TRACE THE
FAILURE
ANCIENT STRUCTURAL FAILURE

Tacitus named two faults.
Rome made a rule for one.

In AD 27 a privately built wooden amphitheatre at Fidenae, a few miles north of Rome, broke apart under a packed crowd at a gladiator show. Tacitus blamed the ground under it and the joints of the frame above. The Senate's new building rule, as he reports it, covered the ground and said nothing about the frame.

Follow the evidence
Illustrative AI image of tall wooden amphitheatre tiers breaking apart in a cloud of dust while people in Roman-era clothing run away; not a depiction from any ancient source.
Illustrative AI image made for the episode; not a photograph or recording of the Fidenae collapse. No ancient depiction of the amphitheatre is known, so its size and design here are invented.
THE COMPANION FILMWhy a Roman Amphitheatre Collapsed With Thousands Inside Watch episode 4:15

Why did the Fidenae amphitheatre collapse?

In AD 27, at Fidenae, a town a few miles north of Rome, a wooden amphitheatre built for a gladiator show collapsed while it was packed with spectators. The Roman historian Tacitus gives two reasons. The builder, a man named Atilius, did not lay the foundations on solid ground, and he did not fasten the wooden structure above them firmly together. Tacitus, Annals 4.62, Latin text; Loeb translation, 1937.

Afterward, Tacitus reports, the Senate ruled that no one worth less than 400,000 sesterces could stage a gladiator show and that an amphitheatre could stand only on ground of tested solidity. As he reports it, the rule said nothing about the timber frame. Atilius was exiled. Tacitus, Annals 4.63.

No investigation report survives for Fidenae. The record is what two Roman historians, Tacitus and Suetonius, wrote roughly 80 to 95 years later, and Tacitus’s two faults are the only physical causes it gives. Neither account describes the size of the building, its layout, the soil or the way the joints were made. Tacitus dates the disaster by naming that year’s two consuls; converted to our calendar, that is AD 27. Livius, “Fidenae”; Pleiades, Fidenae.

Watch the Fidenae episode

The companion episode shows the Latin sentence where Tacitus names both faults, explains how a timber stand carries a crowd’s weight, and sets Fidenae beside a stand that fell in Corsica in 1992. Its scenes of the amphitheatre and of the 1992 stand are AI illustrations, not records of either event; no ancient image of the Roman building is known.

The builder and the crowd

Tacitus calls the builder “Atilius quidam libertini generis”: one Atilius, of the freedman class, meaning he or his family had once been enslaved. Tacitus says Atilius took the project on for sordid gain, not because he had money to spare or wanted standing in his town. That verdict is Tacitus’s own, and these sources record nothing else about Atilius. Tacitus, Annals 4.62, Latin and English.

Under Tiberius, Tacitus writes, Romans had been kept away from such entertainments. The emperor gave no public shows of his own, cut the cost of games and shows and capped the number of gladiator pairs; no source says he banned them. When a show came to a town this close to Rome, men and women of every age crowded in. Neither Tacitus nor Suetonius gives an attendance figure. Suetonius, Life of Tiberius 34 and 47; Tacitus, Annals 4.62.

Two faults: the ground and the frame

Both faults come from one Latin sentence: “neque fundamenta per solidum subdidit neque firmis nexibus ligneam compagem superstruxit.” In English: Atilius neither laid the foundations on solid ground nor built the wooden frame above with firm fastenings. Tacitus, Annals 4.62.

Tacitus says the packed structure broke apart, collapsing inward and spilling outward, and buried both the people watching the show and the people standing around it. It began and ended in a moment, he writes, and this unexpected disaster equalled the losses of great wars. He does not say which part gave way first. Tacitus, Annals 4.62.

Tacitus says nothing more about the structure, so what follows is general engineering, not a reconstruction of Fidenae. A wooden stand rests on posts, and each post carries part of the crowd’s weight down into the ground. If the ground under one post sinks, the frame has to move that weight somewhere else, and it does so through its joints. Joints that were never fastened to carry that load can pull apart instead. This is one way a timber stand can fail; no source says it is what happened at Fidenae.

How many died at Fidenae?

The two ancient writers give different numbers, and they count different things. Tacitus says 50,000 people were maimed or crushed, a figure that includes the injured. Suetonius says more than 20,000 died. Tacitus, Annals 4.63; Suetonius, Life of Tiberius 40.

Neither number can be checked. The 1937 Loeb translation of Tacitus notes that the figures “will not bear a moment’s examination.” Tacitus, Annals 4.63, Loeb vol. IV (1937), LacusCurtius note 14. Both counts come from historians who were not there and wrote decades later.

What Rome changed: two rules, and the fault they left out

The Senate passed two rules for the future. No one worth less than 400,000 sesterces could stage a gladiator show, and no amphitheatre could be built except on ground of tested solidity, “solo firmitatis spectatae.” Atilius was sent into exile. Tacitus mentions no trial. Tacitus, Annals 4.63.

In the days after, Tacitus writes, Rome’s great houses opened their doors, and dressings and doctors were provided for anyone who came. Suetonius says Tiberius had just left for the island of Capri, and that the people’s pleas over Fidenae brought him back to the mainland, where he let anyone approach him. Tacitus, Annals 4.63; Suetonius, Life of Tiberius 40.

