Building a torsion catapult after representations on the Trajan's Column (Rome)


The sequence of functions should correspond to the ancient models and be reliable and reproducible.

The catapult should optically correspond to the representations on the Trajan's Column and implemented as a double armed system with a torsion drive. The arrow- firer should take place on a tripod, an allocation in two part-load - tripod and catapult - had to be possible. The catapult should be transportable by means of box cars. The maximum overall length was thereby submitted on 2,2m.

As models for the arrow projectiles findings from the Syrian Dura Europos served. The arrow length should be between 45 and 50 cm, which amount to weight approx. 350g.

The calibre - defined as the inside diameters of the washers - should amount to 80 mm.

A goal was not to be built it one up to the "last nail “authentic reconstruction.
This was not possible because of the unfavourable source situation - findings clearly the catapult representations on the Trajan's Column to assign can be been missing so far.

For reaching the goals, compromises had to be entered regarding the materials and manufacturing possibilities.

Since no smithy was present, the parts with that available means were manufactured.: milling, bending and drilling and then so worked on that the impression developed of forging.

The cast parts out of bronze (washers) were manufactured from brass. Those corresponds to its alloy after Roman bronze. The treatment takes place by machine through tricks, milling and boring. The "casting effect “was reached by turned radii and formation bevels.


Manufacturing and material choice

The felling trees hurry from oak wood were manufactured. The parts were supplied in their rough dimensions and worked over again.

On animal chords for covering one did without, the manufacturing was obviously a high specialized work and the ancient manufacturing process is still unsettled. Hair hurry as second choice were not available anywhere . Hemp hurry proved by attempts as unfit, the rope stretch too much and are not enough flexible. Twisted plastic hurry, so-called „starter hurry “ proved itself as useful.

The catapult arms were first made of beech wood. Later one changed over to ash.

The clamping chucks were manufactured from brass hollow bars. The cylindrical caps for the torsion stands from brass plate.







Illustration 1

Catapult no. 1, distance of  the
torsion systems 450 mm

Catapult no. 1 did not have torsion stands from wrought steels to separate two cylindrical hollow cylinders  (centrifugal casting) made of bronze. The arc-shaped upper strut  was on the left and on the right riveted in each case. The lower strut was clasped and riveted with 2 locking rings at the torsion sockets. This system not pursued, however was technically feasible.




llustration 2

Catapult no. 1, with torsion systems out of cast bronze.

Illustration 3
Catapult no. 2, with externally working catapult system.
Distance of the torsion systems approx. 600 mm.
The tripod
  was replaced later through a shortened
with wheel set replacing.

The clamping frame

During the construction of the clamping frame findings from the Romanian Orsova and Lyon (France) gave orientation. The lateral torsion stands participate as braced forging implemented construction. This is also valid for the upper strive with its characteristic, centric projection.

The illustrations of the Trajan's Column show the torsion stands as cylindrical construction units with attached top and bottom caps. Today one assumes it thereby acted around sheet metal disguised torsion systems of the same design as with the findings out of Romania and France.

Deviating from the finds a not solvable, riveted execution of the clamping frame was selected. The interpretation at that time of the representations was decisive on the Trajan's Column. The possibility of exchanging torsion stands was therefore deliberately omitted.

Illustration 4
Catapult no.2. Now with two wheels.


The manufacturing of the clamping frame

The clamping frame was made of square and rectangle iron. The typical "grooves" of the bend and lowest vine was in-milled. The latches for binding for the locking collars of the torsion stands were welded. The frame parts were roughly worked over again afterwards in order to reach „a forged “appearance. As corrosion protection a forging pitch was used.



Illustration 5
Clamping frame with riveted props. Version catapult 3, the predecessor
models had smaller distances of  the torsion systems .

Illustration 6
Torsion stand as exploded view. This
hypothetical construction is laid out from the front
in on a clean, cylindrical lining the system.



Illustration 7

Clamping frame without a possibility to change for the torsion stands.
(See also menu  "Reconstruction").

