Construction of the Roman Onager (catapult)

  • Thread starter Zwaf
  • Start date
  • #1
Hi, me and a friend of mine decided to construct a Roman Onager for school.
This is because of some kind of (practical) exam every student has to undertake here in the Netherlands. For example.. some people study the effects of alcohol on the human body and others build a rocket and try to improve it by changing aerodynamics etc.. And we are making a catapult.

I'm sorry if this is in the wrong forumcatagory or something, I couldn't think of a specific one to post it in.
I'll try to explain the problem as good as I can, although my English is not the talk of the town... so might get some trouble using specific scientific terms.

Anyway, as I said before, we are building a Roman Onager or Mangonel, which is some kind of catapult (In case you have no idea where I'm talking about: Image.).
We started off pretty well, construction went as planned. Base was build and so were the poles (which keep the arm from making an angle of 180 degrees, but stopping it at 90).

We then started construction of the rope/tension mechanism. We decided to make 2 wheels on each side with holes in their side to turn them around and we used nylon rope which went through, around a pin, and back (10 times).
This might get complicated so:


Red = Nylon rope
Blue = hole to stick a stick in and twist the wheel around (can be seen in real photo)
Green = hole to stick a stick through, and through the base to secure the wheel from not turning back when tension is reached.

A picture on how it looked for real:

On this one there aren't any ropes yet, and neither is there an arm between them,
so I got another (from the 'inside'):
As you can see, the arm is held in place by the rope around it (10 strings above and 10 below). We made sure the arm was not going to move up to 90 degrees while we twisted the rope for the first time so attached it to some kind of hook.
We got to about 3 full turns with the wheel (each side) before it started to get tough. We touched the rope and tried to twist it a bit with our hands and it did not seem to be so 'strong' yet.
We released the hook that kept the arm down... but nothing happened. The arm was still resting on the base. We discovered it would go though, but only if we gave it a strong push upwards.
We took the machine outside, where we turned the wheels another 1.5 turns (too hard to go further). But still.. the arm was not planning to 'fire'. We had to give it another push to make it go up (of course it required a little less force this time.. because there was more tension in the ropes).

While we were talking we heard some disturbing noises... cracking wood...
We looked at the machine from the back and realized it was curving in.. heavily. The forces of the rope were pulling the base together (11 cm in, from 1.25m to 1.14). We knew this had to do with the extreme forces on the rope as well as the fact that we didn't place a beam in the inside (as you can see on this picture):
Although it might be not so good to see, the wood is bending in the middle, so we placed a support beam there. Also, the arm seems to be up now.. but that's just because we gave it a push

The bending wood wasn't the only problem...
The (iron) pins, that were also experiencing strong forces.. were also bending.. and not just a little:

It appeared there was just not enough force to take the arm up. This might have been caused by the size of it, as well as the weight (data at the end). But even if we made it lighter (of PVC perhaps.. although this material was not available in the early ages) and shorter.. it would still not fire a stone projectile over 100 meters (let alone 250, which the Roman ones could fire).

So now comes the burning question;
How to improve this machine in such as way that it is capable of generating much more force and fire a projectile over 25 meters (required for the research) ?

You might think of ways to improve the entire rope twisting mechanism as well as changing the arm or adding beams (there are no limitations, we got enough time to change things). If you have any questions on the construction or whatever, ask.

Hoping to hear from you,

Answers and Replies

  • #2
Whoa, that's pretty interesting man! I hope some people here can you give some solid advice.

Good luck :)

Also, your English is better than a majority of native English speakers that I have met! Especially since Dutch is drastically different from English, isn't it?
  • #3
the torque energy comes only from the ropes. i suggest not to continue to twist them to infinite (you see the structure has broken) but add more ropes...10 times, 15 times like the pictures on google images.
also, your picture is reversed than the original: the top and bottom woods are inside the left and the right in the real picture, outside in yours, this may have caused the shrink. try to put between the iron pins and the "wheels" a metal disk
  • #4
Gold Member
This has got to be one of the best-documented projects I've ever seen around here. Good job.
I suspect (just a guess) that the elastic properties of nylon or polypropylene aren't suitable to such use. It seems that most of the energy you put in by winding it up is stored longitudinally (the strands naturally 'shorten' as they wrap around each other), without much going into torque. It might have something to do with the nature of long-chain polymers. Have you tried regular fibre rope or leather strips?

