The Warning Outlives the Witnesses

The warning was in the gap. The gap between the dam and the city was fourteen miles. The gap between the dam as built and the dam as maintained was four feet of crest, five discharge pipes, a spillway narrowed with fish screens. The gap between the men who knew those facts and the institutions that needed them was closing. One by one, the witnesses were dying.

John G. Parke had been the club’s engineer. He walked the crest on the morning of May 31, 1889.

He saw the water rise. He watched the crew of laborers try to cut a ditch through the embankment’s face to give the spillway relief. Twice, under orders from club president Elias Unger, Parke rode on horseback to a telegraph office in the nearby town of South Fork to send warnings to Johnstown explaining the dangerous situation unfolding at the dam. He came back. The water kept rising. The crew abandoned their efforts at around 1:30 p.m., fearing that their work was futile and recognizing that the dam was at risk of imminent collapse.

Unger ordered all of his men to fall back to high ground on both sides of the dam where they could do nothing but watch and wait. Parke was there. Between 2:50 and 2:55 p.m., he saw the South Fork Dam breach. He saw the lake go.

Parke died in 1911. His obituary in the Johnstown newspapers was brief. It identified him as a civil engineer and former employee of the South Fork Fishing and Hunting Club. It mentioned the flood. It did not describe what he had seen at the dam face in the final hours. That testimony had been given in 1889, at the coroner’s inquest, in the statements collected by the Pennsylvania Railroad, in the interviews published by the Johnstown and Pittsburgh papers. It existed in print. It existed in the files. The man who had spoken it was gone.

The same year Parke died, the South Fork Fishing and Hunting Club formally dissolved. The legal entity that had owned the dam, that had hired Parke, that had lowered the crest and removed the discharge pipes, ceased to exist. The dissolution papers listed no assets of consequence. The clubhouse on the lake shore had been sold. The cottages were private residences. The lake bed was a marsh. The dam sat in the woods with its breach open, its abutments exposed, its spillway choked with debris and saplings. No sign marked it. No fence enclosed it. The corporation that had created the risk was gone. The risk remained.

That same year, an engineer could open a bound volume of the Transactions of the American Society of Civil Engineers and turn to the report of the committee appointed on June 5, 1889, to investigate the causes of the Johnstown disaster. The committee had been led by James B. Francis, a hydraulic engineer known for his work on the Lowell waterpower system. The other members were William E. Morris, James D.

Francis, and Alphonse Fteley. They had examined the dam’s remains, reviewed the club’s modifications, calculated the spillway capacity, and estimated the volume of water released. Their report was a forensic dissection of structural failure. It described the lowering of the crest. It described the removal of the discharge pipes. It described the spillway’s inadequacy. It described the rock and timber debris clogging the spillway. It concluded that the dam’s failure was caused by the club’s modifications and by the lack of maintenance.

The engineer in 1911 did not need Parke’s living testimony. He had the report. He had the measurements. He had the calculations. The warning that had been spoken in 1889 by men who stood on the dam was now preserved in typeset pages, in bound volumes, in the permanent record of the engineering profession. The witnesses were dying. The file was alive.

The contrast was sharp. On one side, time and forgetting. The men who had built the dam, owned it, inspected it, repaired it, and watched it fail were passing from the scene. The physical site was softening. The breach, raw in 1889, was filling with vegetation. The abutments were weathered. The spillway channel was obscured. A visitor to the site in 1911 would have found it difficult to read the structure’s original form in the tangle of growth and erosion. The landscape was reclaiming the evidence.

On the other side, institutional memory. The ASCE report had been published in the society’s Transactions. Engineering libraries held it, university collections held it, the offices of state engineers and consulting firms held it. Its data did not depend on the survival of any individual. Its conclusions did not require the passion of a survivor. It functioned as a technical document, cited and recited in the professional literature of dam engineering.

The report’s central findings were quantitative. The original dam, as built by the Commonwealth between 1838 and 1853, had a crest high enough to contain the reservoir’s design capacity. The club had lowered the crest by approximately four feet to widen the carriage road. The original dam had five cast-iron discharge pipes running through the base, controlled by valves, allowing the reservoir level to be managed. The club had removed those pipes and sold them for scrap. The original spillway was a channel cut through the rock of the hillside at the eastern abutment. The club had allowed it to fill with debris, timber, and a screen of iron rods designed to keep fish from escaping.

The combined effect of these modifications was to reduce the dam’s capacity to pass floodwaters. The spillway, clogged and undersized, could not carry the inflow from a storm of the magnitude that struck on May 30 and 31, 1889. The water rose above the lowered crest. The earth embankment, saturated and unarmored, washed away.

These were not opinions. They were measurements and calculations, verified by the committee’s own survey of the site. Memory fades. These did not. They sat on the page.

