21 people and was, in the ordinary course, being asked to prove its own losses in a hearing room. It also set the shape of the case, because once sabotage was on the table the plaintiffs had to spend years and money demonstrating a negative about the cause of the failure in addition to demonstrating the company’s negligence in building and operating the tank. The claim was litigated, and it was rejected in the auditor’s findings.

I raise this because the pattern outlives the case. When a large structure fails and the owner’s first public explanation names an outside actor, the correct professional posture is to treat that as an assertion by an interested party until an independent finder of fact has examined it, and to say so plainly in your own situation reports rather than repeating the owner’s theory as background. In 1919 it took a court-appointed auditor and roughly six years to establish that the tank had failed on its own.

The auditor, the record, and the 1925 finding

More than a hundred separate claims, commonly reported as 119, were consolidated and heard under Massachusetts practice by a court-appointed auditor, Colonel Hugh W. Ogden, a Boston lawyer. The auditor’s job in that procedure is to hear the evidence and report findings of fact to the court, which is why the Boston molasses case produced something closer to a public inquiry record than most American civil litigation of the period does. Hearings ran for years, beginning in 1920, and the accumulated testimony and exhibits are measured in tens of thousands of pages, with the exact counts of witnesses and exhibits differing between published accounts, so take those numbers from the record itself rather than from a summary like this one.

Ogden’s report, issued in 1925, found for the plaintiffs. He concluded that the tank was structurally insufficient for the load it was built to carry and that it failed because of that insufficiency rather than because of an explosive device, and he rejected the sabotage defense. The findings also went to the process by which the tank had been approved and put into service, including the absence of competent engineering review of the design and the absence of a proper test before it was filled. Ogden was a lawyer weighing competing expert testimony rather than an engineer running his own analysis, which is worth noting when people describe the 1925 report as an engineering investigation.

The company settled the consolidated claims after the finding. The total paid is reported differently in different accounts, with figures commonly cited in the range of roughly $300,000 to $600,000 in 1920s dollars and per-death payments commonly given at around $7,000, and I would not put a single number in print without pulling the settlement documents, because the published figures do not agree and the composition of what was counted varies. What is not in dispute is that the company lost on the central factual question it had chosen to fight.

What later engineering analysis added

Modern structural analysis has gone back to this failure, and it is consistent with the 1925 finding while being far more specific about mechanism. Ronald Mayville, a structural engineer at Simpson Gumpertz and Heger, examined the tank around the time of the centennial and reported that the wall plates were substantially thinner than the hoop stress at that diameter and head of liquid required, leaving a factor of safety far below what was ordinary practice even in 1915. He also reported that the steel had a low manganese content by later standards, which raises the temperature at which the material stops behaving in a ductile way, and that the rivet holes acted as stress concentrations from which a crack could run.

The rivet detail matters because it explains the noise. A riveted seam under excessive hoop stress with a brittle plate does not weep and then tear slowly; it unzips, which is the mechanism behind the machine gun sound witnesses described and behind the near-instantaneous release of the whole contents rather than a manageable leak. The cold weather in the preceding days sits in the middle of this, since a steel that is already marginal becomes more brittle as it cools, and the fresh warm delivery two days earlier added both load and, by some accounts, fermentation gas pressure.

Present all of that as what it is, namely a careful engineering reconstruction of a century-old failure from surviving drawings, testimony and material knowledge, rather than as a laboratory result from the wreckage. The finding of legal record remains Ogden’s, and the later work is best read in the original if you want the assumptions and the arithmetic.

Stamped plans, licensure, and the gap that is still open

The reform most often attributed to this case is that Boston tightened its permitting so that construction drawings had to be signed and sealed by a registered professional engineer or architect and filed with the city, which closed the specific hole that let a company treasurer approve a pressure-bearing structure in a residential neighborhood. The broader movement toward mandatory engineering licensure was already under way, with Wyoming enacting the first state licensure law in 1907 and the remaining states following over the following four decades, and the molasses case is routinely cited as one of the events that gave that movement its public argument. The National Council of Examiners for Engineering and Surveying publishes the licensure history, and that is the place to check dates rather than relying on the version in any disaster retelling.

