Spoken HistoryReal history. Real evidence.
399 B.C. to A.D. 550

The Roman Scorpion: Rome's Sniper Weapon

RomeArtillerySiege warfareEngineeringGallic WarJewish War

A Gaulish warrior stands on a burning wall at Alesia in 52 B.C., hurling lumps of tallow and pitch down onto the Roman siege works. Then something hits him. A bolt, loosed from a machine he never even saw, goes clean through his right side. He falls. The man beside him steps over the body and takes his place, and the same invisible sting takes him too. A third steps up. Then a fourth.

The Roman siege lines at Alesia, rebuilt at the MuséoParc Alésia: a ditch, a palisade and towers where the machines stood.
The Roman siege lines at Alesia, rebuilt at the MuséoParc Alésia: a ditch, a palisade and towers where the machines stood.Reconstruction at MuséoParc Alésia. Photo: Carole Raddato, CC BY-SA 2.0

Julius Caesar was watching, and he wrote it down because, he said, it was worth remembering. The machine he describes is the Roman scorpion: a device that stored the strength of twisted rope and released it in a single snap. This article takes one apart: how it worked, who built it, how deadly it really was, and what the evidence can and cannot prove. The truth is stranger than the legend.

Julius Caesar. His account of the Gallic War, Book 7, is the eyewitness source for the machines at Alesia.
Julius Caesar. His account of the Gallic War, Book 7, is the eyewitness source for the machines at Alesia.Bust of Julius Caesar: Wilfredor, CC BY-SA 4.0

Born in an arms race

The scorpion was not a Roman invention. It was Greek. Around 399 B.C., Dionysius, the tyrant of Syracuse, gathered the best craftsmen he could find and told them to invent new weapons. The result was the first true artillery: a giant crossbow, braced against the belly of the operator.

A few generations later, engineers discovered something far more powerful. Torsion. Instead of bending a bow, twist a rope and let it spring back. By the time of Alexander the Great, torsion machines were in action at the siege of Tyre, and Rome met them as a victim. A Byzantine encyclopedia, the Suda, says that when the Romans besieged Syracuse, Archimedes' little scorpions fired through tiny openings in the walls, straight at the soldiers below. The Romans took the hint.

Vitruvius and the recipe

By the age of Augustus the scorpion was standard kit, and we know how it was built because an army engineer wrote the manual. Vitruvius spent his career supplying and repairing ballistae, scorpions and other artillery, and he gives the recipe in Book 10 of his work on architecture.

Title page of the 1543 printed edition of Vitruvius, De architectura. The book itself is Roman; this edition is how it reached the Renaissance.
Title page of the 1543 printed edition of Vitruvius, De architectura. The book itself is Roman; this edition is how it reached the Renaissance.Title page, 1543 printed edition (BEIC), public domain

The recipe is beautifully simple, and one number rules everything: the length of the bolt. Divide it by nine and the result is the width of the hole the springs pass through. Every other part of the machine is measured in multiples of that hole. The channel the bolt slides along is nineteen holes long. No blueprint needed, just a bolt and some arithmetic.

Animation: the bolt length divided by nine gives one spring hole, and the bolt channel is nineteen holes long.
Animation: the bolt length divided by nine gives one spring hole, and the bolt channel is nineteen holes long.Original animation, Spoken History

How a torsion engine works

A bow stores its energy in bent wood. The scorpion stores it in twist. On each side of the frame sits a fat bundle of cord: sinew, or, as Vitruvius says, hair, generally women's hair. Each arm is jammed into the middle of its bundle.

Animation: the two spring bundles, with an arm set into the middle of each.
Animation: the two spring bundles, with an arm set into the middle of each.Original animation, Spoken History

Crank the winch and the bowstring drags both arms backward, twisting the bundles tighter and tighter, like wringing out a wet towel. Pull the trigger and all that stored twist snaps back, whipping the arms forward. The string hurls the bolt down its groove.

Animation: the winch draws the string back and the stored twist builds up in both bundles.
Animation: the winch draws the string back and the stored twist builds up in both bundles.Original animation, Spoken History

Tuned like a lyre

Crews did not just build these machines. They tuned them. Vitruvius says each side was tightened until, struck by hand, both springs gave the same note. Roman artillery was tuned by ear, like a lyre.

