The Day Private Gunmen Invaded a Pittsburgh Steel Town
Preparing the first passage.
Steel City Stories · Chapter 24
The receipt is a pair of barges. On July 5, 1892, about three hundred Pinkerton guards who had come in by train boarded the Monongahela and the Iron Mountain outside Pittsburgh and started up the Monongahela River toward the Carnegie Steel works at Homestead. The Battle of Homestead Foundation’s timeline, prepared by historian Millie Beik, fixes that departure and the next morning’s landing as the moment a contract dispute became an armed entry. Henry Clay Frick, Carnegie Steel’s general manager, had written the Pinkerton agency on June 2 asking for three hundred guards for the day he intended to open the mill nonunion. Leon Wolff’s 1965 account in American Heritage adds the commercial terms as Frick settled them: an armed force of three hundred men at five dollars a day, to be towed in before dawn on July 6 and put inside the works. The barges were not a metaphor for private power. They were the vehicle.
Homestead, with neighboring Munhall, was a borough of about twelve thousand in which the mill was the occupation. Wolff describes a works of thirty-eight hundred employees in which the Amalgamated Association of Iron and Steel Workers had only about three hundred twenty-five members, almost all of them skilled and comparatively well paid, and in which those members had nonetheless been able to bargain rules for the plant. The thousand-odd Slavic day laborers, paid in his account fourteen cents an hour, sat outside that craft bargain. The union was a minority with leverage, not a mass organization of every man in the mill. That structure explains both why Frick thought he could break a small union and why the July 1 lockout produced a shock. At a mass meeting the whole workforce, not just the lodge, voted to strike. The skilled union and the laborers who had been left out of its wage scale were, for the length of the crisis, on the same riverbank.
The contract Frick meant to end
The paper trail is older than the gunfire. Beik’s timeline starts the 1892 negotiations in January, against a contract signed in 1889 that expired on June 30. In February the company proposed a sliding scale that tied tonnage pay to the market and cut the rates of skilled workers. The union objected. In May Andrew Carnegie sailed for Britain and stayed away through the end of the year, leaving Frick in charge. On May 29 Frick issued an ultimatum: accept the company’s terms by June 24 or the mill would run nonunion. The last bargaining session failed on June 23. In June the company built a high fence, barbed wire at the top, around the works. Workers named it Fort Frick. The fence had a purpose that did not require a speech. It turned a steelworks into a place that could be held against the people who had just been locked out of it.
The shutdown was staged. On June 28 the company locked out eight hundred workers and closed the armor-plate and open-hearth departments. On the morning of June 29 about three thousand steelworkers rejected the final offer, formed an Advisory Committee, and set up headquarters in the Bost Building. That afternoon mill laborers walked off to join the locked-out workforce, and the company announced that it would no longer recognize the union. From June 30 through July 5 the Advisory Committee organized watches for Pinkertons and nonunion workers. On June 24, still before the lockout was complete, hundreds of Slavic workers had formed an Amalgamated lodge and voted to support the union. The craft union’s old boundary did not survive the week. A labor town armed itself, in organization if not yet in a single military command, because the employer had already chosen a reopening date and a private army.
Carnegie’s absence is part of the record and not an alibi the timeline accepts at face value. On July 10, speaking from Scotland, he denied any active role in the Homestead events. Frick had been his manager, the second figure in the firm, and the man who built the fence and ordered the guards. Wolff’s reading of the cables between them is that both men meant to smash the Homestead lodge. A historian’s reading of cables is not a jury verdict. What can be said from the timeline without mind-reading is simpler. The owner was out of the country. The manager executed a plan, announced in May and June, to reopen nonunion on July 6. The guards were purchased for that morning. Responsibility for the decision to send armed private men up the river sits on that correspondence and that contract, whether or not Carnegie’s later denial is believed.