The washers and  locking bolts

The washers were manufactured from brass. They were not as the original washers cast -, but manufactured mechanically. The treatment took place "roughly “with large tolerances. Thereby the casting was copied by turning of radii and bevels. The pitch diameters for the bolts were drawn by a hand drilling machine. The cut outs for the locking bolts would have sawn out and were able to be filed.  One milled and one over filed afterwards for time reasons. The locking bolts forged in the original were milled and over filed afterwards.



Illustration 8

Washers, locking bolts and the torsion rope.



Illustration 9
Washer, locking bolt and put-in bolt

The clamping key

Around the catapult link on the upper and lower washers must be rotated opposite the torsion stands and locked by means of put-in bolts. For this a clamping key was made. To transfer the strong forces it was clear that the key had to attach positively at the sockets. It would have been offered by means of a tap in form of a hook spanner in the washers to intervene. As the most elegant solution it appeared concerning the setting of the key to use the locking bolts to use. The metal part of the keys received an eyelet by means of the keys to end of a locking bolt was inserted. At the other end a fork attached with the clamping key was engaged when pressing down into the other end of the locking bolt. The metal part of the key received an eyelet at one end to be hung up on the hook.

Illustration 10

Catapult no. 3

Clamping key in „mounted“ position.

The catapult arms

As material for the catapult arms first beech (catapult 1 and 2) was chosen and later ash was selected (catapult 2 and 3). This was stronger than beech.


For safety reasons the arms were always implemented oversized
. Particularly with externally working catapult systems the break of a catapult arm can have devastating effects. The arm is away hurled outward, arc shaped and thereby takes the point of hanging up of the chord as turning point. There is here fore for standing around persons a substantial danger of injury.

There the mass inertia of the clamping arms substantially affects the firing behaviour, at least with external catapult systems . The mass of the arms must to moved, the energy which must be spent for this, cannot any more be transferred to the projectile. With catapult no. 2 for this reason the distance of the torsion systems was increased. This leads with continuous string length to the shortening of the catapult arms and thus to a smaller weight of the same. Unfortunately this measure proved to be fallacy that catapult was less efficient than before. In the next step a change was then made to an internally working catapult system.

Interestingly the weight of the catapult arms does not play a large role with an internal, interior swivelling catapult system. This was proven by computational simulation and practical attempts. (See also: "Simulation technical evaluation of ancient catapult systems")


Illustration 11 and 12

Catapult no. 3

Catapult arms in mounted position.

With the first two catapult variants with conventional, external catapult system could - during appropriate design of the arms - omit an iron reinforcing. Using the variant with the internal system, the load is much larger. So catapult arms of the previous version broke after the 6th shot. Only a reinforcement with iron bounds led to a sufficient durability.




Illustration 13

Catapult no. 2  with external catapult system.

Distance of the chord bundles 600 mm.



Illustration 14
Catapult No. 2 catapult arms out of beech wood
approx. 450 mm long. Despite recesses for the
rope bundles at the end, the construction in

practical proved employment as permanently break proof.

The string

Similar to the covering of the torsion systems a modern material from the bow builders (Dacron) had to be used . Bow builders who make ancient bows to copy, were however afraid of  the risk of the enormous forces the one string for a torsion catapult must bear must. After the conversion to an internal catapult system the previous supplier of Dacron bow- strings also because the enormous stresses - was no longer able to build strings . Inevitably, a self- contained solution had to be sought and was found. Fishing line at first was used which had the desired characteristics like small stretch and durability. One wound it onto a simple wooden device.

The stretching slider

After an analysis of the representations on the Trajan's Column these weapon systems had a stretching slider. These systems are mentioned again and again also in the ancient literature. Deviating from it the catapult 1-3 were equipped with another system. This does not correspond to the well known ancient models.

Technically seen this long, heavy and bulky slider is unreasonable. Is it far more easier is it to guide the catch in a sawed in groove in the stock. A large advantage is that only the relay crucial component namely the closer and not further the great mass of the wooden slider need to be moved.