edit: Just saw your post, B2P. Good ideas.
This thread might in fact get more response if it's moved to Engineering.

re-edit: The above answer would seem to be in error, since a Good Morning Canada segment showed a team from the Maritimes making a device that's sort of a blend of yours and a crossbow. It looked like poly rope they were using, and it worked.
Last edited:
  • #5
Ooh - a topic near and dear to my heart.

As far as the "epizygis" (the metal rod that the skein rope loop around) bending goes: Threaded rods have only as much resistance to flexing as a rod the same diameter of the inside of the thread, so if the original rod was, say, 1/2" diameter without any thread, the useful diameter is only 3/8" (just a guess, depends on the depth of the thread).

A better choice of epizygis material is a bar with a rectangular cross-section. This gives it strength, yet allows more loops of the rope to pass by it as it goes through the "modiolus" - the disk that strengthens the area under the epizygis while allowing it to be rotated. The trick is to keep the epizygis from flopping over sideways.

Check out the bottom two photos on - a small onager I made a while ago. Notice also the way I filed the middle of the inner edge of the epizygis such that it tapers to a wedge in the middle, which allows room for even more loops.

That modiolus and epizygis design was based largely on the "Onager Junior" - see and "Mista Ballista" hardware - .

If you go the simpler route of using a thick-walled steel pipe for an epizygis, you get a handy-dandy hole down it's length for inserting a crowbar for the pre-tensioning of the rope bundle. The drawback is that the wider diameter of the pipe permits fewer loops of rope to go through a given sized hole in the side frame members.

Frame reinforcement crossbars: It wouldn't hurt to have two crossbars, one close to and in front of (downrange from) the "heel" of the throwing arm and slightly high; and another behind and slightly low. You might even cut semi-circular notches in the middle so as not to interfere with the motion of the throwing arm.
In my model, I only used one crossbar because the maple frame was sufficiently thick for the size of the skein I was using.

As Born2Perform suggests, use many more turns in the skein (rope bundle). It could be a bit shorter too, which means a narrower frame. If you were to plot a graph of torque versus arm position (ie. the angle it makes with respect to the horizontal cocked position) you would find that (obviously) as the arm flies up from the cocked to the end-stop position, the torque decreases.
With a short thick skein, the torque falls off more rapidly that for a long thin skein. This is not necessarily a bad thing, because the implication is that the long thin skein might have much more stored energy that will never be used. If you wanted it to turn something for many complete revolutions (eg. similar to a rubber-band driven propellor), a long thin skein would be good.
For an onager however, the short thick skein will provide that short burst of high torque through the first 90 degrees of travel, but with a bit of reserve so that it's not simply coasting through the last 10 degrees. That's exactly what you want. Actually, for maximum range (of the projectile) you want the projectile to be free of the sling at about 45 degrees from the horizontal, so any torque applied beyond that point is much less important than for the first 45 degrees.

Bear in mind though, that a larger diameter skein means larger holes in the side frame members, and it requires a stiffer epizygis because it's spanning a wider hole with more turns of the rope. Hence the reinforcing crossbars, the steel modiolus, the bar-shaped epizygis, a double thickness of wood in the frame sides between the crossbars, etc.

It's hard to tell the exact scale you're working with in the photos, but a skein like the ones you've shown might be more suitable for a throwing arm maybe 50 to 70 centimeters long, depending on the mass you're going to be using for a projectile.

Choice of rope: Several experienced onager builders that posted to the "Catapult Message Board" (which has by now moved to ) have stated a preference for "diamond braided polyester". Failing that, regular twisted polyester (*not polypropylene*) would do. Nylon is somewhere in between polyester (aka "Dacron") and polypropylene for quality. The jury seemed to be still out on the "more turns with thinner rope" vs. "fewer turns with thicker rope" strategies, if I recall correctly.
Something to stay away from is any rope or cable that has no stretch to it at all. A team on "Junkyard Wars" one time built an onager with a steel cable for the skein, and it had no stretch to speak of. Even though they nearly broke the arm and winch trying to haul it down into the cocked position, it would only travel up about 45 degrees under load. An utter failure.

