A Saturated Valley and a Rising Lake

The payment was measured in water. It had been accumulating for months — not in the ledgers of the South Fork Fishing and Hunting Club, not in the correspondence files of Cambria Iron, not in the reports that engineers had written and managers had filed, but in the ground itself. The hills above the Little Conemaugh had been absorbing moisture since the previous autumn. The soil had taken all it could hold. The water that fell now would not soak in. It would move. It would move downhill, into the creeks, into the runs, into the river, and into the lake behind the dam.

The debt was coming due in inches and feet, measured against a crest that had been lowered, against a spillway that had been narrowed, against an embankment that had been patched. The instruments of payment were already in place. They had been in place since the club bought the reservoir and began altering it.

On May 28, 1889, a low-pressure area formed over Nebraska and Kansas. By the time the pattern reached western Pennsylvania two days later, it had become what the U.S. Army Signal Corps would call the heaviest rainfall event ever recorded in that part of the United States, estimating 6 to 10 inches (150 to 250 mm) in 24 hours. The storm was unusual in its intensity and its duration. It was not, however, the sole cause of what followed. The rain landed on ground already saturated, on a reservoir already high, on a dam whose capacity to withstand a major storm had been reduced by years of modifications. The catastrophe of May 31 was the final stage of a physical process that had been under way for months.

The spring of 1889 had been wet across the Allegheny watershed. Rain fell in April. It fell again in early May. The ground absorbed what it could. Then it absorbed no more. The soil in the hills above the Little Conemaugh reached field capacity. Water that struck the ground no longer infiltrated the subsoil. It ran off. Every creek that fed the Little Conemaugh carried more than its usual volume. The tributaries that fed Lake Conemaugh — the South Fork of the Little Conemaugh, Old Beaver Dam Creek, and several unnamed runs — carried the accumulated runoff of a saturated valley directly into the reservoir behind the dam.

The lake had been rising for weeks before the storm arrived.

Downstream, in Johnstown, the spring routines did not change. The Cambria Iron Works ran its furnaces and its rolling mills on schedule. Shift changes moved thousands of men through the gates at the same hours they had kept for years. The Pennsylvania Railroad ran its passenger and freight service along the valley floor. The trains passed through the stone arches of the bridge that would later hold the flood’s debris. Merchants on Main Street stocked for the Memorial Day holiday. The Stony Creek and the Little Conemaugh ran high. Residents noted the water in the channels. The rivers had run high before. The channels held. The banks held. The town went about its business.

The valley below the dam treated the high water as seasonal. The dam above the valley was treated as permanent.

At the South Fork Fishing and Hunting Club, the season was opening. Members from Pittsburgh planned their weekend trips to the lake. The club’s resident manager, John Parke, maintained the grounds, the cottages, the boathouses, and the dam. He had been employed to watch the property. He watched the lake.

Parke had been with the club since 1887. He knew the dam’s history. He knew the crest had been lowered. He knew the spillway had been screened and partially blocked. He knew the club had patched the embankment where the old culvert had been. He also knew the dam had held. In his experience, the water rose in the spring and fell in the summer. The spillway carried the excess. The lake stayed within its banks. The structure held.

The club’s account of conditions at the dam in the spring of 1889 held to this position. The dam was sound. The spillway was adequate. The water had never come close to the crest. The resident manager monitored the level. The structure performed as it had performed for decades.

The later engineering investigation told a different story.

The American Society of Civil Engineers committee that examined the dam after the flood found that the spillway capacity had been reduced below what the original design intended. The Commonwealth’s dam, built between 1838 and 1853, had included a spillway cut through rock at the eastern abutment. The original spillway was designed to carry excess water when the reservoir rose above a controlled level. The club had narrowed this spillway. It had installed screens to keep fish from escaping. It had allowed debris to accumulate. The screens and the debris reduced the effective opening. Water that should have passed through the spillway instead backed up into the lake. The lake rose higher than the original design anticipated.

The ASCE committee also found that the club had lowered the crest by several feet during its modifications. The original dam stood higher. The lowered crest meant less freeboard — less distance between the normal water level and the top of the embankment. Less freeboard meant less margin. Less margin meant less time between the water rising and the water overtopping the dam.

The club’s defenders disputed parts of this account. They argued the spillway had been adequate for normal conditions. They argued the dam had held for years without incident. They argued the storm was unprecedented. They argued no structure could have withstood the volume of water that fell on May 30 and 31.

The ASCE committee disagreed. Its 1891 report calculated the spillway’s effective capacity and found it insufficient. The committee found that even a properly maintained spillway of the club’s dimensions would have been overwhelmed by the storm. But the committee also found that a dam built to the original specifications — with the original crest height and an unobstructed spillway — would have had enough freeboard to hold the flood. The water would have risen to the crest. It would not have overtopped. The dam would have stood.