As Tacitus reports it, the decree named no permit, inspector or engineer. Laws on bad building were also far older: the Code of Hammurabi made a builder answerable with his life if a house he built badly fell and killed its owner. Code of Hammurabi, laws 229–233.

Tacitus had named two faults, the ground and the frame. The Senate’s building rule, as he reports it, covered the ground. Nothing in his summary says how the timber above had to be built or fastened.

Furiani, 1992: the ground was approved and the stand still fell

More than nineteen centuries later, a temporary stand in France collapsed after the one favourable opinion it received had covered only its ground. On May 5, 1992, minutes before a French Cup semi-final between SC Bastia and Olympique de Marseille, the stand gave way at the Stade Armand-Cesari in Furiani, on Corsica.

The French Court of Cassation’s 1997 decision restates the facts the courts established. The club had the stadium’s 800-place north stand torn down and hired a firm to build a temporary stand of about 9,000 places in its place; with a recently added west stand, that was meant to take the stadium from 6,800 to nearly 18,000 spectators. Nobody applied to the mayor of Furiani for a permit. Construction began on April 28 and finished on the day of the match. The upper part of the stand separated from the lower part along its whole length and fell, with 3,500 people on it. Cour de cassation, criminal chamber, June 24, 1997, no. 96-82.424, facts and procedure.

The toll depends on the source. The 1997 decision, restating the appeal court’s findings, says 15 of the 3,500 died and nearly 2,000 were injured. Cour de cassation, facts and procedure. A 2020 French National Assembly report gives 19 dead and 2,357 injured; the sources do not explain the difference. Assemblée nationale, report no. 2655, PDF pp. 5, 21.

An outside inspection firm gave one favourable opinion, on April 28, and it covered only whether the ground could carry the stand’s loads. The firm’s local director climbed onto the stand on May 2, and diagonal braces were added to its back afterward. The safety commission’s last meeting, on May 5 shortly before the crowd arrived, gave no definitive opinion; by then no inspection of the stand had been carried out, and workers were still on it. Court-ordered experts found that the collapse was inevitable because of errors in how the stand was designed and assembled, and because of the loads it had to carry. The appeal court called the abnormally short assembly time one of the essential causes, and said a simple calculation could have shown the stand’s probability of failure was on the order of 100%. Cour de cassation, June 24, 1997, facts and procedure, grounds and operative part.

In 1997 the Court of Cassation rejected the appeals of the inspection firm’s local director, the prefect’s chief of staff and the head of the French Cup’s central commission against their convictions for involuntary homicide and injury. It also rejected the appeal of a club vice-president convicted over a document that falsely claimed the safety commission had approved the match, used to obtain extra tickets. The case against the French federation’s director general ended with his death in 1996. Cour de cassation, June 24, 1997, operative part.

Rome’s answer to Fidenae was a rule about the ground. At Furiani, the inspection firm’s only favourable opinion concerned the ground, nobody signed off on the structure above it, and the stand still fell because of errors in how it was designed and assembled in about a week. The parallel is in the records; nothing links the two events directly.

For a collapse that investigators could trace to a single connection, read how one design change doubled the load on a Hyatt Regency walkway connection.

Primary sources and visual provenance

  1. Tacitus, Annals 4.62–4.63 (Latin). The Latin Library. The builder, the two faults, the collapse, the toll, the Senate decree and Atilius’s exile. Latin text.
  2. Tacitus, Annals IV.57–75, Loeb Classical Library translation (1937). Hosted by LacusCurtius. Chapters 62–64 and the translator’s notes, including note 13 on the town’s distance from Rome and note 14 on the casualty figures (LacusCurtius numbering). English text.
  3. Suetonius, Life of Tiberius 34, 40 and 47. Games cut back (34, 47); the toll of more than 20,000 and Tiberius’s return from Capri (40). Loeb translation by J. C. Rolfe (1913), hosted by LacusCurtius: English text. The Latin Library: Latin text.
  4. Livius, “Fidenae”. J. Lendering. Used for the year, AD 27. Article.
  5. Pleiades, place 422923, Fidenae. The site’s coordinates, a few miles north of Rome. Gazetteer entry.
  6. Code of Hammurabi, laws 229–233. L. W. King translation, Yale Avalon Project. Text.
  7. Cour de cassation, criminal chamber, June 24, 1997, no. 96-82.424. Facts established by the Bastia Court of Appeal (December 13, 1995), the grounds of appeal and the operative part. Juricaf.
  8. Assemblée nationale, report no. 2655 (2020). Committee report by M. Castellani; Furiani figures on PDF pp. 5 and 21. Report PDF.

The article image and preview thumbnail are illustrative AI images made for the episode; the thumbnail adds editorial typography. No ancient drawing or description of the Fidenae amphitheatre’s size and design is known, so the building in these images is invented. The episode uses AI narration and AI-generated illustrative scenes, including its scenes of the 1992 Furiani stand, alongside the Latin text of Tacitus, the Loeb translator’s note and the 1997 French court decision.