The decade that followed the turn of the century saw the gradual extinction of the flood’s principal actors. Daniel J. Morrell, the Cambria Iron manager who had sent engineer John Fulton to inspect the dam in 1880 and whose warnings had been ignored, had died in 1885, four years before the flood. Fulton himself lived until 1903. He had examined the dam, found the discharge pipes removed, found the spillway impaired, found the crest lowered, and reported to the club that the structure was dangerous. His report survived in the Cambria Iron correspondence. It was a warning in writing. The club had ignored it. It was now part of the record.

Fulton’s death removed the last Cambria Iron engineer who had personally inspected the dam before the flood. His report, however, was preserved. It existed in the company’s files. It existed in the ASCE committee’s citation of it. It existed in the testimony given before the coroner’s inquest. The warning he had delivered in 1880 to the club’s officers — that the dam was dangerous and required repair — outlived him by decades.

The club’s members, the Pittsburgh industrialists and financiers who had owned the dam and treated it as private scenery, also passed. Benjamin F. Jones, the iron manufacturer, died in 1904. Henry Clay Frick died in 1919. Andrew Carnegie, who had been listed in some accounts as a member though his involvement was disputed, died in 1919. The men who had paid for the dam’s purchase, who had authorized the modifications, who had enjoyed the lake for fishing and sailing, who had received Fulton’s warning and declined to act, were gone.

Their private correspondence, where it survived, did not contain detailed discussions of the dam’s maintenance. The Absentee Ledger — the record of dues and assessments, of fish stock and clubhouse improvements kept by proprietors who never lived within sight of the embankment — was silent on the dam’s condition. The ledger recorded the cost of fish. It did not record the cost of the discharge pipes’ removal. It did not record the cost of the crest’s lowering. It did not record the cost of the spillway’s obstruction. Those costs had been passed to the public. The ledger stayed balanced. The valley paid the difference.

The legal record, too, was closed. The lawsuits filed against the club in the years after the flood had failed. The courts had held that the dam was an improvement to a natural watercourse, that the club was not liable for the acts of God, and that the corporate form shielded the individual members from personal liability. The Pennsylvania Supreme Court had affirmed these holdings. The United States Supreme Court had declined to review them. The law had spoken. It had said that the catastrophe was, in legal terms, no one’s fault. The dissolution of the club in 1911 was a final administrative act. There was nothing left to sue. There was no one left to charge.

But the technical record was open. And it was expanding.

In the years after 1900, the engineering profession in the United States confronted a new generation of dam projects. The Reclamation Act of 1902 authorized federal construction of dams in the western states. The Roosevelt Dam in Arizona was completed in 1911. The Arrowrock Dam in Idaho, begun in 1912, would stand 350 feet high. The Hetch Hetchy project in California, approved in 1913, involved the construction of a dam in a national park. In the East, industrial dams multiplied. Tailings ponds, water-supply reservoirs, and mill ponds dotted the landscape of Pennsylvania, West Virginia, and Ohio. Each of these structures presented the same question that the South Fork Dam had presented: could a private owner be trusted to maintain a structure whose failure would kill people downstream?

The ASCE report on Johnstown was not a statute. It did not regulate anything. It did not compel any inspection. It was a document in a professional journal. But it functioned as a precedent. When engineers debated the design of a new dam, they cited the South Fork failure. When they argued for spillway capacity, they cited the South Fork failure. When they discussed the need for discharge valves, they cited the South Fork failure. The report’s data — its measurements of the dam’s geometry, its calculations of the reservoir’s volume, its analysis of the spillway’s capacity — became canonical numbers. They appeared in engineering textbooks. They appeared in the lecture notes of professors at Lehigh, Rensselaer, and the Massachusetts Institute of Technology. They appeared in the specifications drawn up by consulting firms for new dam projects.

The warning had migrated from living memory into institutional custody. It no longer depended on Parke’s recollection of what he saw on the morning of May 31. It no longer depended on Fulton’s memory of what he found when he walked the crest in 1880. It no longer depended on Unger’s account of the desperate work at the spillway, the recognition of futility, the order to retreat. It lived in the measurements. The measurements did not die.

The migration was not automatic. It required the active intervention of engineers who understood the report’s value and who chose to cite it, teach it, and apply it. These men were not survivors. They were not litigants. They were not politicians. They were professionals who saw in the Johnstown disaster a case study in what happened when private ownership met public risk and no inspection stood between them.

One of those engineers was James B. Francis himself. Francis had served as the chairman of the ASCE investigating committee. He was a man of exacting method. His work on the Lowell waterpower system had established the formula for flow over weirs that engineers still used. His name was attached to the Francis turbine, a waterwheel design of extraordinary efficiency. When the ASCE appointed him to lead the Johnstown investigation on June 5, 1889, it chose a man whose professional reputation lent authority to whatever findings the committee produced.