Animation: left and right springs tightened until they sound the same note.
Animation: left and right springs tightened until they sound the same note.Original animation, Spoken History

The real thing: a machine from Xanten

There is a real machine to point to. In 1999, pieces of a small Roman bolt-shooter were collected from a gravel pit near Xanten, in Germany: a first-century machine with its wooden frame almost intact. Replicas, built to the millimetre, show what it was. Compact, plated with metal, and light enough, says the engineer Alan Wilkins, to carry on one shoulder.

A spring and its washer on a replica scorpion built from the evidence.
A spring and its washer on a replica scorpion built from the evidence.Replica: Alan Wilkins. Photo: Dot Boughton, Portable Antiquities Scheme, CC BY-SA 2.0
A modern model of a cart-mounted bolt-shooter (carroballista).
A modern model of a cart-mounted bolt-shooter (carroballista).Modern model. Photo: Ron L. Toms, CC BY 3.0

A legion on the march

The writer Vegetius, late and describing an idealised older army, says every legion had fifty-five of these on carts, one for every century, with mules to pull it and a squad of eleven men to crew it. They stood behind the heavy infantry. Vegetius makes a bold promise: that against their onslaught, neither armoured cavalry nor shield-bearing infantry could stand. Treat that with care, because he wrote long after the period he praises.

Animation: fifty-five carroballistae per legion, one for every century (Vegetius).
Animation: fifty-five carroballistae per legion, one for every century (Vegetius).Original animation, Spoken History

Trajan's Column, dedicated in A.D. 113, shows these carts in the field. Mules in front, crews at work. Unlike Vegetius, this is contemporary Roman evidence.

A mule-drawn cart on Trajan's Column, from a cast of the reliefs.
A mule-drawn cart on Trajan's Column, from a cast of the reliefs.Trajan's Column cast: photograph, CC BY-SA 3.0, relief is contemporary Roman evidence

Shock weapons: Britain, 55 B.C.

These machines were not only killers. They were shock weapons. In Britain in 55 B.C. Caesar's transports were stuck off a beach crowded with British warriors, so he ordered his warships, loaded with slings, arrows and engines, to row around to the enemy's open flank. The Britons, Caesar says, were startled by the strange ships, the movement of the oars, and the engines, which were new to them. They fell back, just long enough for the eagle-bearer of the Tenth Legion to jump into the surf.

How deadly was it, really?

Out to about a hundred metres, a scorpion could pick off a single man. Lobbed in an arc, its bolts could carry four hundred metres or more. One Greek engineer claimed six hundred and forty metres with a big machine, a record that the writer who preserved it admitted was not easily believed.

Animation: aimed fire at about 100 metres, arcing fire at 400 metres or more.
Animation: aimed fire at about 100 metres, arcing fire at 400 metres or more.Original animation, Spoken History

The punch was real. Alan Wilkins's replica bolt-shooters have driven bolts through iron shield bosses backed with eighteen millimetres of plywood. Against steel plate, he says, even a man in body armour would have been put out of action.

What mattered most was outranging the enemy. In A.D. 62 the general Corbulo drove back the Parthians with catapults because, as Tacitus tells it, Roman stones and bolts flew further than Parthian arrows.

Jotapata, A.D. 67

Now imagine that, multiplied. In Judaea in A.D. 67 the future emperor Vespasian besieges the fortress of Jotapata and lines up around one hundred and sixty machines: bolts whistling through the air, stones weighing a talent (about 26 kilograms), fire, a storm of arrows. The Jewish commander inside was Josephus, and it is his account that survives. The noise of the engines, he says, was terrible. One stone took the head off a man standing beside him and flung the skull three furlongs. Josephus may be exaggerating, but this is how it felt on the receiving end.

A reconstructed stone-thrower at Yodfat, the site of ancient Jotapata.
A reconstructed stone-thrower at Yodfat, the site of ancient Jotapata.Reconstruction at Yodfat. Photo: Tombah, CC BY-SA 4.0

Cremona, A.D. 69

Artillery could even turn a battle. At Cremona in A.D. 69 Roman fought Roman, in the dark. Tacitus tells us that one immense engine, belonging to the Fifteenth Legion, was mowing down the enemy line with huge stones. The slaughter would have been enormous if two soldiers had not snatched up shields from the enemy dead, crept forward and cut the ropes that held its weights. They were cut down at once. Their names, Tacitus says, are lost, but the deed is certain.

