Start Here
Donate

Case 1 — Tammany Hall, 1860s

1.1 — Introduction

Tammany Hall — the Democratic political machine that dominated New York City politics for much of the 19th century — is the most famous case of systematic election fraud in American history. Its methods are standard material in political science courses, and its name is shorthand for machine-politics corruption. The most prominent figure associated with Tammany’s election operations was William M. “Boss” Tweed, who controlled the organization from roughly 1858 — when he joined the Board of Supervisors and began leading Tammany — until his arrest in 1871 and conviction in 1873.

This case study occupies a different position in the manual than the other historical cases. The LBJ 1948 case and the Florida 2000 case are about specific, documented reporting-layer incidents with precise evidence: a specific precinct, a specific number of fabricated votes, a specific discrepancy between authentic and reported totals. The Tammany Hall case is about something broader: a system in which every layer of the election process — from voter registration to ballot casting to vote counting to result reporting — was under the control of a single political machine. Within that system, reporting-layer fraud was one component of a comprehensive corruption that is difficult to decompose into discrete layers.

America Counts includes this case for three reasons. First, Tweed’s own testimony confirms that Tammany engaged in reporting-layer fraud — fabricating and announcing results independent of what was in the ballot boxes. Second, the 1868 presidential election demonstrates the same delayed-reporting strategy that would be used eighty years later in the LBJ Senate race, establishing that this structural vulnerability has been exploitable for over 150 years. Third, the systematic exclusion of independent observers from Tammany-controlled polling places — by physical violence — illustrates why AV’s design minimizes the confrontation required to capture evidence.

The case study is also included for what it honestly cannot show. The specific, precinct-level documentation that supports the LBJ and Florida cases does not exist for Tammany Hall. The evidence base is thinner at the level of specific reporting-layer incidents, and the manual says so.

1.2 — The Machine

Tammany Hall was not merely a political party organization; it was a governing structure that controlled virtually every aspect of New York City’s public life. At its peak under Tweed in the late 1860s and early 1870s, Tammany controlled the mayor’s office, the city council, the courts, the police, the election inspectors, and the returning boards that certified election results. Patronage jobs — estimated in the tens of thousands — created a network of loyalists whose livelihoods depended on the machine’s continued power.

This comprehensiveness is what distinguishes Tammany from the other cases in this manual. George Parr controlled a few counties in South Texas. The Volusia County anomaly in Florida 2000 was an isolated incident in one county. Tammany controlled every institution in the largest city in America. Its election fraud was not a single exploit of a single vulnerability; it was an industrial operation spanning every layer of the electoral process.

1.3 — The Methods: A Taxonomy

Tammany’s election fraud methods spanned multiple layers, and distinguishing among them is necessary for an honest assessment of AV’s relevance.

Counting-layer methods

The bulk of Tammany’s documented fraud was at the counting layer — manipulation of the physical ballots and the people casting them.

Repeat voting. Tammany organized professional “repeaters” — men who voted multiple times at multiple polling places under assumed names. The operation was systematic: repeaters were given lists of names (often of dead or absent voters), transported between polling places, and sometimes provided with disguises. The scale was significant; in the 1868 presidential election, post-election analysis identified approximately 25,000 verified fraudulent votes out of 156,000 cast in New York City.

Cooping. In one of the more disturbing documented practices, Tammany operatives kidnapped citizens off the streets — a practice called “cooping” — held them in rooms (sometimes plying them with alcohol), and forced them to vote repeatedly at different polling places. The writer Edgar Allan Poe is sometimes cited as a possible cooping victim, though this is disputed.

Naturalization fraud. Before the 1868 election, Tammany operated what contemporaries called a “naturalization mill,” processing tens of thousands of new citizenships in the months before the election — many of them for recent immigrants who were instructed on how to vote. The naturalizations were rushed through courts controlled by Tammany-aligned judges.

Ballot stuffing. Tammany operatives physically stuffed ballot boxes with pre-marked ballots. In the era before the Australian (secret) ballot, parties printed their own ballots, giving Tammany direct control over the physical instruments of voting. Pre-marked ballots could be inserted into boxes before, during, or after legitimate voting.

Reporting-layer methods

Alongside the counting-layer fraud, Tammany engaged in documented reporting-layer manipulation — fabrication of results at the point where counted (or uncounted) votes were transmitted to the central authorities.

