The Last Ordinary Weeks

The clubhouse shutters swung open on their new paint. A caretaker swept the porch. A delivery wagon climbed the road from the South Fork station, carrying provisions for the opening weekend. The date was May 25, 1889. The lake lay flat and full behind the dam. The season had begun.

Six days before the rain started, the South Fork Fishing and Hunting Club was conducting its annual business. The ledger entries ran to coal for the boilers, paint for the cottages, fishing line for the tackle boxes. The amounts were modest. The handwriting was steady. The dam appeared nowhere in the column for repairs. It appeared nowhere because it was not being repaired. It was being maintained. The distinction was the kind of distinction that fills a page in a ledger without changing anything in the ground.

Three days before the rain began, the valley below the dam carried on its work. The date was May 27, 1889. In Johnstown, the Cambria Iron works operated its furnaces and rail mills on continuous shifts. Crews changed at six and six. The whistle marked the hours. Smoke from the open-hearth furnaces drifted over the valley. Along Washington Street, shopkeepers unlocked their doors. Children walked to school. The streetcars ran on their schedule. The Stonycreek River ran high with spring rain, as it did every year. No one in the valley had reason to look eighteen miles up the Little Conemaugh, toward a reservoir held back by an earth dam that had been lowered, patched, and stripped of its discharge pipes.

The club treated the dam as a settled fact. The structure held the lake. The lake held the fish. The fish held the members’ interest. This was the order of things. The club’s minutes from the spring of 1889 recorded seasonal preparations: grounds cleanup, cottage openings, provisions ordered, boats launched. The dam appeared in the maintenance log the way a fence might appear in the log of a farm. It was checked. It was patched where patches were needed. It was not assessed for its capacity to withstand a storm of record. The distinction mattered. Checking a dam for leaks is not the same as evaluating its hydraulic sufficiency. Patching a slump is not the same as restoring spillway capacity. The club performed the first and called it the second.

The dam’s physical condition in the spring of 1889 was the product of choices made over the preceding decade. The crest had been cut down by as much as three feet to widen the carriage road. The five discharge pipes that the Commonwealth had installed through the base of the embankment had been removed and never replaced.

The spillway — a channel cut through rock and earth at the dam’s eastern end — had been narrowed by the construction of a viaduct for the Pennsylvania Railroad, and its screens, designed to keep fish from escaping, reduced its effective capacity further. The breast of the dam, where water pressed against it, had been patched with earth and rock in spots where slumping had occurred. The patches held against ordinary water levels. They had never been tested against a flood. The club’s alterations had also eliminated an emergency spillway on the western abutment, a feature whose existence was later documented by topographic data from 1889.

A hydraulic analysis published in 2016 confirmed what the engineers who inspected the dam in the 1880s had observed: the changes made to the dam by the South Fork Fishing and Hunting Club severely reduced its ability to withstand major storms. Lowering the dam by as much as three feet and failing to replace the discharge pipes stripped the structure of the safety margin its original designers had built into it. The spillway alone could not compensate. The analysis ran the numbers. The numbers said the dam, in its altered state, could not pass the volume of water that a large storm would deliver. The club did not have this analysis. The club had something else: the testimony of years. The dam had held since 1881. The water had never reached the crest. The patches had not failed. This was the evidence the club trusted — the evidence of seasons passed without incident.

The problem with survival as evidence is that it proves nothing about the next storm. It proves only that the storms so far were not large enough. The dam’s survival from 1881 through the spring of 1889 was real. It was also provisional. The dam had survived because the rainfall in those years had not exceeded the spillway’s reduced capacity. The margin between what the dam could pass and what a storm could deliver had been narrowed by the club’s own alterations. The margin was real. It was invisible. No gauge on the dam measured it. No alarm on the crest signaled when it had been crossed. The dam’s safety depended on the difference between two numbers — the spillway’s capacity and the inflow from a storm — and only one of those numbers was known to the people who owned the dam. The other number belonged to the weather.

In Johnstown, the spring of 1889 brought the usual concerns. The Stonycreek and the Little Conemaugh ran together at the point where the valley narrowed. The rivers rose with March snowmelt and April rain. They rose again in May. Families who lived near the banks watched the water. Some moved belongings to upper floors. This was routine. The rivers had flooded before — in 1887, in 1888, in years before that. The floods were nuisance floods. They wetted cellars. They covered streets in low-lying sections. They did not kill. They did not destroy neighborhoods. They were the price of living in a narrow valley, and the valley’s residents had been paying that price for decades without catastrophe.

The Cambria Iron works occupied the valley floor between the rivers. The works employed thousands. The furnaces ran day and night. The rail mills turned out steel rails for the railroads that crossed the continent. The company had built its own hospital, its own stores, its own housing. Cambria Iron was Johnstown.

