In 1977 a new tower opened in Manhattan: fifty-nine storeys, sharp-topped, and standing on what looked like four giant table legs. The legs were not at the corners. They were at the middle of each side, nine storeys tall, because a church owned one corner of the site and kept its place there. The tower’s corner simply cantilevers over it.

A building standing on four mid-side columns needs a clever way of collecting its loads, and it had one. Stacked chevrons of diagonal bracing ran down each face, gathering the forces from the floors and delivering them to the four legs. On the roof sat a block of concrete of about four hundred tonnes, on a thin film of oil, pushed by hydraulics so that it moved against the building’s sway — one of the first tuned mass dampers in a skyscraper. The structural engineer was William LeMessurier, and the building was, by every account, a brilliant piece of work.

In the spring of 1978 he learned something about it he had not known. In the summer, a student asked him a question. In the autumn, steel plates were welded over more than two hundred of its joints at night, while thousands of people worked in the offices by day and never found out.

The change nobody showed him

The original design called for the diagonal braces to be joined with full-penetration welds. During construction, the steel contractor proposed bolted joints instead — cheaper, faster, perfectly normal. The engineering office approved it. LeMessurier himself only learned about the substitution in May 1978, after the building was finished.

If this sounds familiar, it should. The last note was about a connection changed during fabrication and approved by the office without the designer’s calculation. Two notes in a row, the same villain: a change that looked like the same thing.

At first, bolts did not worry him. The building had been designed for the wind loads the code required, and the bolts were checked for those. Then came the question.

The question

In June 1978 a student contacted his office about the building. The question, as LeMessurier told it, was about those mid-side columns: wasn’t the building vulnerable to quartering winds — winds that hit it diagonally, on a corner, rather than square on a face?

He explained, patiently, why it was fine. And then, being an engineer, he went back and checked it properly. On 24 July 1978 he ran the numbers for diagonal winds. For four of the eight tiers of chevrons, a quartering wind increased the force in the braces by about 40 per cent — and the force at the bolted joints by about 160 per cent. Turn the wind yourself:

Then he asked the question every engineer should ask and most of us are afraid to: how likely is it? Weather records said that a wind strong enough to fail the weakest joints would come along, on average, once every fifty-five years — if the mass damper on the roof was working. If it lost power, which storms are good at arranging, the number dropped to once every sixteen years.

A one-in-sixteen chance, every year, for an occupied skyscraper in the middle of Manhattan.

What he did

He had options. He could have stayed quiet: the building had passed every check anyone had asked for, and nobody else knew. Instead he went to the building’s owner and told them that the tower he had designed could fall down.

What followed was one of the most remarkable repair jobs in construction history. Every night from August to October 1978, welders worked behind plywood enclosures, welding thick steel plates over the bolted joints, and cleaned up before the office workers arrived in the morning. Emergency generators were installed so that the damper could not lose power. Strain gauges went onto critical members. Evacuation plans were drawn up for the surrounding neighbourhood. A newspaper strike happened to be under way in the city, which helped keep it out of the press.

In September, with about half the joints repaired, Hurricane Ella was off the coast and heading north. The city was hours away from the evacuation plan. Then the storm turned out to sea.

The repairs were finished in October. The building is still there. LeMessurier’s story, told publicly for the first time seventeen years later in a long magazine article, became the standard case in engineering ethics courses: the engineer who found his own mistake and told the truth about it at enormous personal risk.

All of that is true. Now the parts that usually get left out.

Hard truth one: seventeen years of silence

The people who worked in the building were not told. The people who lived around it were not told that an evacuation plan existed with their streets in it. The public learned about it in 1995.

Was that right? Here is my opinion, labelled as one. The repair was right, and it was brave. The silence is much harder to defend. There is an argument that public panic could have done more harm than the risk itself — and that argument is always available to whoever holds the information, which is exactly why it should make us uneasy. An evacuation plan for people who don’t know they are in it is not really a plan for them. It is a plan about them.

Hard truth two: the student had a name

In the 1995 article, the caller was “a young man, whose name has been lost.” A good story needs only one anonymous student, so for a while there was one.

In fact, an engineering student at Princeton, Diane Hartley, had written her undergraduate thesis on the building that year and calculated the effect of quartering winds herself. She talked to an engineer at the firm; she never spoke to LeMessurier. For years, the version everybody taught left her out. And in 2022 a former architecture student, Lee DeCarolis, said he was the one who had made the phone call.

So the most famous phone call in engineering ethics now has two candidates for the student, and the one who did the most thorough calculation never spoke to the engineer at all. The story we teach about honesty spent decades not getting its own facts straight. That is not a reason to stop teaching it. It is a reason to teach it with the footnotes.

Hard truth three: maybe it never needed saving

This is the one that hurts. A reassessment by a researcher at the U.S. National Institute of Standards and Technology, published around 2019–2020 with modern methods, concluded that the quartering wind loads were not the threat that LeMessurier and Hartley had believed. The weak spot in the 1978 reasoning was the assumption in my little dial above: that the peak pressures on two adjacent faces arrive at the same time and simply add up. They don’t, quite. The reassessment suggested that whether the retrofit was really needed deserves a fresh look.

Does that make the story wrong? I don’t think so — and this is where it gets interesting for engineers. In 1978, with the tools and data he had, LeMessurier could not show that the building was safe. When you cannot show that an occupied building is safe, you do not get to bet that it is. You round in the safe direction. Archimedes did it with square roots; LeMessurier did it with two-inch steel plates.

The hard truth is not that he was wrong. It is that engineering decisions are made with the knowledge available at the time, and they can be right even when the fear behind them turns out to be bigger than the danger. The opposite mistake — a confident calculation that the building is fine, when it isn’t — does not get a magazine article. It gets an investigation.

---

What I take from it

1. The most important calculation is the one about your own work. Checking other people’s buildings is easy. LeMessurier’s real achievement was running the numbers on his own building after a stranger questioned it — and believing them.
2. Substitutions get the designer’s eyes. Welds to bolts, one rod to two: if the force path or the connection type changes, the person who understands the force path checks it. Approval by the office is not approval by the calculation.
3. Answer the question you were asked, not the one you were expecting. His first answer to the student was the confident one. His second answer, a month later, was the true one.
4. Write down your assumptions so they can be questioned. The 1978 analysis was reassessed forty years later only because its assumptions were clear enough to disagree with. A calculation that hides its assumptions cannot be corrected. It can only be trusted or not.

Good engineers find their mistakes. Great ones go looking. The best ones also tell everybody — including the people in the building.