A stone-throwing engine on Trajan's Column. A later monument, but the same kind of machine Tacitus describes.
A stone-throwing engine on Trajan's Column. A later monument, but the same kind of machine Tacitus describes.Trajan's Column, relief is contemporary Roman evidence (c. A.D. 113)

What the ground shows: Hod Hill and Maiden Castle

At Hod Hill in Dorset, excavators in the 1950s found seventeen iron bolt heads, most of them still standing in the chalk exactly where they had landed. Eleven were inside a single round hut. Only one had overshot. The archaeologist Sir Ian Richmond thought the Romans had picked out the chieftain's hut from somewhere between 130 and 170 metres and fired, to show the defenders what concentrated fire could do, and that the hill surrendered on the spot. A perfect demonstration of the scorpion's accuracy. Perhaps.

Animation: where the bolt heads were found, hut by hut (schematic).
Animation: where the bolt heads were found, hut by hut (schematic).Original animation, Spoken History

Another reading says the bolts were fired much later, as target practice, from the Roman fort built in the corner. Nobody can prove it either way.

Hod Hill today.
Hod Hill today.Photo: Dave Dunford, CC BY-SA 4.0

And the famous skeleton at Maiden Castle, with an iron point lodged in the spine, long labelled a Roman ballista bolt? A 2025 study from Bournemouth University says it is a spearhead, and that the Roman massacre never happened the way the textbooks told it.

Maiden Castle from the air, 1935.
Maiden Castle from the air, 1935.Aerial photograph, 1935: Major George Allen, public domain

Be careful when you hear that this bolt or that skeleton proves the scorpion's reach. The best evidence is in the writers, the carvings and the machines themselves.

How the scorpion changed

By the second century the scorpion was changing: iron frames and compact, arch-shaped bracing, which can be seen on Trajan's Column. In the Greek east the hand-held version became the cheiroballistra, literally the hand-ballista. Small iron frames like these have turned up along the Danube.

Reconstruction drawing of a cheiroballistra.
Reconstruction drawing of a cheiroballistra.Cheiroballista reconstruction drawing: Victor Prou, public domain
Ballista parts found at Ampurias, published in 1918.
Ballista parts found at Ampurias, published in 1918.Erwin Schramm, 1918, public domain

Even the name moved

To Vitruvius, a scorpio was a bolt-shooter and a ballista threw stones. Four centuries later, bolt-shooters were being called ballistae. And Ammianus Marcellinus, a veteran officer, wrote that scorpion now meant a different machine: the one-armed stone-thrower we call the onager. Its raised arm, he said, was the scorpion's sting.

Decline

The great torsion engines had weaknesses. Twisted-cord springs hated damp and swings in temperature. After Rome's western empire collapsed, the technology faded across western Europe. In the east it lived on. Procopius, writing in the sixth century, describes a bolt from a Roman ballista passing through a Gothic warrior's armour and body and pinning him to a tree.

The verdict

A sniper built from wood, iron and hair, tuned like a lyre. Accurate at a hundred metres, reaching four hundred, and frightening enough to stop an enemy in his tracks. But Rome had a bigger machine: one arm, a kick like a wild donkey, and boulders for ammunition. Next up: the onager, Rome's one-armed stone-thrower, and the machine that stole the scorpion's name.

Prefer to watch? The video on YouTube covers the same story with the full animations. Jump to a part:

  • 0:00 The wall at Alesia, 52 B.C.
  • 0:46 What is a Roman scorpion?
  • 1:08 Born in an arms race
  • 2:04 Vitruvius and the recipe
  • 2:49 How a torsion engine works
  • 3:28 Tuned like a lyre
  • 3:43 The real thing: a machine from Xanten
  • 4:10 A legion on the march
  • 4:48 Shock weapons: Britain, 55 B.C.
  • 5:22 How deadly was it, really?
  • 6:15 Jotapata, A.D. 67
  • 7:00 Cremona, A.D. 69
  • 7:36 What the ground shows: Hod Hill and Maiden Castle
  • 8:54 How the scorpion changed
  • 9:16 Even the name moved
  • 9:40 Decline
  • 10:06 Verdict, and what comes next

