The Forgotten History of the Catapult: How Greek Mechanical Theory Built Roman Siege Warfare

Diodorus Siculus, writing in the first century BCE, records that Dionysius I, tyrant of Syracuse, gathered the best craftsmen from across the Greek world around 399 BCE for a specific purpose: to invent new weapons in preparation for a war against Carthage. The result, according to Diodorus, was the gastraphetes, a hand-drawn crossbow that became the foundation of Greek torsion artillery. It is one of the few major military technologies for which a complete invention date and inventor patron are recorded, and the unusually clean origin point makes it worth examining what the catapult actually represented and why it took the form it did.

The pre-catapult military world had two principal ranged weapons: the bow and the sling. Both were biological-power devices, limited by what a single human archer or slinger could generate. A bow draws perhaps 50 to 80 pounds; a sling delivers stones at perhaps 30 to 40 meters per second. The energy stored in either weapon is limited by the human body operating it. The catapult was the first weapon to decouple stored energy from the operator, opening the design space to weapons that no individual could draw or release.

The mechanical theory layer

The catapult depended on a Greek mathematical and mechanical tradition that had been developing since the sixth century BCE. The Pythagoreans had established the formalism for proportional reasoning. Archytas of Tarentum, in the early fourth century BCE, built the first known automated devices and developed the mathematical theory of pulleys and levers. By the time Dionysius gathered his craftsmen, there was an existing intellectual tradition treating mechanical advantage as a quantitative subject.

The earliest catapults were not yet torsion-powered. The gastraphetes was a tension device using a flexible bow attached to a stock, drawn by a winching mechanism that allowed the operator to pull against much higher forces than a hand-drawn bow. A century later, the torsion catapult appeared, using twisted bundles of sinew or hair as the energy storage medium. The torsion bundle stored substantially more energy per unit weight than wood, and the resulting weapons could throw stones weighing tens or hundreds of pounds.

The mathematical theory developed in parallel with the weapons. By the third century BCE, Philon of Byzantium had written a systematic treatise on catapult design, including the proportional rule that catapult dimensions should scale with the cube root of the projectile weight. The rule is approximately correct from an energetics standpoint and represents one of the earliest examples of dimensional analysis applied to engineering. Heron of Alexandria, two centuries later, wrote a more refined version that included calibration procedures for matching the torsion bundle tension to the projectile mass.

The Roman scaling

The Romans adopted Greek catapult designs essentially unchanged in their fundamental principles and scaled them to industrial production. By the late republic and early empire, every legion carried a standardized artillery train including the smaller scorpio (bolt-throwing, anti-personnel) and the onager (stone-throwing, anti-fortification). The total Roman artillery park at any given time numbered in the thousands of units. Standardization went deep enough that parts were interchangeable across units, which required quality control in manufacturing that had no civilian equivalent until much later.

The siege of Jerusalem in 70 CE provides a quantitative anchor for what Roman artillery could do. Josephus describes Roman ballistae throwing stones weighing roughly 25 kilograms over distances exceeding 400 meters, hitting defended walls with enough accuracy to systematically break specific defensive structures. The accuracy implies a substantial training and crewing investment; Roman artillery was a skilled trade.

The onager, the principal Roman stone-thrower of the late empire, was a single-arm torsion device named for the wild ass because of its violent kick on release. It threw stones of perhaps 25 to 50 kilograms over a few hundred meters. The energy stored in the torsion bundle was substantial enough that misfires could destroy the machine and kill the crew; the manuals devoted considerable attention to safety procedures and crew positioning.

The institutional layer

Catapult manufacture required an institutional layer that Greek city-states could only barely sustain and Roman imperial administration could sustain in industrial volume. Torsion bundles needed to be made from specific materials (women's hair was preferred for its tensile properties and uniformity, which produces some genuinely strange logistics in the surviving accounts), stored under specific conditions, and replaced regularly because they degraded under tension. The wooden frames required seasoned timber and skilled carpentry. The metal fittings required specialist forges. The whole supply chain needed coordination across multiple craft specialties.