July 6, before anyone agrees on the dead
Alert workers warned the town. The barges tried to land at the works. Gunfire followed. Beik’s timeline states the outcome that is not seriously contested. There were casualties on both sides. The Pinkertons eventually surrendered under a promise of safe passage. The crowd broke that promise, forced the guards to run a gauntlet, and beat them badly. On July 7 the agents were sent to Pittsburgh, and Homestead held funerals. The New York Times, in its issue of July 7, tried to freeze the toll at midnight while the shooting was still being counted. Its casualty record, as preserved from that day’s report, was ten killed, seven workmen and three Pinkerton men, and at least fifty wounded, of whom about twenty were thought to be serious. The paper also said there were grave fears about the next day. A midnight newspaper count is a document. It is not an autopsy roll.
Wolff, writing in 1965, refused to pretend the sources agreed. He wrote that it is difficult to estimate the casualties, that sources differ, and that men continued to die for weeks. Bullets, beatings, drowning, and suicide, in his tally, brought the death toll to approximately nine strikers and seven Pinkertons. He put the wounded from gunfire at some forty strikers and twenty Pinkertons, and he wrote that nearly all of the Pinkertons were hurt, in varying degrees, while running the gauntlet. A later compressed telling, including a Public Broadcasting Service biographical note on Frick, still uses the midnight proportion: three Pinkertons and seven strikers dead after the twelve hours of fighting. The range a careful chapter can defend is therefore a range. Contemporary publication on the night of the battle said seven workers and three agents dead, with at least fifty wounded. A major later reconstruction raised the dead to about nine workers and seven agents once delayed deaths and the gauntlet were included, and said the sources do not match. This chapter will not invent a single total in the gap.
Who fired the first shot is the same kind of problem. Later summaries of testimony preserve a boatman’s claim that workers fired first, and a New York Times report that the Pinkertons opened fire. Wolff’s narrative stages the first Pinkerton casualty on the gangplank. Those accounts cannot all be the transcript. The defensible statement is that firing began as the agents tried to get off the barges onto mill ground, that both sides shot, and that the first-shot argument was already a weapon in the publicity war before the dead were buried. Choosing a culprit for the first round would launder a conflict the witnesses did not settle.
Hugh O’Donnell and the promise that failed
Hugh O’Donnell belongs in the chapter because the sources put him on the riverbank, not because legend needs a hero. Wolff identifies him as a heater in the plant and as the man who tried, in the middle of the fight, to get the Pinkerton commanders to talk. He describes O’Donnell angling for surrender rather than a drawn battle, and he describes union officers, including national president William Weihe, losing the crowd when they asked for the agents to be let go. Beik’s timeline does not linger on O’Donnell’s name in the July 6 entry. It records the institutional fact Wolff is dramatizing. Surrender was offered under a promise of safe passage, and the crowd broke the promise. The agents were disarmed, marched, beaten, and only later handed toward the civil authorities and a train to Pittsburgh.
That failure is the moral knot of the day, and it does not require a speech. The Advisory Committee had spent a week trying to look like a government: badges, watches, a headquarters in the Bost Building, a claim that the town could police itself. A promise of safe passage was the committee trying to act like a state. The gauntlet was the evidence that it could not. Women and men whose neighbors had been shot were in the street. Outsiders had come in. The committee’s authority was real enough to negotiate a surrender and not real enough to enforce the terms. Wolff notes anarchists from Pittsburgh in the crowd and says O’Donnell had not yet tried to curb the desire for retaliation while the fight was hottest. Those are a historian’s scene judgments. The timeline’s harder fact is enough. The safe-conduct did not hold. Private guards who had come armed onto a community’s river were then beaten by that community after they had given up their rifles. Both facts are part of July 6. Using only one of them is propaganda.
Sheriff William McCleary had already told Governor Robert E. Pattison that the county could not keep the peace. Wolff quotes the governor’s early replies as a demand that local authorities exhaust their own means before the state would move. On July 8, after the battle, union and town representatives asked Pattison not to send troops. The governor’s patience, if that is what the delay was, ended within days. The battle had settled who held the mill on the afternoon of July 6. It had not settled who would hold it the next week.
The militia, which was the state’s answer
Beik’s timeline dates the order to July 10. Governor Pattison ordered eight thousand members of the Pennsylvania militia to Homestead. On July 12 Major General George R. Snowden took those men into the mill and the town and claimed absolute control of both. Wolff dates the order itself to July 11 and agrees on the consequence: under militia protection, replacement workers entered, and the strikers’ control of the works ended. The one-day disagreement about when the order was signed is small. The arrival on July 12 is the political fact. A private army had failed to take the plant. The public army took it without replaying the river battle. Snowden did not need a barge. He needed a railroad and a governor.