Francis and his colleagues visited the site. They surveyed the remains. They measured the breach. They calculated the volume of water that had been impounded. They estimated the flow rate at the dam face and at the point where the flood struck Johnstown. They examined the spillway and found it obstructed. They examined the embankment and found it unarmored on the upstream face, contrary to the original design. They examined the abutments and found them of inadequate width. They reviewed the club’s modifications and found them to have reduced the dam’s safety factor below acceptable limits.

The committee’s report was not a brief for the prosecution. It did not assign blame. It did not name the club’s members. It did not argue that the flood was an act of God or an act of man. It presented the physical evidence, described the modifications, calculated the consequences, and stated the engineering conclusion: the dam failed because it was altered from its original design in ways that reduced its capacity to withstand the storm of May 30 and 31, 1889.

This restraint was the report’s strength. It was not a polemic. It was a technical analysis. It could be cited without controversy. It could be assigned in a classroom without ideological dispute. It carried the authority of the ASCE and the reputation of Francis. It entered the literature of civil engineering as a foundational document in the field of dam safety.

The report also preserved the counter-argument. The committee noted that the rainfall of May 30 and 31 was extraordinary. It noted that the reservoir was designed to hold a specific volume and that the inflow from the storm exceeded that volume. The implication was clear: even a well-maintained dam might have been overtopped.

The committee did not state this conclusion. It presented the data and allowed the reader to draw it. The data showed that the storm was severe. The data also showed that the modifications had reduced the margin between the reservoir’s capacity and the crest’s height. The data showed that the spillway, if cleared and unobstructed, might have passed enough water to prevent overtopping. The data showed that the discharge pipes, if present and functional, would have allowed the reservoir level to be lowered before the storm arrived.

The committee did not say that any single modification caused the failure. It said that the modifications, taken together, made the failure probable.

This was the warning that outlived the witnesses. Not a single sentence. Not a slogan. A body of data, presented in measured language, preserved in a professional journal, available to anyone who consulted the bound volume.

The counter-argument, the act-of-God defense, was also preserved. The committee’s own data supported the claim that the storm was exceptional. The rainfall records from the U.S. Army Signal Corps stations in the region showed totals that exceeded any previously recorded storm. The Little Conemaugh watershed was small. The reservoir was large relative to the watershed. The inflow was rapid. A dam built to the full original height, with all five discharge pipes installed and functional, and a clear spillway, might have held. It might not have. The committee could not say with certainty. The data did not permit a definitive answer.

But the data did permit a clear conclusion about the modifications. They reduced the margin. They narrowed the gap between safety and failure. They converted a structure that might have survived the storm into one that almost certainly would not. The deferred-maintenance debt — the accumulated cost of repairs not made, pipes not replaced, spillways not cleared, crests not maintained — was the difference. The debt was four feet of earth, five iron pipes, and a screen of fish-trap rods.

The report made this calculation visible. It made it permanent. It made it citable.

In the 1910s, the calculation began to matter in new ways. Pennsylvania was a state pockmarked with industrial dams. The anthracite fields of the northeast held hundreds of culm banks and water reservoirs. The bituminous fields of the west held tailings ponds and slurry impoundments. The steel towns of the Monongahela and Allegheny valleys held water-supply reservoirs for the mills. Each of these structures was a potential South Fork. Each was owned by a private company. Each was subject to no state inspection. Each was maintained, or not maintained, at the discretion of its owner.

The Johnstown precedent hung over these structures. Not as a memory. Not as a story told by survivors. As a technical document. An engineer who inspected a dam in 1915 and found the spillway obstructed could cite the ASCE report. He could point to the South Fork Dam and say: this is what happened when the spillway was blocked. He could point to the discharge pipes and say: this is what happened when they were removed. He could point to the crest and say: this is what happened when it was lowered. He did not need to argue. He needed only to present the numbers.

The numbers were in the file. The file was in the library. The library was open.

The men who had made the numbers were not. Parke was dead. Fulton was dead. Morrell had been dead since before the flood. Unger, the club president who had ordered the retreat to high ground, who had watched the dam go, who had given his testimony to the inquest and the newspapers, was gone. The Italian laborers who had dug at the spillway with picks and shovels in the final hours were unrecorded. Their names were never entered in any ledger. Their work, their fear, their recognition of futility, existed only in the paraphrased testimony of others. The testimony was in print. The men were in the ground.