Where it happened

  • Alesia (Alise-Sainte-Reine, France) (52 B.C.): The siege where Caesar saw the machines at work, 52 B.C.
  • Syracuse, Sicily (399 B.C.): The first true artillery, about 399 B.C.; Archimedes' small scorpions, 213 B.C.
  • Tyre (332 B.C.): Alexander's siege, 332 B.C.
  • Xanten, Germany (A.D. 100): The real find: parts of a scorpion from Xanten.
  • Jotapata (Yodfat), Israel (A.D. 67): Vespasian's siege, A.D. 67.
  • Cremona, Italy (A.D. 69): The night battle, A.D. 69.
  • Hod Hill, Dorset, England (A.D. 44): Iron bolt heads found among the huts (the reading is disputed).
  • Maiden Castle, Dorset, England (A.D. 44): A skeleton with a bolt in the spine (the reading is disputed).

Sources

Ancient writers and modern work

Ancient writers

  • Julius Caesar, Gallic War 7.25 (Alesia) and 4.25 (Britain, 55 B.C.)
  • Vitruvius, On Architecture, Book 1 preface and Book 10, chapters 10 to 12 (the scorpion's proportions, sinew and hair, tuning by ear)
  • Vegetius, Epitome of Military Science 2.25 (one carroballista per century, eleven men, mules). Late, and describing an idealised older army.
  • Flavius Josephus, The Jewish War, Book 3 (Jotapata, A.D. 67). The figure of about 160 machines comes from one English translation, so treat it as approximate.
  • Tacitus, Histories 3.23 (Cremona, A.D. 69); Annals 15.9 (Corbulo, A.D. 62), as cited by Alan Wilkins
  • Ammianus Marcellinus, Book 23 (the scorpion as the one-armed engine)
  • Procopius, Gothic War (a bolt pinning a Goth to a tree), known through secondary summaries
  • The Suda (Byzantine encyclopedia), on Archimedes' small scorpions at Syracuse
  • Modern work

  • Alan Wilkins, writing on Roman bolt-shooters, including "The bolt-shooter: accuracy, range and effects" (author's PDF, 2015): replica trials, ranges, the Xanten-Wardt find
  • G. MacDhnouill and Nanzydon the Traveler, "A Modern Reconstruction of Vitruvius' Scorpion" (Bucknell University project binder)
  • Excavation reports and later reassessments for Hod Hill (Richmond, 1968 reading; Russell, 2019 reading) and Maiden Castle (Wheeler, 1943; Smith, Russell and Cheetham, 2025)
  • Conrad Cichorius, Die Reliefs der Traianssäule (1896-1900), and plaster casts of Trajan's Column
  • Where the evidence is disputed, the video says so: the Hod Hill bolts and the Maiden Castle skeleton are open questions, not proof.

    Where the evidence is disputed, the article and the video say so. Reconstructions and replicas are labelled as such.

    Image credits

    • Modern model. Photo: Ron L. Toms, CC BY 3.0
    • Museum reconstruction. Photo: John E Ryelea, CC BY-SA 4.0
    • Photo: Dave Dunford, CC BY-SA 4.0
    • Photo: MatthiasKabel, CC BY-SA 3.0
    • Photo: Proa 500, CC BY-SA 4.0
    • Re-enactment at Yodfat. Photo: Tombah, CC BY-SA 4.0
    • Reconstruction at MuséoParc Alésia. Photo: Carole Raddato, CC BY-SA 2.0
    • Reconstruction at Yodfat. Photo: Tombah, CC BY-SA 4.0
    • Replica scorpion (Legio VIII Augusta). Photo: Pierre C, CC BY-SA 3.0
    • Replica: Alan Wilkins. Photo: Dot Boughton, Portable Antiquities Scheme, CC BY-SA 2.0
    • Reproduction at Alésia. Photo: Rolf Krahl, CC BY-SA 2.0
    • Trajan's Column: plates by Conrad Cichorius (1896) and photographs by Joe Mabel (CC BY-SA 3.0); the reliefs are contemporary Roman evidence
    • Ballista parts from Ampurias: Erwin Schramm, 1918, public domain
    • Cheiroballistra reconstruction drawing: Victor Prou, public domain
    • Vitruvius: title page of the 1543 printed edition (BEIC) and a medieval manuscript (Metropolitan Museum of Art, CC0)
    • Maiden Castle aerial photograph 1935 and excavation 1937: Major George Allen, public domain
    • Bust of Julius Caesar: Wilfredor, CC BY-SA 4.0