The company’s executives watched the rivers too, but their concern was operational. Floodwater in the lower yards meant delayed shipments. Floodwater in the furnaces meant a longer shutdown and a costly restart. The company had dealt with these interruptions before. The interruptions were part of the cost of doing business in the valley.

No one at Cambria Iron connected the spring floods to the dam eighteen miles upstream. The dam was the club’s property. The dam was the club’s problem. The valley’s relation to the dam was the relation of a downstream neighbor to an upstream landowner — a relation defined by property lines and legal boundaries, not by shared risk.

That relation was the relation the law recognized. The dam sat on private land. The reservoir covered private water. The club owned both.

The law of nuisance and the law of trespass gave downstream owners a right to sue if the dam caused harm, but the right was retrospective. It applied after the harm was done. No law required the club to maintain the dam to any standard. No law required the club to submit to inspection. No law required the club to warn anyone downstream.

The warnings that had been generated — by engineers, by Cambria Iron’s own correspondence, by the visible condition of the dam itself — existed in the space between knowledge and authority. The people who knew the dam was compromised did not have the authority to repair it. The people who had the authority to repair it did not believe it needed repair.

This was the gap. The gap was not a failure of law. It was a feature of the law. The law distributed risk along property lines, and the property lines put the risk downstream.

The club’s members, most of them, lived in Pittsburgh. They were manufacturers, bankers, railroad executives, men whose names appeared in the social registers and the financial columns. They came to South Fork for weekends and holidays. They fished. They ate. They walked the grounds. They did not walk the dam. The dam was maintained by a caretaker, John Parke, a civil engineer by training who served as the club’s resident manager. Parke kept the buildings in repair, the grounds in order, and the lake stocked. He also watched the dam. He had reported concerns about the dam’s condition in previous seasons. He had noted the slumping. He had supervised the patches. His reports went to the club’s secretary in Pittsburgh. The secretary filed them. The pattern was established. The caretaker observed. The secretary recorded. The members fished. The dam held.

This was the absentee ledger: a private account of profits and pleasures, maintained in a city fifty miles from the risk. The ledger recorded the costs of upkeep, the dues of members, the stocking of fish, the wages of servants. It did not record the cost of the dam’s reduced capacity. It did not record the risk that the lowered crest and the missing pipes and the narrowed spillway had created for the valley below. That cost did not appear in any column. It was not priced. It was not acknowledged. It was carried — invisibly — by the people who lived downstream, who had not consented to it, who did not know its dimensions, and who had no means to demand its payment before the bill came due.

The spring of 1889 was wetter than most. Rain fell across western Pennsylvania in the second week of May. The rivers rose. They receded. They rose again in the third week. The pattern was familiar. Spring in the Alleghenies meant rain. Rain meant high water. High water meant inconvenience. The valley absorbed the inconvenience and continued its work. The Cambria Iron works kept its furnaces running. The Pennsylvania Railroad kept its trains moving. The streetcars ran. The schools held classes. The shops sold goods. The newspapers reported local events — a wedding, a funeral, a council meeting, a dispute over a street paving contract. None of these reports mentioned the dam. The dam was not news. The dam was a reservoir on a private lake above a fishing club. It had been there for years. It would be there tomorrow.

The club’s season opened in mid-May. Members arrived from Pittsburgh on the Pennsylvania Railroad. The train stopped at South Fork, where a road climbed to the clubhouse. The cottages were aired and stocked. The lake was ready. The fish were biting. The dam held the lake, and the lake held the fish, and the fish held the members’ attention.

The dam was infrastructure in the way a retaining wall is infrastructure: present, functional, unconsidered. The club treated it as a finished object. It was not a finished object. It was a structure in a constant state of alteration — altered by the club’s own work, altered by weather, altered by time.

The earth of the embankment settled. The patches eroded. The spillway screens clogged with debris. The lake level rose and fell with the seasons. The dam was not static. It was dynamic.

The club’s maintenance treated it as static. This was the error. Not a single decision, but a habit of thought: the dam was fine, the dam had always been fine, the dam would continue to be fine.

The habit of thought had a name. It was normalization. The dam’s weaknesses had been identified, reported, and discussed for years. Each year that the dam held, the weaknesses became less alarming. Each season without failure made the next season without failure more likely in the minds of the people who owned and operated the structure. The warnings did not disappear. They were absorbed. They became background. They became the kind of knowledge that people possess and do not act on, not because they have forgotten it, but because it has been overtaken by the evidence of continued function. The dam held. The dam held. The dam held. Three years of holding outweighed a report that said the spillway could pass perhaps half the flow of a probable maximum flood. The report was filed. The dam held. The filing was sufficient. The holding was proof.

The spring rains of 1889 were not extraordinary. They were not the storm that would break the dam. Not yet. The rains that fell in the second and third weeks of May were the kind of rains that fell every spring. They raised the rivers. They filled the low places. They tested the drainage. They did not test the dam. The dam’s spillway passed the water easily. The lake rose a few inches. The lake fell. The dam held. This was the last confirmation. The dam held through the spring rains of 1889, and its holding was taken as evidence of its soundness, when in fact it was evidence only of the rains’ modesty. The dam had not been tested. The dam had been spared. The distinction was the distinction between survival and sufficiency, and no one at the club made it.