Illustration 15
Stretching system without slider, which takes
place guidance in a groove in the stock.
In the back the linear "hanging thing bolting
device system" with 2 pawls


As a locking a linear system was selected, the slider engages gradually at a metal border with riveted lock bolts. If the handles accumulate on the notched pins with stretching, these are raised by the diagonal approach surface and fall - with advanced clamping stroke - into the free gaps, in order to then lock at the pins. If the slider is pushed forward, one could fold the two handles forward and embraces both with the hand.



Illustration 16
The centric guide groove for the slider in the shank is good to recognize.

Stock and clamping crank

The stock made out of oak wood. The arrow gutter is semicircular and implemented into the top side. A sawed slot serves for the lateral guidance of the slider. At the rear end in an iron timber construction a hasp wheel was installed. Into this a double armed pivot can be slid. This lever can be removed for transport or by a shot.



Illustration 17

Linear bolting system with roll and crank.


In practice this principle with our catapult has amazingly worked well and satisfactorily and permitted us with a two men catapult operation a rate of fire of 6 shot per minute, such a firing power would have never been attainable with a system with a slider .

Note: Systems without slider for the antiquary have not yet been demonstrated..

Reconstruction attempt of a linear locking system with stretching slider.

Illustration 18 shows the concept of a locking system which is installed on a stretching slider. The function is implemented with the system, altogether without slider.


Illustration 18
With the exception of a "dove-tail guide"  Moving this slider
corresponds to the design
of the executed catapult design .


Due the construction dependently the clamping stroke is larger with external catapult systems than with an internal catapult system. The chord stands in a state of rest before the clamping frame must be pulled by this through to the rear. There it easily comes to a clamping stroke of 1,1 - 1,2m.

The slider would come including remaining arrow guidance gutter before the string (at least arrow length with 45 cm) and the dimension of the catch on a length of 1,6 m.  In the advanced condition the slider would stand out approx. a meter over the catapult and the stock must be accordingly extended . An extremely unmanageable affair and also a weight problem. Estimated such a slider might weight easily 6-8kg.

Insignificant mass would not have to be moved . Even if the stock was extended for the purpose of the basic function this gave a disproportionately large catapult construction. In addition,  this system is prone to jam and the wooden parts can be forged , which can prevent a very difficult sliding operation.



Illustration 19

Slider in front position.

Duncan Cambell notes in his description of the Hatra ballista already to this connection:


Duncan B.Campell, Greek and Roman Artillery 399 BC-AD 363 Orsprey Publishing , New Vanguard, 2003,

ISBN 1-84176-634-8


“The wide set springs and empty frame could have had no other purpose than allow interior- swinging arms, an arrangement which has certain implications for the design of the stock. For example, the forward swing of the arms requires that the stock must project well beyond the front oft the spring- frame, in order to support the slider while the slider trigger catches the bowstring.” 


Despite these disadvantages all ancient catapults - at least after the present state of research - seem to have had slider systems.

The graphic reconstruction (menu option 2) is implemented in such a way. For a planned revision of the cannon model (No. 4) a slider valve construction is intended.

The catapult stand

Catapult 2  (see illustration 2)

To set up the catapult now another condition device had to be manufactured. Lateral swivelling and an elevator arranging should be  possible. We decided ourselves for a tripod construction, which provides high stability and which corresponds to the catapults depicted on Trajan's Column . Wood was selected as material . The catapult was inserted into an iron fork with cross holes and held by a pin, which is struck by the two iron latches of the fork and in the catapult. The pins are easily solvable, so that the bolt firer in a two part load can be transported. The iron fork is stored mobile  with a centre section of the tripod, with which swivelling of the catapult is possible with standing tripod.

Underneath the stand has a mobile edition bar with a mortised slot division over one laterally and in the height. In this resting a support intervenes, which is riveted mobile at the rear part of the arrow rail. So the direction of height can be made through support of the shift. Catapult no. 3 (see illustration 20)

To improve mobility the stand was shortened again and built on and set onto a framework with 2 wheels with spokes. The catapult is thereby only fastened only with 2 iron clips on the wheel set and can be removed. This execution does not correspond to the ancient execution.



