Scientists sometimes get a little excited when an experiment doesn't work the way they expect it to. It's an opportunity to learn something new about the world, and they can always change the experiment a bit to try again next time. Engineering... is a little bit different. Designers and engineers generally prefer to find out something's wrong before it's built instead of after, because engineering failures can be disastrous.

But engineering failures have happened and, if nothing else, they're an opportunity to learn what not to do next time.

When it opened in 1878, Scotland's 3.2 km Tay Bridge was the longest bridge in the world. It was so impressive that Queen Victoria knighted Thomas Bouch, the Tay Bridge designer, after she took the train across it. But on December 28, 1879, a train was crossing the Tay Bridge during a violent storm when the middle of the bridge collapsed. The bridge, the train and the 60 or so people on board crashed into the water below— there were no survivors.

An official investigation found so many problems with the bridge's construction and design, that they couldn't even figure out what failed first. But according to their report, Bouch did just about everything wrong. They tested a lot of the bridge's surviving components, for example, and found that they couldn't stand anywhere near the pressure they were supposed to.

So the bridge would have failed eventually, anyway, and the storm just accelerated things. And when investigators looked at the columns that had been holding the bridge up, they saw that they were broken at the bottom, so the columns might have been responsible for the bridge's collapse. The winds could have made the train rock as it went along, maybe even forcing it against the bridge's walls.

If that had happened, it would have put a lot of extra stress on those columns, breaking them and bringing the bridge down. Whatever the reason the Tay Bridge failed, engineers found even more problems when they looked at some of Bouches' other projects. Some of their components were either just as badly-made as the ones on the Tay Bridge or they were arranged in ways that meant they couldn't hold much weight.

Apparently, he just wasn't a very good engineer. The Tay Bridge disaster made it clear that those projects were also ticking time bombs and that bridge inspections had to be way more thorough to stop these kinds of things from happening. The other bridges were either quickly repaired... or destroyed to prevent another disaster.

The Tay Bridge, of course, was not the last bridge to collapse. The Quebec Bridge was set to be the longest cantilever bridge in the world when construction began in 1900, but it collapsed twice before it eventually claimed the title in 1917. Cantilever bridges are made of big bulky parts that support relatively flat sections between them. And during construction, workers kept noticing that some of the support pieces were bent more than they should have been.

They were holding way more weight than they were designed to, partly because the bridge ended up longer than it was originally planned to be, but to save time and money, extra support wasn't added. The designers hoped that the problem could be fixed as construction went on, but at the end of the workday on August 29, 1907, parts of the bridge collapsed under their own weight, killing 75 workers. Construction resumed after investigators figured out what happened.

Their main conclusion was that the bridge's components just weren't strong enough. So the bridge was rebuilt, this time much bigger and stronger.

But, all that extra metal made it much heavier, too. And as the middle section was hoisted into place in 1916, the equipment bringing it up broke and they plunged into the river below. This time 13 people died.

The Quebec Bridge disaster was a reminder that it's important to make sure that the thing you're building can carry the weight you're putting on it, and it probably would not have happened today. Over the last century, engineers have developed tons of new mathematical techniques and computer programs that probably would have caught the first collapse long before the bridge was even built.

Even before it opened in 1940, people noticed something strange about Washington State's Tacoma Narrows bridge: it moved. Gentle winds could make the long suspension bridge flex up and down by meters at a time, making it hard to drive or even walk across. The bridge's designers tried to keep it from bouncing so much by installing shock absorbers, but they weren't very effective. And then about four months after it opened to the public, one of its cables snapped from how high the bridge was bouncing in a strong wind, and the other cables started to slide around.

This let the bridge start twisting back and forth instead of bouncing, and eventually the middle broke off and fell into the water below. Unlike the Tay and Quebec Bridge disasters, there was only one fatality this time: a dog named Tubby. Tacoma Narrows failed because of something called aeroelastic flutter.

When that first cable snapped, its side of the bridge fell slightly because it was less supported. Then other cables on that side pulled up like stretched-out rubber bands, and the bridge started twisting back and forth. Then the wind gave each twist a little boost.

Those small boosts built up over the course of the morning, and eventually twisted the bridge apart. Engineers learned their lesson pretty quickly and reinforced the Bronx Whitestone bridge in New York, which is also known to flex in the wind. And that bridge is still standing today.

But not all failed construction projects have been bridges, though. The 40-story Hyatt Regency Hotel opened in Kansas City Missouri in 1980 and it quickly became known for its innovative atrium which featured three long walkways that seemed to float in the air. The hotel called them skywalks. Instead of being supported by pillars, each skywalk was suspended from long rods hanging down from the ceiling.

The second floor skywalk hung below the fourth floor skywalk, with the third floor skywalk off to the side on separate supports. Their unique design made the huge atrium an ideal place to host events and parties, like the one that the hotel hosted on July 17, 1971. About 1,600 people attended. Hundreds danced and talked in the atrium while others milled around the hotel and checked out the famous skywalks.