    Transcript

    Read the full narration

    The Roman Scorpion. Rome's sniper machine, taken apart. Fifty-two B.C. A Gaulish warrior stands on a burning wall, hurling lumps of tallow and pitch down onto the Roman siege works. Then something hits him. A bolt, loosed from a machine he never even saw, goes clean through his right side. He falls. The man beside him steps over the body, and takes his place. The same invisible sting takes him too. A third steps up. Then a fourth. Julius Caesar was watching, and he wrote it down, because, he said, it was worth remembering. This is the Roman scorpion. A machine that stored the strength of twisted rope, and released it in a single snap. Today we're taking one apart. How it worked. Who built it. How deadly it really was. And what the evidence can prove, and what it can't. Because the truth is stranger than the legend. The scorpion wasn't a Roman invention. It was Greek, born in an arms race. Around three ninety-nine B.C., Dionysius, the tyrant of Syracuse, gathered the best craftsmen he could find, and told them to invent new weapons. The result was the first true artillery. A giant crossbow, braced against the belly. A few generations later, engineers discovered something far more powerful. Torsion. Instead of bending a bow, you twist a rope, and let it spring back. By the time of Alexander the Great, torsion machines were in action at the siege of Tyre. And Rome met them as a victim. A Byzantine encyclopedia, the Suda, says that when the Romans besieged Syracuse, Archimedes' little scorpions fired through tiny openings in the walls, straight at the soldiers below. The Romans took the hint. By the age of Augustus, the scorpion was standard kit. And we know how it was built, because an army engineer wrote the manual. Vitruvius spent his career supplying and repairing ballistae, scorpions and other artillery, and he gives the recipe in Book Ten of his great work on architecture. It's beautifully simple. One number rules everything. The length of the bolt. Divide it by nine. That's the width of the hole the springs pass through. Every other part of the machine is measured in multiples of that hole. The channel the bolt slides along? Nineteen holes long. No blueprint needed. Just a bolt, and some arithmetic. Now here's the genius. A bow stores its energy in bent wood. The scorpion stores it in twist. On each side of the frame sits a fat bundle of cord. Sinew, or, as Vitruvius says, hair. Generally women's hair. Each arm is jammed into the middle of its bundle. Crank the winch, and the bowstring drags both arms backward, twisting the bundles tighter and tighter, like wringing out a wet towel. Pull the trigger, and all that stored twist snaps back, whipping the arms forward. The string hurls the bolt down its groove. And crews didn't just build these machines. They tuned them. Vitruvius says each side was tightened until, struck by hand, both springs gave the same note. Roman artillery was tuned by ear, like a lyre. And we can hold the real thing. In 1999, pieces of a small Roman bolt-shooter were collected from a gravel pit near Xanten, in Germany. A first-century machine, with its wooden frame almost intact. Replicas, built to the millimetre, show what it was. Compact. Plated with metal. Light enough, says the engineer Alan Wilkins, to carry on one shoulder. Now picture a legion on the march. The writer Vegetius, late, and describing an idealised older army, says every legion had fifty-five of these on carts. One for every century. Mules to pull it. And a squad of eleven men to crew it. They stood behind the heavy infantry. And Vegetius makes a bold promise: that against their onslaught, neither armoured cavalry nor shield-bearing infantry could stand. Trajan's Column, dedicated in A.D. one hundred and thirteen, shows these carts in the field. Mules in front. Crews at work. These machines weren't only killers. They were shock weapons. Britain, fifty-five B.C. Caesar's transports are stuck off a beach crowded with British warriors. So he orders his warships, loaded with slings, arrows and engines, to row around to the enemy's open flank. The Britons, Caesar says, were startled by the strange ships, the movement of the oars, and the engines, which were new to them. They fell back. Just long enough for the eagle-bearer of the Tenth Legion to jump into the surf. So how deadly was it, really? Out to about a hundred metres, a scorpion could