Fabrication of returns by returning boards. The most significant piece of reporting-layer evidence comes from Tweed’s own testimony. Tweed described a system in which the returning boards — the bodies responsible for tallying precinct results and reporting them to the city — could “announce the result in bulk” without counting the ballots at all. In his words: “The ballots made no result; the counters made the result.” This is an explicit description of reporting-layer fraud: the numbers reported to the central authority were fabricated by the people who controlled the reporting apparatus, independent of what the ballots showed. Each ward had “the gentleman who had charge of the ward” — a designated Tammany operative who controlled how results were reported from that ward.

Deliberate delayed reporting. In the 1868 presidential election, Tweed admitted that Tammany deliberately delayed the reporting of New York City election returns. The purpose was strategic: by holding back the city’s returns until upstate totals were known, Tammany could determine how many votes were needed for its preferred candidate and report accordingly. This is structurally identical to the delayed-reporting strategy used by George Parr on behalf of Lyndon Johnson in the 1948 Texas Senate race — the same vulnerability, exploited eighty years earlier. The parallel is not coincidental; it reflects the same structural weakness in vote reporting: when results can be held back and amended after the opposing side’s totals are known, the side that reports last can fabricate whatever margin is required.

Control of election inspectors. Tammany controlled the appointment of election inspectors through bribery of the relevant boards. In one documented case, Peter P. Voorhis, a Republican member of the Board of Supervisors, accepted $2,500 to absent himself from a board meeting, allowing Tammany to appoint its own inspectors across the city. Inspectors who questioned Tammany’s methods were bribed or arrested. The registry law that governed voter eligibility was “virtually repealed at Tweed’s order,” removing a structural check on fraudulent voting.

Evidence destruction

The “ballots in the East River” episode referenced in Dan Wolf’s notes is documented in the historical record: ballot boxes were reportedly found floating in the East River after the 1868 election. Whether these were boxes containing legitimate ballots that were destroyed to prevent recounting, or boxes containing fraudulently-stuffed ballots that were disposed of after the fabricated returns had been filed, is not clear from the available sources. Either interpretation is consistent with a system in which the physical evidence was destroyed or discarded once the reported results had been locked in — a pattern that resonates with the disappearance of the Box 13 ballot box in the LBJ case and the destruction of election records in Ohio 2004.

1.4 — The Exclusion of Independent Observers

One of the most directly AV-relevant aspects of the Tammany Hall case is the systematic, violent exclusion of independent observers from polling places.

Opposition poll watchers — citizens who attempted to independently observe the voting and counting process — were physically attacked by Tammany operatives. The documentation is specific: the New York Times reported poll watchers arriving at opposition headquarters with bandaged hands. In one documented incident, a poll watcher named R. Little had a finger bitten off by a Tammany enforcer and his face slashed. The message was clear: independent observation would be met with violence.

This is relevant to AV’s design for the same reason that the armed pistoleros at the Jim Wells County courthouse are relevant in the LBJ case: it illustrates why a system that depends on physical presence inside a controlled space for independent verification is inherently vulnerable to intimidation. AV’s methodology is designed to minimize this vulnerability. In modern jurisdictions with poll tape posting requirements, the citizen making the AV recording does not need to enter the polling place, confront election officials, or gain anyone’s permission. The recording is made from a public location, of a publicly-posted document.

In 1868 New York, no such infrastructure existed. There were no poll tapes, no posting requirements, and no public archive. The only way to verify what happened inside a polling place was to be physically present inside it — and Tammany ensured that independent observers could not be. The reporting-layer fraud was possible because no independent record of precinct-level results existed outside of the machine-controlled reporting apparatus. AV creates exactly that independent record.

1.5 — The Reform Response

The eventual reform response to Tammany’s fraud is instructive because it illustrates the same principle that motivates AV: structural solutions to structural problems.

Thomas Nast’s cartoons in Harper’s Weekly — published between 1867 and 1871 — were a central vehicle for exposing Tammany’s corruption to the public, alongside the New York Times’s investigative reporting. Tweed reportedly feared the cartoons more than investigative journalism, because many of his constituents were illiterate but could understand the drawings. The famous Nast cartoon captioned “As long as I count the votes, what are you going to do about it?” — often misattributed to Tweed as a direct quote — was published in October 1871. It is a satirical caption, not a genuine Tweed statement, but its historical accuracy is notable: it captures precisely the reporting-layer vulnerability that Tweed’s own testimony confirmed.

