When Lake Conemaugh’s earthen dam gave way on the afternoon of May 31, 1889, Johnstown’s fate was sealed.

But even as those who had gathered at the dam site looked on in horror, they could not have known that a few nondescript bridges located miles downstream would make the flood even more devastating.

A new study shows that, at two points, the raging flood backed up behind bridges and then was released with even stronger force. In fact, researchers say the flood’s peak flow happened not at the dam but at a railroad bridge more than three miles downstream.

“It’s approximately equivalent to the average flow at the mouth of the Mississippi River,” said Carrie Davis Todd, an assistant professor of geology at Pitt-Johnstown.

“So basically, it was like the Mississippi River emptying into the valley.”

The basic facts of the infamous 1889 Flood – and the more than 2,200 lives the deluge claimed – are well known. The story has been immortalized in books and on film.

But some say there is relatively little comprehensive scientific research, with historical accounts of the flood relying heavily on anecdotes that are sometimes inconsistent and difficult to verify.

“Trying to find references is unbelievably difficult,” said Uldis Kaktins, a longtime UPJ geology professor who retired last year.

So Davis Todd, Kaktins and former UPJ adjunct instructor Neil Coleman – with help from former UPJ student Reed Myers – set about changing that. They’ve used tools including GPS technology, extensive field research and historic archives over the course of about a year of research.

Numerical models

It was Coleman, a Somerset County native, who first suggested the project. And Davis Todd said Coleman, while studying other dam failures, developed numerical modeling that has been valuable to the effort.

The researchers are not interested simply in the water’s velocity, but rather in its “discharge rate” – a measurement that shows both the volume and speed of the flood.

Their findings so far have been surprising: At the South Fork Dam site near present-day St. Michael, the water’s peak discharge rate when the structure failed is now estimated at 8,500 cubic meters per second.

There is no doubt that is a devastating wave: That is the equivalent of nearly 2.25 million gallons flowing by an object each second.

The water poured down the South Fork of the Little Conemaugh, picking up more debris as it flowed. But the deluge then encountered, at an oxbow on the Little Conemaugh River, the sturdy Conemaugh Viaduct.

The single-arch stone railroad bridge held for a time, and Kaktins says the flood built up behind it – possibly to a depth of more than 80 feet.

“A new lake forms, basically,” Kaktins said.

“What brought the bridge down was nothing more than the pressure of the water – the weight of the water itself.”

When the viaduct failed, it spurred what Davis Todd calls “another catastrophic release” at a discharge rate the researchers now estimate at 12,000 cubic meters per second. That’s equal to 3.17 million gallons each second.

That was 41 percent more flow than at the dam upstream, and it is equivalent to unleashing the modern-day Mississippi.

Six stories high

“When the water is released, it’s (a wave) about 65 feet high,” Kaktins said. “That’s about 30 feet higher than it was at the dam.”

To a lesser extent, the same thing happened again farther downstream, where the water and debris was held back by what was called Bridge No. 6 near Staple Bend Tunnel.

Researchers have not calculated a flow rate for when that structure failed. But Kaktins said the flood wave again reached more than 45 to 48 feet high – taller than it was at the South Fork Dam break.

The new research gives statistical backing to the observations of railroad men who were interviewed shortly after the 1889 Flood.

M. Trump, an assistant superintendent of the Pennsylvania Railroad’s Pittsburgh division, was in the area when the flood happened. In response to questioning from a railroad representative, Trump said the two bridge failures were the equivalent of two additional dams breaking.

At the larger Conemaugh Viaduct, Trump said, “The creek is 85 feet below the track, and the water dammed up at this viaduct until it finally gave way, and then it all came down in a rush.”

He added that “It was just like a dam going out, and that same action was repeated at Bridge 6” downstream.

“After (the water) was liberated from Bridge 6, it went down the valley there in an immense wave, and that was what caused the heavy waves so close to Conemaugh and Johnstown,” Trump said.

“In other words, it was just as if there were two dams like the South Fork Dam between it and Johnstown.”

Rolling wave

It was the heavily populated city and its suburbs that took the brunt of the force generated by those additional “dam” breaks. Witnesses described a rolling wave or wall of debris and water slamming into the city.

But until now, there have been few hard, scientific numbers assigned to the volume and force of that wave as it coursed through the river valley.

What started as a “little project,” Davis Todd said, has grown significantly and will continue.

Some findings are preliminary and could change.

For instance, the researchers’ calculation that Lake Conemaugh contained about 15 million cubic meters of water – which is less than some previous estimates – could change upon confirmation of the dam’s exact height on May 31, 1889.

“This is not done by any means,” Davis Todd said.

Still, the study’s results are getting some notice. Data were presented last month at a Geological Society of America conference in Portland, Ore., leading to a short article in The New York Times.

The overall goal simply is to gain a better understanding of what happened on that tragic day in Johnstown’s history. But there also may be practical uses for the researchers’ findings.

Not long after the Oregon conference, Davis Todd received a call from a dam-safety official in Michigan. The idea behind the 1889 flow data, he said, could have implications for present-day emergency planning in communities that lie downstream from dams.

“For him, that was a very interesting point – that there are downstream influences that can increase the discharge rate,” Davis Todd said.