Illustration 20
Catapult No.3 (present stage of development)
The wheel set does not correspond to the ancient original.
The arrow projectiles

The projectile pins were reconstructed according to ancient models. The shank and the wings consist of wood, the arrow-heads are forged of iron. Length approx. 45-50 cm, weight approx. 350g.



Illustration 21

Projectile pin in firing position

Constructional caused old projectile pins for the internal catapult system had to be reworked, there in the strained condition of the string angles - by the catapult arms lying now inside - were clearly smaller and affected the string of the wings. Newly manufactured arrows received an extended, flattened shank at the end and approx. 15 mm of far forward transferred wings. The problems with the small chord angle were solved.


Illustration 22
Projectile pins. Above and centre approx. 50 cm long, down approx. 30 cm.

In illustration 22 the tapered wings are to be recognized by the rear arrow. This was a necessary adjustment to the catapult with an internal catapult system. The middle arrow received wing set in front and an extended shaft end to far and they let themselves also begin with fully strained catapult and small string angle.

The lower bolt opposite the original findings - double the amount increases. These projectiles were supposed with „manuballisten “- small portable torsion arm breasts - in use. While the two arrows with a wing tail unit flew stable, the lower arrow with its cone tail unit was commuting during the flight. Presumably it was too heavy and reached only ranges of 40 - 50m. Here we planned the point to be smaller, what might not only lead to a weight reduction, but also - during an adjustment of the front shank - to a better stabilizing incident flow of the rear stabilization cone.

Illustration  23

Short bolt with feathers.

Fig. 23 shows the lower bolt from illustration 20 with reduced point. The shank is now in front more thinly worked. The length remained alike. For stabilization for a test feather set are attached.

Fire projectiles

Normal projectiles unfold their effect on the target by their kinetic energy. Fire projectiles serve with inflammable materials attached to set fire. Instead of a point with these arrows a fire basket is attached.
The fire basket is attached with oil, bitumen or pitch soaked rags or other inflammable materials.

lllustration 24

Reconstructed fire-arrow with a fire basket at the top.

Through the air stream compounds were distinguished during the flight. With additives of sulphur, iron splinters, lime and saltpetre this was prevented

Illustration  25 and 26

Fire basket without and with incendiary compound.

New preparation of projectile pins for an internal catapult system.

The arrows are appropriate for the employment for an internal catapult system. The wings were therefore shifted further forward. The length of the two flattenings for the clamping claw is doubled.

Illustration 27
Treatment of the shaft ends with the wings.

Illustration 27 shows the work.

From right to left: Make the flattenings for the chord clamping claw.

Train the slots for the wings.

Preparation of the wings.

Loops of the wing edges and assembly.

Illustration 28

Conical turned shaft end for the sleeve of the arrow-head.

Illustration 29
Left place, old bolts with changed wings.  Right place, arrows of new production.

View:  New building 2009

It is a new catapult as " carroballista" with car to be built. The weapon is to receive a slider with linear bolt device system.

Slider and stock

Begun with one " carcass" by stock and slider, the parts were manufactured and supplied after design of a carpenter's workshop. Oak wood was selected as material. The guidance was as " gliding seat" with appropriate tolerances laid out.

Illustration 30

Slider and stock installed. The arrow gutter
and the structure for the barrier are still missing.

Illustration 31
Stock and slider. "dove guidance" is good to recognize. The dimensions
correspond  in substantial as determined under "Reconstruction".

Illustration  32

Slider mounted in the stock.

Illustration 33 shows the "dove profile" well and the proportions of stock and slider.

Illustration 33
Slider in the theoretical  necessary foremost position.