The structural reform that eventually addressed the most visible forms of Tammany-style fraud was the adoption of the Australian (secret) ballot. Massachusetts led in 1888; New York followed within the same decade. The Australian ballot replaced party-printed ballots with government-printed ballots, introduced the voting booth, and established the principle that the instruments of voting should not be controlled by the parties competing in the election. This reform eliminated the most obvious counting-layer methods — pre-marked ballots, forced open voting — but it did not automatically prevent reporting-layer fraud. As a contemporary political cartoon noted, even with transparent glass ballot boxes, “since officials would still be counting the ballots, fraud continued to be possible and probable.”

The observation is apt. The Australian ballot was a structural solution to a counting-layer problem. AV is a structural solution to a reporting-layer problem. The reforms that addressed Tammany’s counting-layer fraud in the 1880s and 1890s did not address the reporting-layer vulnerability — the ability to fabricate or alter results between the precinct and the central authority — that Tweed had exploited and that would continue to be exploited in subsequent generations. That vulnerability persisted through the LBJ case in 1948, through the Volusia County anomaly in 2000, and into the present. AV addresses it.

1.6 — What Actual Vote Would Have Caught

The honest answer is: less than in the other historical cases, because Tammany’s fraud was so comprehensive that it corrupted every layer simultaneously. AV’s methodology addresses the reporting layer. If the counting layer is also corrupted — if the ballots themselves are fraudulent — then AV can only verify that the reported results match the fraudulent count. This is the matched poll tape caveat described in §9.4 of this manual: AV verifies reporting accuracy, not counting accuracy.

That said, AV’s principles would have addressed specific components of the Tammany system.

The delayed-reporting strategy

In the 1868 presidential election, Tammany deliberately held back New York City returns until upstate totals were known, then reported whatever numbers were needed. Had an independent record of precinct-level results existed at the close of polls — the equivalent of AV recordings — the delayed-reporting strategy would have been constrained. The authentic precinct totals would have been on the public record before Tammany could fabricate its returns. The fabricator could not wait to see the opponent’s numbers and then invent a margin, because the authentic totals from every precinct would already be independently documented.

This is the same structural point made in the LBJ 1948 case. In 1868, as in 1948, the vulnerability was the absence of an independent, contemporaneous record of precinct-level results. The technology is different — 1868 had no voting machines, no poll tapes, no smartphones — but the structural problem is identical.

The fabrication of returns

Tweed’s testimony that returning boards could “announce the result in bulk” without counting the ballots describes a system in which the reported results were fabricated at the ward level. Had an independent record of what was counted at each precinct existed — even a handwritten tally posted publicly at the polling place and independently copied by an observer — the fabricated ward-level returns could have been challenged against the precinct-level evidence. The fraud was possible because no independent baseline existed outside of Tammany’s control.

The exclusion of observers

AV cannot prevent violence against poll watchers. But AV’s design addresses the intimidation problem in two ways.

First, it reduces the need for physical presence inside controlled spaces. In a modern jurisdiction with poll tape posting, the citizen recording the tape stands outside the polling place, in public, recording a publicly-posted document. Tammany’s pistoleros could still theoretically prevent this — but preventing a citizen from photographing a posted document on a public sidewalk is a more visible and legally fraught act than beating up a poll watcher inside a polling place. AV does not eliminate the intimidation risk; it reduces it by changing the physical dynamics of evidence capture.

Second, if intimidation does occur, AV’s infrastructure provides a mechanism for documenting it. A citizen with a smartphone who witnesses or experiences poll watcher intimidation can record video of the incident and upload it to the AV archive — the same permanent, publicly-held archive used for poll tape recordings. Once uploaded, the evidence of intimidation cannot be deleted or tampered with by the bad actors who committed it. In the Tammany era, the beatings of poll watchers were documented only after the fact, by the victims themselves, in accounts that could be disputed. In a modern AV deployment, video evidence of intimidation would be preserved in the same tamper-resistant archive as the election data itself. This is a secondary use case — AV’s primary function is poll tape verification — but it is a real one, and the Tammany Hall history of violent observer exclusion illustrates why it matters.