pick off a single man. Lobbed in an arc, its bolts could carry four hundred metres, or more. One Greek engineer claimed six hundred and forty metres with a big machine, a record that the writer who preserved it admitted was, quote, not easily believed. And the punch was real. Alan Wilkins's replica bolt-shooters have driven bolts through iron shield bosses backed with eighteen millimetres of plywood. Against steel plate, he says, even a man in body armour would have been put out of action. What mattered most was outranging the enemy. In A.D. sixty-two, the general Corbulo drove back the Parthians with catapults, because, as Tacitus tells it, Roman stones and bolts flew further than Parthian arrows. Now imagine that, multiplied. Judaea, A.D. sixty-seven. The future emperor Vespasian besieges the fortress of Jotapata, and lines up around one hundred and sixty machines. Bolts whistling through the air. Stones weighing a talent, about twenty-six kilograms. Fire. A storm of arrows. The Jewish commander inside was Josephus, and it's his account that survives. The noise of the engines, he says, was terrible. One stone took the head off a man standing beside him, and flung the skull three furlongs. Josephus may be exaggerating. But this is how it felt, on the receiving end. Artillery could even turn a battle. Cremona, A.D. sixty-nine. Roman fights Roman, in the dark. Tacitus tells us that one immense engine, belonging to the Fifteenth Legion, was mowing down the enemy line with huge stones. The slaughter would have been enormous, if two soldiers hadn't snatched up shields from the enemy dead, crept forward, and cut the ropes that held its weights. They were cut down at once. Their names, Tacitus says, are lost. But the deed is certain. But what does the ground itself show? At Hod Hill, in Dorset, excavators in the nineteen-fifties found seventeen iron bolt heads, most of them still standing in the chalk exactly where they had landed. Eleven were inside a single round hut. Only one had overshot. The archaeologist, Sir Ian Richmond, thought the Romans had picked out the chieftain's hut, from somewhere between one hundred and thirty and one hundred and seventy metres, and fired. To show the defenders what concentrated fire could do. And that the hill surrendered on the spot. A perfect demonstration of the scorpion's accuracy. Perhaps. Another reading says the bolts were fired much later, as target practice, from the Roman fort built in the corner. Nobody can prove it either way. And the famous skeleton at Maiden Castle, with an iron point lodged in his spine, long labelled a Roman ballista bolt? A twenty twenty-five study from Bournemouth University says it's a spearhead, and that the Roman massacre never happened the way the textbooks told it. So be careful when you hear that this bolt, or that skeleton, proves the scorpion's reach. The best evidence is in the writers, the carvings, and the machines themselves. By the second century, the scorpion was changing. Iron frames. Compact, arch-shaped bracing, which you can see on Trajan's Column. In the Greek east, the hand-held version became the cheiroballistra. Literally, the hand-ballista. Small iron frames like these have turned up along the Danube. Even the name drifted. To Vitruvius, a scorpio was a bolt-shooter, and a ballista threw stones. Four centuries later, bolt-shooters were being called ballistae. And Ammianus Marcellinus, a veteran officer, wrote that scorpion now meant a different machine: the one-armed stone-thrower we call the onager. Its raised arm, he said, was the scorpion's sting. The great torsion engines had weaknesses. Twisted-cord springs hated damp, and swings in temperature. And after Rome's western empire collapsed, the technology faded across western Europe. In the east, it lived on. Procopius, writing in the sixth century, describes a bolt from a Roman ballista passing through a Gothic warrior's armour and body, and pinning him to a tree. So that's the Roman scorpion. A sniper built from wood, iron, and hair. Tuned like a lyre. Accurate at a hundred metres. Reaching four hundred. And frightening enough to stop an enemy in his tracks. But Rome had a bigger machine. One arm. A kick like a wild donkey. And boulders for ammunition. Next up: the onager. Rome's one-armed stone-thrower, and the machine that stole the scorpion's name. It's on your screen right now. Tap it, and I'll see you behind the wall.

    More episodes