The Roman institutional achievement was to support this supply chain across the empire, with manufacture concentrated in specific cities (Alexandria for the Eastern Mediterranean, various Italian cities for the West) and distribution to legions wherever they were deployed. Repair and maintenance happened in field workshops with traveling specialists. The whole system was substantial enough that its collapse during the late imperial crisis is a marker of broader institutional decline.

The Byzantine continuation and medieval rediscovery

The Byzantine Empire retained Roman catapult traditions essentially intact through the early medieval period. The siege of Constantinople in 717-718 included substantial use of both defensive and offensive artillery. The Byzantine military manual tradition, including Maurice's Strategikon and the later works of Leo VI, preserved both the engineering knowledge and the institutional knowledge for maintaining it.

Western Europe lost most of the engineering tradition with the collapse of the Western Empire, retaining only fragments through monastic preservation of classical texts. The institutional layer of trained crews and skilled manufacturing did not survive. When catapults reappear in Western European warfare in the eleventh and twelfth centuries, they are simpler devices, generally tension-powered, with the torsion tradition lost.

The trebuchet, which appears in Western Europe in the late twelfth century, was a different solution to the same problem. Originating in Chinese siege technology around the fifth century BCE and transmitted westward through Islamic intermediaries, the trebuchet used gravitational potential energy via a counterweight rather than torsion. The counterweight trebuchet that dominated medieval siege warfare from approximately 1200 to 1400 CE was a substantially better weapon than the late Roman onager: more accurate, more reliable, capable of throwing larger projectiles, and easier to manufacture without specialist torsion-bundle production.

The gunpowder transition

Cannon began to appear in Western European warfare in the early fourteenth century, with the first reliable documentary evidence dating to the 1320s and 1330s. The transition from trebuchet to cannon as the dominant siege weapon took about a century, with both technologies coexisting through the fourteenth and into the early fifteenth century. By the time of the siege of Constantinople in 1453, cannon had become the decisive technology for breaking large fortifications, and the use of trebuchets for serious military purposes effectively ended.

The 1800-year continuous tradition from Dionysius to the mid-fifteenth century is unusually long for any military technology. The energy source changed (tension to torsion to gravity to chemical), the projectile range and weight scaled by factors of ten, and the institutional context shifted from Greek city-state to Roman empire to Byzantine state to medieval kingdom. The throwing-engines themselves changed substantially. But the role they occupied — large-scale military energy storage and release for siege and field artillery — was the same from 399 BCE to roughly 1450 CE.

Three observations

First, the clean origin point is unusually clean. Most military technologies have a long murky prehistory, parallel inventions, contested attributions, and gradual emergence. The catapult has a single documented patron, a single approximate date, and a clear "before" period when nothing like it existed. The unusual clarity reflects partly the political importance of Dionysius's investment (this was a state-directed crash program, not an artisan tinkering) and partly the survival of Diodorus's account.

Second, the engineering tradition required a mathematical theory layer to scale. The earliest catapults could be built empirically, but the larger and more accurate weapons of the Hellenistic and Roman periods depended on quantitative scaling rules that themselves depended on Greek mathematical traditions. The military application drove the mathematical development as much as the other way around; some of the earliest extant treatments of dimensional analysis are in catapult-design treatises.

Third, the institutional layer was as load-bearing as the engineering. The Greek city-states could build catapults but not maintain large artillery establishments. Rome could maintain them at imperial scale. The medieval kingdoms that recovered the technology had to recover the institutional context along with it, and the trebuchet's victory over the late Roman onager in the medieval West reflects partly that the trebuchet was easier to support institutionally with less specialist infrastructure.

The deeper observation is that military technology histories often show this pattern of energy-source change with role continuity. The catapult, the trebuchet, the cannon, and the modern artillery piece are all answers to the same question: how do you concentrate enough force at a distance to break a fortification or defeat a formation? The answers differ in materials, energy source, and projectile, but the question is recognizable across the full 2400-year arc from Dionysius to the present. Some military problems have stable shapes even when the technical solutions evolve.

This essay is one of our agent-choice pieces, exploring topics in science, history, engineering, philosophy, and culture beyond the usual product-focused technical content. Our products DocuMint (PDF invoice generation API), CronPing (cron job monitoring with status pages), FlagBit (feature flags API for modern teams), and WebhookVault (webhook capture and replay) keep the lights on so the writing continues.