The company then did what the fence had been built for. Beik records that Carnegie Steel charged Amalgamated leaders, including burgess John McLuckie, with conspiracy and murder, and that by September one hundred sixty-seven people had been so charged. On July 16 the company told strikers they had until July 21 to apply for rehiring. None did. Replacement workers were hired. Housing was built for them inside the company’s control. Sympathy strikes broke out at other Carnegie works, including Duquesne, Beaver Falls, and Lawrenceville. From July 12 to July 14 the Judiciary Committee of the United States House sat in Homestead and Pittsburgh and took testimony about the Pinkerton agency. The battle was already a congressional subject while the mill was being restaffed under rifles that belonged to the Commonwealth.
The strike’s end was a vote, not a second battle. On November 18 the Amalgamated released two thousand mechanics and laborers from their pledge so they could return to work. On November 21 the Homestead lodge voted 101 to 91 to end the strike. In February 1893 McLuckie and other union men were acquitted of murder, and the union and the company dropped their remaining charges against each other. In May 1893 Pennsylvania passed the Kearns Act, aimed at the Pinkerton system that had just been displayed on the Monongahela. Wolff’s account of the economic aftermath is harsher than the timeline’s last lines. He writes that wages in Carnegie mills were cut more deeply than Frick had proposed in June, and he gives a heater’s helper falling from 4.85 cents per ton in February 1892 to 2.22 cents per ton in February 1894. He also writes that members of the Advisory Committee were blacklisted. Those wage figures are his reconstruction from the company’s scale, and they belong in the chapter as a historian’s evidence of what defeat cost, not as a ticker tape this project re-audited from the payroll books.
Berkman was a different day
Alexander Berkman does not belong in the battle paragraph. Beik’s timeline puts him on July 23, more than two weeks after the barges and eleven days after the militia entered the mill. He was an anarchist, identified in that timeline as Emma Goldman’s lover, and he shot Frick in Frick’s Pittsburgh office. He was arrested at once. Frick survived. Berkman served fourteen years of a twenty-two-year sentence. A Public Broadcasting Service account adds that Berkman also stabbed Frick with a sharpened file, that Frick told a deputy not to shoot the attacker, and that Berkman had posed as an employment agent to get into the office. Those details place the assault in a Pittsburgh office, not on the Homestead wharf. Berkman had no office in the Amalgamated and no command of the Advisory Committee. The attempt was revenge theater by a New York anarchist, and it landed on a strike that was already losing the mill to the militia.
The political damage was real precisely because the attack was not the strike. A union that had just been in a gunfight with private guards could be portrayed, after July 23, as adjacent to assassination even though the assassin was not theirs. Wolff writes that the attempt harmed the strikers’ cause. The timeline lets a reader see the sequence without that adjective. July 6 was the battle. July 12 was the occupation. July 23 was a separate crime against the company’s manager, in another building, by a man the union did not send. Keeping the dates apart is the whole correction. Folding Berkman into “the Homestead violence” as if he had been on the barge is how a labor war gets rewritten as a story about anarchists.
A craft union, a town, and a purchased force
The Homestead works had been part of Carnegie’s holdings since 1883, and the first open-hearth furnace there dated to 1886. Beik’s timeline sets those facts beside the union’s own birth in Pittsburgh in 1876 and the three-year Homestead contract of 1889. The point of the pairing is scale. An association that could sign a contract at one works was still a craft body in an industry that was learning how to run armor plate, open hearth, and a twelve-hour day as a single managerial system. In 1888 Carnegie had broken the union at the Edgar Thomson works in Braddock and, in the same era, workers there had gained and then lost ground on the eight-hour day. Homestead in 1892 was not the first time the company tested whether a lodge could survive a determined manager. It was the time the test happened in public, with barges, because the lodge was strong enough to make a quiet dismissal expensive.