What the reform addressed was the design approval step. It did not, by itself, create a regime for inspecting a tank that is already standing and already leaking, which is the part of the 1919 story that killed people. That gap got filled much later through a combination of industry standards and federal rules, including the American Petroleum Institute’s standard for in-service inspection, repair and alteration of aboveground storage tanks, generally known as API 653, the National Fire Protection Association’s code covering flammable and combustible liquids storage, and the Environmental Protection Agency’s Spill Prevention, Control and Countermeasure requirements for oil storage facilities. Which of these applies to a given tank depends on what is in it, who owns it and what your state has adopted, so confirm the applicable requirements with your state fire marshal and state environmental agency rather than assuming coverage.

Notice the shape of that dependency, because it is the live issue for a modern jurisdiction. A tank holding a commodity that is neither flammable nor classified as oil may fall outside all of those regimes and be subject only to whatever the building code required on the day it was permitted, which in practice can mean no scheduled integrity inspection for the life of the structure. Molasses was a food-grade commodity in a neighborhood that had no reason to think of it as a hazard, and the same description fits a great many large tanks standing today holding water treatment chemicals, fertilizer solutions, food-grade liquids, brewery and distillery product, and process water.

What to do at your agency

• Have your fire prevention officer produce a written list of every aboveground bulk liquid storage tank in the jurisdiction over a threshold your department sets, with contents, capacity, year built, containment volume and the date of the last documented integrity inspection, and flag every tank where that last column is blank.
• Ask your building official or plans examiner, in writing, for the permit file on the largest tank on that list, and confirm whether the drawings were sealed by a licensed professional engineer and whether a full hydrostatic test was witnessed before the tank went into service.
• Put one item on your existing local emergency planning committee agenda asking who in the jurisdiction receives a citizen report of a leaking or bulging tank, what record that report creates, and how somebody would retrieve three years of such reports for one address.
• Have your communications supervisor or QA officer run a query in the CAD or records system for the address of your largest storage facility over the last three years and read what comes back, because a system that returns nothing on a facility the neighbors talk about has a recording problem rather than a quiet facility.
• Assign a company officer to write or update a pre-incident plan for that facility this month, covering contents, quantity, containment, the downhill path of a catastrophic release, the occupancies inside that path, and which of your own stations or radio sites sit in it.
• Add a paragraph to your existing hazardous materials annex covering a non-flammable bulk liquid release, naming the heavy rescue and technical rescue resources you would request for entrapment in a viscous medium and the mutual aid path to reach them.
• Have your public information officer adopt a standing rule, in writing, that any cause attributed by a facility owner in the first hours goes into releases and logs as an unverified statement by that owner with the owner named.

Takeaways

• The Commercial Street tank failed on 15 January 1919 and killed 21 people, with injuries usually counted at about 150, though published injury totals differ depending on how the compiler handled people treated and released.
• The tank’s construction was overseen by a company treasurer with no technical training, the drawings received no competent independent engineering review, and the tank was put into service without a proper hydrostatic test, all of which was established in the litigation that followed.
• The tank leaked visibly for three years, neighbors collected the leakage in pails, and the company had the shell painted brown, which demonstrates that the observation of a defect is worthless without a channel that connects it to somebody with authority to act.
• United States Industrial Alcohol defended the case by blaming anarchist sabotage, an assertion that shifted years of cost onto the plaintiffs and that the court-appointed auditor rejected.
• Auditor Hugh W. Ogden reported in 1925 that the tank was structurally insufficient and failed for that reason, and the company settled afterward, with published settlement totals differing enough that the figure should be taken from the case record rather than from any summary.
• Later structural analysis by engineer Ronald Mayville reported wall plates too thin for the hoop stress and steel composition that behaved brittly in cold weather, which is a reconstruction consistent with the 1925 finding rather than a replacement for it.
• The main documented reform was procedural, requiring sealed and filed construction drawings in Boston and reinforcing the national case for engineering licensure that began with Wyoming in 1907, and it addressed design approval rather than in-service inspection.
• A tank holding a commodity that is neither flammable nor classified as oil can fall outside API 653, NFPA’s flammable liquids code and EPA spill prevention rules alike, so confirm with your state fire marshal and state environmental agency which tanks in your jurisdiction are actually subject to a scheduled inspection.