Then, at 7:05 pm, two of the skywalks suddenly collapsed without warning. One hundred fourteen people were killed and more than 200 were injured, making the incident the deadliest structural collapse in US history until 9-11 happened 20 years later. Investigators found a small last-minute change to the rods holding the two collapsed skywalks that meant that instead of each holding its own weight, they were linked together. Originally, rods went from the ceiling all the way through the fourth floor skywalk and down to the second floor one, with pieces underneath each skywalk supporting its immense weight.

But after the change, the second floor skywalk hung from the fourth floor one. So supports for the fourth floor skywalk weren't just holding up one 29 metric ton skywalk, they were holding up two of them. It also didn't help that to save some money, the building materials themselves weren't quite as strong as they should have been.

And on the night of the party, the extra weight of a couple dozen people standing on the skywalks proved to be too much. What seems like a tiny change to their construction turned out to have terrible consequences.

Manhattan's Citicorp power, which is now just called 601 Lexington Avenue, proves that projects with problems don't always end in disaster. The skyscraper's triangular top makes it stand out from afar, but you notice something else if you were just walking by: Citicorp Tower was built on stilts, which turned out to be a problem.

The stilts were built to accommodate a church next door that essentially said they could build a skyscraper on the property as long as it didn't block the church, so the building designers just decided to start most of the tower nine stories up.

They knew that such a tall building with so little on its bottom floors could be prone to toppling over in strong winds, so they wanted to make sure that it could withstand even the strongest winds blowing against its windows. So they added extra weight to the top of the tower that moved in the opposite direction of any lean the wind might cause.

Wind against a skyscraper's corners usually just sort of slides past the building, so Citicorp Tower's designers didn't even bother checking for them, and that probably would have been fine if the stilts were on the corners, but because that weird thing with the church, the building stilts are in the middle of its sides.

And in 1978, about a year after the center opened, a college engineering student figured out just how unstable the building was. The calculations showed that storms strong enough to topple the tower hit Manhattan about every 16 years. Pretty alarming.

The building's designers didn't tell anyone other than the New York City Police Department and the Red Cross. who helped them secretly work out an evacuation plan while the building's internal structure was quietly but frantically improved every night after the office workers who were using the building went home.

Braces that had been bolted together during construction were  now welded steel plates, making the connections between them stronger so Citicorp Center would be sturdier in the wind. In the end, everything turned out fine. The repairs were finished after just three months, and the only major northern hurricane of the year missed New York City.

Almost no one knew about any of this until 1995 when someone finally spilled the beans on just how close New York came having a skyscraper fall over in a powerful storm.

London's Millenium Bridge was another project with a major design flaw that got fixed before it led to disaster. The bridge opened on June 10th of 2000 as a way for pedestrians to cross the river Thames. It was touted as an engineering accomplishment at the time; a sleek new bridge to mark London's entry into the 21st century.

Then it promptly closed for repairs on June 12th, 2000. And no, I did not say that wrong. It was open for a grand total of three days. The problem wasn't weight, or wind, or waves. The bridge was plenty strong enough to deal with everything it was supposed to hold; instead it was the people themselves who were unintentionally making the bridge sway back and forth like a pendulum.

Some sort of random coincidence would get a wobble started, like a bunch of people randomly stepping on the bridge on the same side at the same time. And with up to 2,000 people crossing at once, that sort of coincidence was bound to happen sometimes. Then, once the wobble got started, it was easier for people to walk with that wobble than to fight against it. So many people walking in sync made the wobble get even bigger, and soon it got so extreme that people couldn't walk across the bridge without grabbing the side railings.

The bridge probably wasn't in danger of failing, because people would stop walking when the wobbles got too big and wait for them to die down before continuing. But the wobble would have still weakened the bridge over time, and the bridge's engineers knew what happened to Tacoma Narrows; they weren't taking any chances. So they decided to close the bridge pretty much immediately before the wobbles started causing other problems.

They spent the next two years adding dampers all along the bridge, which absorb most of the force from pedestrian steps to keep the bridge from swaying. The bridge reopened in 2002, with all the dampers firmly in place, and it's been a famous tourist attraction ever since.

Designers and engineers are human, and occasionally they make a mistake that isn't caught, or there's a factor that no one anticipated. Sometimes, like with the Citicorp Tower and the Millennium Bridge, the issues are fixed in time. In other cases they can lead to catastrophe. But generations of engineers have studied and learned from these failures, and there's no telling how many potential disasters have been avoided because of those lessons.

Thanks for watching this episode of SciShow, which was brought to you by our patrons on Patreon. If you wanna help support this show you can go to patreon.com/SciShow and don't forget to go to youtube.com/SciShow and subscribe.