The site itself was becoming unreadable. In 1889, the breach was a raw wound. The abutments stood exposed. The spillway channel was visible. The embankment’s cross-section was open to inspection. A trained eye could read the structure’s history in the layers of earth and rock. By 1910, the breach was soft. Vegetation covered the slopes. The spillway was a ditch filled with leaves and silt. The abutments were mossy. The embankment’s face was obscured by growth. A visitor without the report would not know what the structure had been. A visitor with the report could match the description to the remains. But the remains were fading.

The ASCE committee had surveyed the site in 1889. Their measurements were precise. They recorded the height of the embankment, the width of the crest, the dimensions of the spillway, the slope of the upstream and downstream faces. They recorded the location and diameter of the discharge pipe openings. They recorded the composition of the embankment material. They recorded the angle of the breach. These measurements were the last accurate survey of the dam as it existed at the time of failure. After 1889, the site changed. Erosion filled the breach. Vegetation covered the surfaces. The structure that the committee measured was disappearing. The measurements were not. They were fixed in the report. The site could erode. The data held.

This was the transfer. The warning moved from the physical to the archival. It moved from the living to the documented. It moved from the valley to the library. It moved from the men who saw the dam fail to the pages that described why it failed. The transfer was not complete. It was not sufficient. The report did not regulate anything. It did not compel any inspection. It did not prevent any future owner from making the same modifications. It was a document. It sat on a shelf. It was consulted by engineers who already understood its importance. It was ignored by everyone else.

But the transfer meant that the warning could not be lost. It could be forgotten. It could be ignored. It could be dismissed as ancient history. But it could not be lost. The measurements were in the Transactions. The calculations were in the textbooks. The conclusion was in the professional record. As long as the engineering profession existed, the warning existed. As long as engineers consulted the literature of their field, the South Fork Dam was there. Not as a memory. As a case study. As a set of numbers. As a calculation of what happened when the margin was erased.

The act-of-God defense required a well-maintained structure overwhelmed by an unforeseeable event. The committee’s data showed a poorly maintained structure overwhelmed by a foreseeable event. The storm was severe but not unprecedented. Storms of comparable intensity had struck the region before. The dam had been tested before, in its original configuration, and had held. The modifications had not been tested. The storm of May 30 and 31 was the first test of the modified dam. It failed.

The warning was specific. Not that storms were dangerous — engineers knew that. Not that dams could fail — engineers knew that. The warning was that the modifications made to the South Fork Dam had converted a structure that could withstand a severe storm into one that could not. The warning was that private owners, acting without engineering oversight, had reduced the safety margin of a dam impounding a large reservoir above a populated valley. The warning was that no law required them to do otherwise.

The law had not changed. The lawsuits had failed. The courts had held that the club was not liable. The Pennsylvania legislature had passed no dam safety law. The state had no inspection program. The Commonwealth that had built the dam in the 1840s and sold it to private owners in 1875 had not reclaimed any authority over its maintenance. The legal gap between the dam and the city was the same in 1911 as it had been in 1889. The gap was fourteen miles. The gap was the absence of regulation. The deferred-maintenance debt was still outstanding. No court had collected it. No legislature had appropriated it. The valley had paid it in lives.

In 1913, an engineer in Harrisburg opened the Transactions of the American Society of Civil Engineers to the Francis committee’s report. He was reviewing the technical literature on dam failures for a memorandum to the state legislature. The legislature was considering a bill to authorize the inspection of dams in Pennsylvania. The bill had been introduced before and had failed. It would be introduced again.

The engineer copied the committee’s measurements of the South Fork Dam into his memorandum. He copied the spillway dimensions. He copied the discharge pipe specifications. He copied the calculated reservoir volume. He copied the estimated flow rate at the breach. He noted that the modifications had reduced the dam’s discharge capacity by approximately half.

He noted that the storm, while severe, was not beyond the range of probable events. He noted that the original dam, unmodified, would likely have withstood the storm. He cited the report. He closed the volume. He submitted the memorandum.

The legislature did not act on the bill. The memorandum went into a file. The file went into a cabinet. The cabinet sat in an office in Harrisburg. The report sat in the library. The dam sat in the woods. The city sat below. The warning sat in both places. The knowledge was in the file. The law was not. The witnesses were dead. The numbers were alive. The legislature held its silence. The numbers held their places on the page. The dam held its breach, softening. The city held its ground, below.

The engineer in Harrisburg had done what Parke had done on the morning of May 31, 1889. He had read the signs. He had written them down. He had sent the message. The message had gone to the people who could act. The people had not acted. The dam had failed. The city had drowned. The message had survived. The messenger had died. The file held. The warning held. The gap held. The legislature held its silence. The dam held its shape in the woods, eroding. The city held its place below. The warning was in the file. The file was in the cabinet. The cabinet was in Harrisburg. Harrisburg was one hundred and fifty miles from Johnstown and fourteen miles from the dam. The distance was the gap. The gap was the warning. The warning was waiting for someone to read it and act.