The difference between these two accounts was the difference between a dam that could hold the storm and a dam that could not. That difference was measured in feet of earth removed from the crest and cubic feet of spillway capacity lost to screens and debris.

The club had made both changes. The changes were documented. The consequences were calculable.

The disagreement between the club’s defenders and the ASCE committee turned on a question of design intent. The original engineers who built the dam for the Commonwealth in the 1840s and 1850s had specified a crest height and a spillway capacity for a reason. They were building a reservoir for a canal system. They were not building a recreational lake. They were not building a fish pond. They were building infrastructure. Infrastructure is designed for the worst case, not the average case. The original crest height provided freeboard for storms that exceeded the historical record. The original spillway provided outlet capacity for floods that exceeded the normal runoff. The engineers who built the dam understood that the structure would be tested. They designed it to be tested.

The club’s modifications removed that design margin. The club lowered the crest to widen the road. It installed fish screens to keep its stocked bass from escaping. It allowed debris to accumulate because no one was paid to clear it regularly. Each change was small. Each change was made for a practical reason — a wider road, a better fishery, a lower maintenance cost. Each change reduced the structure’s capacity to perform under the conditions for which it had been built.

The club treated the dam as a feature of its property. The original engineers treated the dam as a hydraulic structure. The difference mattered. A feature of property is maintained for appearance and convenience. A hydraulic structure is maintained for capacity and safety. The club maintained the dam the way it maintained the cottages and the boathouses. It kept them in working order for the season. It did not keep them in working order for a hundred-year storm.

By late May, the lake was already high. Spring runoff had filled it above its normal seasonal level. Parke, the resident manager, noted the rise. The water sat within the range he had seen before — high, but not unprecedented. The spillway was carrying water. The screens were partially clogged with debris, but water was passing through. The dam was holding.

The lake continued to rise.

On May 30, the storm reached western Pennsylvania. The low-pressure system that had formed over Nebraska and Kansas arrived in the Allegheny Mountains. Rain fell. It fell steadily through the day. It fell heavily through the night. The saturated ground could not absorb it. The creeks that fed the lake carried it directly into the reservoir. The lake rose.

Rain also fell on the watershed below the dam. The Little Conemaugh and the Stony Creek rose. The channels in Johnstown that carried the rivers through the town began to fill. Water reached the tops of the banks. In some places, it overflowed. Low-lying streets near the rivers saw standing water. Basements flooded. The fire department reported water in the streets near the canal basin. The Pennsylvania Railroad reported water on the tracks in the valley.

None of this was unprecedented. Johnstown had flooded before. The rivers had overflowed before. The town had been built on a floodplain at the confluence of two rivers in a valley hemmed by steep hills. Flooding was a known condition. Residents moved goods to upper floors. They waited for the water to recede. It always receded.

Upstream, at the dam, the water was not receding. It was rising.

The lake’s surface area was approximately 450 acres. The watershed feeding it was larger. The dam impounded a reservoir that held, at its normal level, roughly 20 million tons of water. As the lake rose, the volume increased. Each inch of rise represented thousands of additional tons pressing against the embankment. The dam held the weight. It had held the weight for years. But the weight was increasing, and the dam’s structure had been altered.

The embankment was earth — puddled clay and rock fill, built in the 1840s by the Commonwealth’s contractors. The club had repaired sections. The repairs used earth and rock, placed without engineering supervision. The patched sections were not compacted to the original specifications. They were not as dense. They were not as strong. The ASCE committee later examined the breach and found that the dam had failed where the structure was weakest — where modifications had compromised the original construction.

The club’s repairs were not malicious. They were not designed to fail. They were designed to be cheap. They were designed to get the dam back into service after the old culvert collapsed, using material that was available and labor that was unskilled. The club did not hire an engineer to supervise the repairs. It did not test the compaction. It did not verify the strength. It filled the hole. It smoothed the surface. It moved on.

The dam held. It held because the water had never come close to the crest. The structure had never been tested. The repairs were adequate for the conditions the dam normally faced. They were not adequate for the conditions the dam was built to withstand.

On the night of May 30, the rain continued. The lake continued to rise. The spillway carried water, but it could not carry enough. The screens and debris reduced its capacity. Water that should have exited the reservoir stayed in it. The lake rose faster than the spillway could drain it.

The freeboard narrowed.

The dam’s lowered crest sat lower than the original design. The lake’s surface climbed toward that lowered crest. The distance between the water and the top of the dam shrank. What had been feet of margin became inches. What had been inches became a film.

Parke watched the water through the night. He later reported that he observed the lake rising. He checked the dam. He checked the spillway. The structure was holding. The water was not over the top. But it was close. He knew it was close.