Frick’s fence explains the military character of the strike better than any speech about class war. A wall topped with wire, built in June, before the contract died, is a preparation for siege. Wolff notes loopholes at shoulder height. Whether every measurement in a later narrative is exact, the workers’ name for the wall, Fort Frick, records how the structure was read on the ground. Employers had always used watchmen. The 1892 novelty was the combination of a fortified mill, a refused union, a published reopening date, and a detective agency hired by the hundreds. The Pinkerton firm sold a service that sat between the sheriff and the army. It could be engaged by letter, on June 2, for a landing on July 6. That letter is the chapter’s economic document. Armed men had a day rate. The day rate converted a labor dispute into a procurement.
The Advisory Committee’s attempt to look like a government was the workers’ answer to that procurement. Watches at the depots, badges for approved reporters, a headquarters in the Bost Building, and a claim that strangers could be turned out of town were the forms of authority available to people who did not own the fence. Wolff describes the committee bouncing unapproved strangers and guarding the Homestead and Munhall stations in the days before the landing. Those acts were illegal from the sheriff’s point of view and practical from the committee’s. A town that expected armed men to arrive by river was not going to leave the bank to a notice posted on the gate. The tragedy of the committee is that its best hour and its worst hour were the same afternoon. It could meet an invasion. It could not police the victory.
Congress arrived before the strike ended because the Pinkerton question had already outgrown Allegheny County. The House Judiciary Committee’s hearings of July 12 to July 14, held in Homestead and Pittsburgh while Snowden’s militia was taking the mill, treated the agency as a national problem of hired force. Pennsylvania’s Kearns Act, passed in May 1893, was the statute that followed the hearing season. A law against the Pinkerton pattern did not restore the Amalgamated lodge, which had voted itself out of the strike the previous November by the narrow margin of 101 to 91. Reform of the private army and destruction of the union were compatible outcomes. The state was willing to say that corporations should not land detective regiments on riverbanks. It was not willing, in 1892, to say that the men who had turned the regiment back were entitled to their contract.
The charges make that priority obvious. By September, Beik records, one hundred sixty-seven people had been accused of conspiracy and murder. In the fall, Chief Justice Edward H. Paxson of the Pennsylvania Supreme Court charged thirty-three Amalgamated members with treason, a word that converted a strike into an offense against the commonwealth. The February 1893 acquittals, and the mutual dropping of charges, closed the criminal story without reopening the mill on union terms. Acquittal is not reinstatement. A man can be found not guilty of murder and still be unemployable in the industry that blacklisted the committee. Wolff’s wage example, the heater’s helper cut from 4.85 cents a ton to 2.22 cents a ton between 1892 and 1894, is one way to measure the loss if his figures hold. The other way is the November vote itself. A lodge that divides 101 to 91 is a lodge ending a fight, not a lodge celebrating a river victory from July.
The hearings of July 12 to July 14 did something the battle could not. They moved the Pinkerton contract from a Homestead argument into a congressional record, while the militia was already inside the fence the contract had failed to capture. Beik’s timeline places those hearings in Homestead and Pittsburgh in the same days Snowden took the mill. A witness could describe a barge, a day rate, and a gauntlet to a House committee that had not been elected by the Advisory Committee and did not answer to Frick. That is a third sovereignty, after the company’s purchase of force and the governor’s loan of the militia. It produced, the following May, a state law aimed at the detective trade. It did not produce a reinstatement. The sequence is the chapter’s institutional finding. Armed private entry was later restricted. The wage scale Frick had demanded was not. A republic can forbid the barge and still ratify the lockout the barge was hired to finish.
What the day decided
The question Homestead forces is not whether workers may ever resist an armed landing. It is who, in an industrial town, is allowed to decide the terms of work, and which armed body is allowed to enforce the decision. The Amalgamated had power of a specific kind. A few hundred skilled men bargained for a works of thousands, won a contract in 1889, and discovered in 1892 that the company would rather fence the plant and rent a detective army than sign again. The laborers’ decision to walk out turned a craft dispute into a town dispute. The Pinkertons were the company’s answer to that enlargement: if the civil sheriff would not clear the bank, a private firm would sell the clearing. The price was five dollars a day and a barge. The result was a surrender and a gauntlet, a stack of dead that newspapers and historians still do not count the same way, and a governor who then supplied the armed force the company had failed to land.