Illustration 34 shows - already under " Analysis" and " Reconstruction" addressed problem of the internal catapult system. The slider exceeds approx. 1,2m over the stock and hangs easily diagonally tilted in its guidance. Like already responded, this might be moderated in practice thereby that the chord - under the smaller pre loading of the torsion bundles with the internal catapult system - can withdraw itself quite 20- 30 cm around her at the slider to hang up. Therefore the slider mustn't shown as far in illustration 31.

Illustration 34

Stock end with up-glued cheeks for the winch.



Illustration 35

Stock end slider.

Illustration 36
Stock with mounted slider.

Fig. 36 shows the slider installed in the stock. Underneath the locking block elaboration for the iron rest border of the barrier is to be recognized in the stock. On the slider the wood block for the pawls and for attachment for the bracing cable is up-glued. The drillings for the axle of the pawls and for the transverse pin of the bracing cable are already brought in.

Illustration 37
Slider with trained guidance
utter for the projectile pin.

 Illustration 38
Make the idler pulley out of oakwood.
The first lateral bearing tap is turned.

Illustration 39
The outer contour is finished. Are missing still the turning for the
reinforcing rings and 
the centric drilling for the square drive.

Illustration 40
Test assembly of the idler in the stock.

The role runs in front on wood. The stresses are taken up later in the half-moon-like wood cheeks. The retaining iron - here still bolted - when final assembling with the cheeks are riveted.

Illustration 41
Test assembly of the ratchet ledge with pawl.

Illustration 42 
Parts in the carcass.

Left one sees the locking collars and props for the torsion stands. In the center the slot borders for the ratchet system lie. Right one sees the idler with the retaining iron in front, down one of the pawls with the associated bearing bolt. The metal parts are still white. The parts are later with forging pitch corrosion proof.

Illustration 43
Slot border with pawl, above the retaining claw for the string.

The slot border is fastened by means of two, continuous rivets in the stock. The clevis mounting for the string claw is riveted in the slider. Here are only for a test put in rivets.

Illustration 44
Bolting device system in bird view

Illustration 44, slider with pawl in rear positon. The retaining claw for the string is installed, the release lever is missing still.

Illustration 45
Ttorsion stands.

Illustration 46
3d- representation of the torsion stands (Illustration 1, menu option reconstruction) . 

Illustration 47, which is locking collars with both clamping frames already riveted. The left stand has already its eyes for the under and arched strut. Right: Assembly situation for the eyes before riveting.    

Illustration 47
Torsion stand in side view.

Illustration 48, the recesses for the eyes are to be recognized.

Illustration 48
Torsion stand in side view.

Illustration 48, complete mounted torsion stand. The rievted eyes are well recognized.


Illustration 49
Arched strut with clamps at both ends.


Illustration 50
Lower strut. Existing out of two riveted parts.






Illustration 51
Clamps of the struts in a detail view.





Illustration 52
Clamps in hanged up position.

lllustration 52, shows the props in mounted position. The attachment takes place via hit wedges.





Illustration 53
Lower wedge connection. The rear wedge is not hit yet.





Illustration 54
Clamping frame with mounted chuck and clamping bolt.

Illustration 55
Complete clamping frame.


Illustration 56
Stock end with assigned slider and installed idler. The handle for the slider is mounted.





Illustration 57
Idler with burned square.

Illustration 58
Stock with slider. The attaching brackets for the clamping frame are installed.













Illustration 59
Torsion stand with drawn in bracing ropes.

Illustration 60
The two protection sleeve halves for one torsion  stand.
Right: The front protective sleeve halv with the catapult arm bag.

The protecting sleeves are from brass. After that the cutouts for the retaining eyes and the catapult arm were brought in, the parts were circular rolled.

Illustration 61
The two protection sleeve halves and the torsion stand before mounting.

In the installed condition the two case halves are fixed with the attachtment wedges for the props.












Illustration 62
Right torsion stand with mounted protecting sleeves.







Illustration 63
Right torsion stand. Backside view with with put in catapult arm.






Illustration 64
Right torsion stand comlete mounted with the clamping frame.





Illustration 65
The protecting sleeve is held by the hit wedges.






Illustration 66
Rivet connection between stock and clamping frame.