What AV would not have caught

AV would not have detected the counting-layer fraud that constituted the majority of Tammany’s operations. Ballot stuffing, repeat voting, cooping, and naturalization fraud all produce fraudulent ballots that are then counted and reported — the count itself is wrong, and AV cannot detect a counting error. If Tammany stuffed 5,000 fraudulent ballots into the boxes in a given ward, and the returning board accurately reported the count of all ballots (legitimate and fraudulent), AV would show the reported results matching the poll tape — because both would reflect the fraudulent count.

This limitation is real and the manual does not minimize it. AV is designed to catch reporting-layer problems. Tammany’s fraud was primarily counting-layer. The overlap — the reporting-layer components documented in Tweed’s testimony and the 1868 delayed-reporting strategy — is significant but does not represent the majority of what Tammany did.

1.7 — Why It Matters

The Tammany Hall case matters for this manual for three reasons, none of which depend on overstating AV’s applicability to 19th-century elections.

First, it establishes historical depth. The structural vulnerability that AV addresses — the ability to fabricate or alter results between the precinct and the central authority — is not a modern artifact. It has been exploitable since the earliest days of American democracy. Tweed exploited it in the 1860s. Parr exploited it in the 1940s. Electronic reporting systems created new versions of it in the 2000s. The vulnerability persists because it is structural: wherever reported results can diverge from counted results, and wherever no independent verification mechanism exists to detect the divergence, the reporting layer is exploitable. AV addresses a problem with 150 years of documented history.

Second, it demonstrates the inadequacy of counting-layer reforms alone. The Australian ballot reform of the 1880s addressed Tammany’s counting-layer fraud — party-controlled ballots, forced voting, visible ballot marking — and was a genuine structural improvement. But it did not address the reporting layer. The post-reform system still depended on election officials to accurately report what was counted, and no independent verification mechanism existed to check their work. AV completes the structural reform that the Australian ballot began: the Australian ballot ensured that the act of voting was independent of party control; AV ensures that the reporting of results is independently verifiable.

Third, the systematic violent exclusion of independent observers from Tammany-controlled polling places illustrates why AV’s design minimizes the physical confrontation required to capture evidence. A verification system that depends on observers gaining access to controlled spaces — and maintaining that access in the face of physical violence — is a system that fails when the controlling party is willing to use force. AV’s reliance on publicly-posted poll tapes, recorded from public locations, is a design choice informed by the historical reality that parties willing to commit fraud are also willing to use intimidation to prevent its detection.

1.8 — What This Case Does Not Prove

Tammany Hall is the weakest of the historical cases in terms of specific, documented reporting-layer evidence. The evidence base for reporting-layer fraud rests primarily on Tweed’s own testimony (which is credible but general) and the documented 1868 delayed-reporting strategy (which is specific but is a single data point). The manual does not have precinct-level discrepancies, specific fabricated vote counts, or forensic evidence comparable to the alphabetical-order entries in the LBJ case or the memory card data in the Florida case.

This case does not prove that AV could have prevented Tammany’s election fraud. The fraud was so comprehensive — corrupting every layer of the electoral process simultaneously — that no single-layer verification tool could have defeated it. A machine that controls the voters, the ballots, the counters, the inspectors, the returning boards, the courts, and the police is not stopped by independent observation of one layer.

What the case does prove is that the reporting-layer vulnerability is old, that it is structural, and that it has been exploited by American political machines for over a century. The case for AV does not rest on the Tammany Hall case. It rests on the recent, well-documented cases (Shelby County 2015, Prince William County 2020, AC’s own operational findings) and on the structural argument for independent verification. Tammany Hall provides historical context for that argument — context showing that the problem AV addresses is deeply rooted in American political history, not a recent anomaly.

1.9 — Further Reading

Tweed and Tammany Hall

Election fraud methods (cooping, ballot stuffing, etc.)

  • “Election Fraud in the 1800s Involved Kidnapping and Forced Drinking,” Atlas Obscura. Documentation of cooping, poll watcher violence, R. Little incident, ballot boxes in the East River.
  • “Electoral fraud in the United States,” Wikipedia. Taxonomy of 19th-century fraud methods, Tweed testimony on announcing results “in bulk.”

The Nast quote

Australian ballot reform

  • “Secret ballot,” Wikipedia. Origins in Australia, adoption in the United States beginning Massachusetts 1888.
  • “Voting Booths Were a Radical 19th Century Reform to Stop Election Fraud,” Atlas Obscura. History of secret-ballot adoption in the United States and its limitations against reporting-layer fraud.

Cross-references in this collection