The last week of May arrived. The club’s ledger recorded the usual entries. The Cambria Iron works recorded its usual output. The rivers ran at their usual spring levels. The weather was overcast. The barometer was falling. No one in the valley read the barometer with the dam in mind. No one at the club read the barometer with the dam in mind. The barometer was an instrument for sailors and farmers. It was not an instrument for fishermen.

The low pressure system that moved into western Pennsylvania on May 29 was a weather event. It was not, in the language of the valley or the club, a dam event. The dam was not in the weather’s vocabulary. The weather did not know about the dam. The weather did not care about the dam. The weather did what weather does: it dropped its moisture on the Alleghenies, and the Alleghenies shed the water into the streams, and the streams fed the rivers, and the rivers fed the Little Conemaugh, and the Little Conemaugh fed the lake above the dam.

The lake was Lake Conemaugh. It stretched across the valley behind the dam, covering a surface area of roughly 450 acres at normal pool. It held about 20 million tons of water. The water came from the watershed above — the streams that drained the hills of Cambria County.

The watershed was not large. It covered perhaps 48 square miles. But the watershed was steep. The hillsides were logged. The ground was saturated. When rain fell on the watershed, it ran off fast. It collected in the streams. It concentrated in the Little Conemaugh. It poured into the lake.

The lake’s level rose. The water pressed against the dam’s breast. The dam’s crest, lowered by three feet, offered less freeboard than the original design. The spillway, narrowed and screened, offered less capacity. The discharge pipes, removed and never replaced, offered no relief. The water had one path through the dam: over the spillway.

If the spillway could not pass the inflow, the water would rise to the crest. If the water reached the crest, it would flow over the top. If water flowed over the top of an earth dam, the earth would erode. If the earth eroded, the dam would fail.

This sequence was not speculative. It was mechanical. It was the way earth dams fail. It had been understood by engineers for decades.

The club’s members were not engineers. The club’s caretaker was an engineer, but his reports had been filed, and the filing had been treated as action, and action had been deferred to routine, and routine had been deferred to the next season, and the next season had arrived.

The last week of May 1889 was the last interval in which the flood could have been a near miss.

If the dam had been repaired — if the crest had been raised, if the discharge pipes had been replaced, if the spillway had been widened and its screens removed — the storm of May 30 and 31 would have tested the dam and the dam would have held. The lake would have risen. The spillway would have passed the water. The rivers below would have flooded, as they always did. Cellars would have been wetted. Streets would have been covered. The inconvenience would have been absorbed. The valley would have gone on. The dam would have held.

This was the near miss that did not happen. It did not happen because the repairs were not made. The repairs were not made because the authority to make them belonged to a club that treated the dam as scenery. The club treated the dam as scenery because the dam had held. The dam had held because the storms had been modest. The storms had been modest because the weather, until May 30, had cooperated. The weather was about to stop cooperating.

The low pressure system that entered the Ohio Valley on May 29 was not unprecedented. It was large. It was slow-moving. It carried moisture from the Gulf of Mexico and the Atlantic. It stalled over western Pennsylvania.

The rain began on the evening of May 30. It fell steadily through the night. It intensified on the morning of May 31. The rainfall was heavy across the region.

Estimates of the total varied, but the consensus among observers and later investigators placed it at six to ten inches in the watershed above the dam, falling in roughly twenty-four hours — what would be termed the heaviest rainfall event ever recorded in that part of the United States. This was not a probable maximum flood. It was not a storm of biblical proportion.

It was a large storm — the kind of storm that occurs in the Alleghenies every few decades. The dam had been built to withstand such storms. The dam, in its original configuration, would have held. The dam, in its altered configuration, could not.

The rain fell on May 30. The ground was already saturated from the spring rains. The streams rose. The Little Conemaugh rose. The water poured into Lake Conemaugh. The lake level began to climb.

The dam’s spillway passed what it could. The screens clogged with debris from the swollen waterline. The water rose higher.

The caretaker, John Parke, saw the water climbing toward the crest. He sent men to clear the screens and unclog an iron grate and broken fish trap obstructing the spillway. They worked in the rain. The water kept rising.

The dam’s lowered crest offered less freeboard than the original design. The margin between the lake’s surface and the top of the dam was narrowing. The margin that the club had not measured, had not maintained, had not restored — that margin was being consumed by the rain, inch by inch, hour by hour. The deferred-maintenance debt was coming due. The debt was measured in feet of freeboard and cubic feet of spillway capacity. The payment was measured in water.

The rain was falling on May 30. The lake was rising against the dam’s altered structure. The ordinary weeks had run out.