The club’s members were in Pittsburgh. The lake was their property. The dam was their responsibility. The water rising against it was their liability. But they were not there. They were in their offices, their homes, their meeting rooms. The absentee ledger did not carry an entry for the cost of a dam failure. It carried entries for dues, for cottage maintenance, for fish stocking, for groundskeeping. It did not carry an entry for freeboard. It did not carry an entry for spillway capacity. It did not carry an entry for the risk downstream.

The ledger was balanced. The lake was not.

Down in the valley, the night passed. The rain fell. The rivers rose. In Johnstown, residents went to bed with water in the streets. The town had seen high water before. The Cambria Iron Works ran its night shift. The railroad ran its trains, though some routes were delayed by water on the tracks. The stone bridge carried traffic across the Little Conemaugh. The streets were quiet. The rain was loud.

At the dam, the lake rose through the night. The spillway worked. It was not enough. The water climbed past the level Parke had seen in previous springs. It climbed past the level the spillway could manage. It climbed toward the crest.

The dam had been built with a margin. The club had consumed that margin. It had lowered the crest. It had narrowed the spillway. It had patched the embankment with inferior material. Each decision had been small. Each decision had been made for a reason — a wider road, a fish screen, a quick repair. Each decision had reduced the structure’s capacity to hold what it was designed to hold.

The rain fell on all of it at once.

The storm was large. The ASCE committee acknowledged its magnitude. The rainfall on May 30 and 31 was the heaviest recorded in the region. But the committee’s finding was specific: a dam built to the original specifications would have held. The storm was unprecedented, but the dam had been designed for unprecedented storms. That was the purpose of freeboard. That was the purpose of a spillway. The original engineers had built both. The club had reduced both.

The water rose. The freeboard disappeared. The dam’s structure absorbed the pressure. The patched sections held. They were not designed for this. They were not engineered for this. They were repaired with fill and rock and the assumption that the water would never come.

The water came.

The storm that reached the Allegheny Mountains on May 30 was not a local event. It tracked across several states. It carried moisture from the Gulf of Mexico and the Atlantic. It met cooler air over the mountains. The air was forced upward. It cooled. It released its moisture. The process was ordinary. The volume was not. The rain fell at rates that exceeded anything the region’s instruments had recorded. The ground could not absorb it. The creeks could not carry it. The river could not pass it. The lake could not drain it. Every component of the watershed was full. Every component was doing what water does when it has nowhere to go. It was rising.

The valley below the dam was a confluence. The Little Conemaugh ran from the east. The Stony Creek ran from the south. They met at Johnstown. The combined flow moved west toward the Conemaugh and the Allegheny. The town sat at the junction. The junction was a natural collection point for water from three directions. The hills that hemmed the valley compressed the flow. The narrow channel at the stone bridge compressed it further. The town was built on the one piece of flat ground in a valley of steep slopes. That flat ground was a floodplain. It had been a floodplain before the town was built. It remained a floodplain after the town was built. The streets and the mills and the houses sat on ground that water had claimed before and would claim again.

The dam above this floodplain held a lake. The lake held water from a watershed that drained into the Little Conemaugh. The dam was the barrier between the lake and the river. The barrier was earth. The earth had been altered. The alterations had reduced the barrier’s capacity. The storm was testing the altered barrier against a load the original barrier had been designed to withstand.

The test was under way.

On the morning of May 31, Elias Unger, president of the South Fork Fishing and Hunting Club, awoke in his farmhouse on the hill above the dam. The rain was still falling. He looked out at the lake. The water was nearly at the crest. Unger went outside. He walked to the dam. He saw the water. He saw the spillway. He saw the screens clogged with debris. He saw the lake.

He acted. He summoned men to clear the spillway. He sent messengers downstream. The messengers rode toward South Fork and toward Johnstown. They carried warnings. The dam was in danger. The water was at the top. The town below should prepare.

The messengers rode. The water rose. The dam held. It held against a load it had not been designed to carry, on a structure that had been altered from its design, with a spillway that could not pass the volume, and a freeboard that no longer existed.

In Johnstown, the Friday shift at Cambria Iron began. Workers walked through water in the streets. The rivers were over their banks. The rain was still falling. A wet morning — not, by the valley’s standards, an extraordinary one.

The messengers had not yet arrived.

The lake was at the dam’s crest. The spillway was clogged. The screens held debris. The water had nowhere to go. It rose against the lowered embankment. It pressed against the patched earth. It found the weak points. It found the places where the original construction had been cut and filled and repaired without engineering. It found the places where the club had saved money. It found the places where the ledger had recorded profits and not costs.

The water found the dam’s limits.

The messengers rode toward a valley that had minutes. The dam held. The water climbed over the crest. It seeped through the embankment. It weakened the structure from the inside. The earth softened. The fill shifted. The patched sections began to move.

The rain fell. The lake was full. The dam was at its limit. The messengers rode. The valley below went about its morning. The water pressed against the earth. The earth held. Then it did not.