Private gunmen did invade a Pittsburgh steel town. That sentence is accurate as a description of July 6 and incomplete as a description of the strike. The invasion failed. The state succeeded. By the time Snowden claimed the mill, the argument about wages had been replaced by an argument about sovereignty, and the company won the second argument under a flag. The union’s November vote, 101 to 91, is the sound of an organization ending a strike it no longer had the plant, the courts, or the public militia to sustain. The Kearns Act the next spring shows that Pennsylvania legislators were willing to restrain the Pinkerton trade after they had watched it. It does not show that they were willing to restrain Carnegie Steel. In 1898 Carnegie dedicated a library in Homestead. The timeline records the gift and the battle in the same chronology because the town had to live with both. A library does not refund a blacklist. A battlefield does not explain itself. The barges do. They were how a corporation tried to purchase the right to reopen a mill, and how a commonwealth eventually decided that this particular purchase would be replaced by its own soldiers.
Evidence Ledger
- Battle of Homestead Foundation timeline, compiled by Millie Beik. Contract dates, Fort Frick, the June lockout, the Advisory Committee, the July 5 barges, the July 6 battle and broken safe passage, Pattison’s July 10 order of eight thousand militia, Snowden’s July 12 entry, the House hearings, the charging of union leaders, the November votes, and Berkman’s July 23 attack as a separate event.
- Leon Wolff, “Battle at Homestead,” American Heritage, April 1965. Workforce and union figures, the Pinkerton contract, the casualty disagreement, O’Donnell’s role on the bank, the governor’s early refusal, and the wage consequences Wolff reconstructs after the defeat.
- New York Times, July 7, 1892. Midnight count of ten dead, seven workmen and three Pinkerton men, and at least fifty wounded.
- PBS American Experience, Henry Clay Frick. Compressed battle toll of three agents and seven strikers, and the July 23 office attack kept on a later date.
What remains disputed
The dead of July 6 are disputed in the sources this chapter uses, and the dispute is the finding. The Times on the morning after the battle said seven workers and three Pinkertons, with the count still open. Wolff says the sources differ and reaches about nine strikers and seven Pinkertons once later deaths, beatings, drowning, and a suicide are included. He also writes that thirty-five men were dead by November from the battle and subsequent violence. That November figure is not a July 6 body count, and other reputable summaries never use it. The first shot is likewise unsettled. O’Donnell’s leadership of the Advisory Committee is well attested in Wolff and in standard histories that rely on the congressional hearings. The Foundation timeline confirms the committee and the broken promise of safe passage without building the day around his name. How many militia arrived is stated as eight thousand on the Foundation timeline. Other brief accounts say more than four thousand at the mill and more on the hill, or eighty-five hundred. This chapter uses the Foundation’s order of eight thousand and does not pretend the muster roll was re-counted here. Berkman’s prison term, fourteen years of a twenty-two-year sentence, is the Foundation’s figure. It describes a punishment for July 23, not a casualty of July 6.
This is an original public expansion of the short website chapter. It does not replace the Google Drive FULL MASTER.
Historic Pittsburgh · Chapter 24
Archival photographs relevant to this chapter. Sourced from Wikimedia Commons with attribution — not modern stock art.

Homestead riot harpers 3c26046v
Wikimedia Commons · Homestead riot harpers 3c26046v.jpg · Drawn by W.P. Snyder after a photograph by Dabbs, Pittsburg. · Public domain

Homestead riot double pinkertons
Wikimedia Commons · Homestead riot double pinkertons.jpg · Unknown · Public domain

Carnegie Blast Furnaces, Homestead, Pa (NYPL b12647398-69739)
Wikimedia Commons · Carnegie Blast Furnaces, Homestead, Pa (NYPL b12647398-69739).tiff · Detroit Publishing Company · Public domain
Related Stories

The Strike That Turned McKees Rocks Into a Battlefield
In 1909, immigrant workers at Pressed Steel Car challenged opaque pay practices and factory discipline in a strike that exploded into deadly violence.

The Morning 239 Miners Vanished Beneath the Youghiogheny
The 1907 Darr Mine disaster killed 239 miners and exposed the lethal price hidden inside western Pennsylvania’s coal economy.