I would recommend some sort of padded backstop board running vertically from the top crossbar down to near the skein. Otherwise, your throwing arm can snap off near the tip where it strikes the top crossbar. Just imagine that you are swinging a hockey stick as hard as you can in a chopping motion down onto some sort of steel railing. It's going to snap. The vertical board should distribute the shock along the length of the arm, with more of a slap than a chop.
Also add diagonal braces from the tops of the vertical side posts to the front of the frame.

Generally, it's a process of beefing up one component or another, see what breaks, and then redesign or reinforce, and try to break it again.

For throwing arm materials, think hockey sticks, axe handles, sledge hammer handles, skinny baseball bats, etc. You should have some kind of knob or short crossbar on the heel of the throwing arm so that it will not easily slip out of the skein.

It needs to have so much torque and speed that it is quite dangerous to the operators. I've heard of 1-foot long models breaking fingers when something or other failed. Suffice it to say that one must exercise great caution whenever near a cocked onager. Treat them like you would a large powerful firearm.

Make sure bystanders are well away from it (especially in front and rear)while it's being cocked or fired. A full-face lexan shield would be a good idea for the experimental skein tensionings (where the side frame members might snap) and for the firings.

For additional notes on torsion-powered siege engines, check out ballistae, such as this one at . They are, essentially, a pair of onagers arranged base-to-base. Oddly enough, the Romans developed their ballista long before the onager, even though it's a much more complicated bit of engineering.

For construction and operation of high-power "pumpkin-class" onagers, Bob Carbo's the man - see and click on the "construction" links.
Last edited by a moderator:
  • #6
Ohh great, thanks.

- As I've seen on some of the pages and pictures you showed I also realized it would be much better to have an epizygis with a much greater diameter.. therefore it won't bend in as fast as ours did.

- Still got a question about your system here though:

The iron epizygis is 'stuck' inside the 'skein tunnel parts' (as you refer to them). But how are you capable of turning these around ? I guess you just use your fingers to grab the epizygis and turn them both (or do you lift the epizygis out of the gold part and then turn it?) . But anyway, how does this turning work in a scenario with much more force required, like here:
I guess it's not quite as safe (could shoot back) or easy to turn it for let's say.. the 10th time.
I'm not spotting any holes to put a stick or lever through and turn it (like we did) but I guess this won't be possible as well, because it's not as wide as ours is.

- There was no modiolus present in our onager. It was somewhat replaced by the wheel itself (which was so wide). But if we wish to make a epizygis as the ones shown we will need one, not only to make the spinning easier, but also to spread the forces a bit (correct?), meaning it won't hurt to make it as large as the height of the sidebar itself ?

- The crossbars near the rope system... yea we were planning that.. but we were first trying without (kind of stupid huh..).

- I understand the point you are making with the long thin skein instead of the short thick one... going to work this out as well.

- 'padded backstop board' . We are aware of this and were planning to create this, but after we made sure the firing system itself worked.

It's hard to tell the exact scale you're working with in the photos, but a skein like the ones you've shown might be more suitable for a throwing arm maybe 50 to 70 centimeters long, depending on the mass you're going to be using for a projectile.
To what aspect? The fact that it is not twisted around enough or that it is just the wrong kind of rope?

About the scale... I said in the first post I would supply you with data at the end .. but forgot I see, so here it is:
2.5 meters in length, 1.25 width, 1.15 high.
Arm length: 2.5 meters (meaning the entire structure would be around 3.75 meters in length, since the arm is attached in the middle).
I am aware of the fact that the arm is too long, but we have to do an experiment that involves changing the length of the arm..

- Last thing I wanted to ask is if you have any advice on the shape of the arm. For example.. would it be wiser to make a round one, since the skein would then have more 'grip' on the arm at the bottom, or am I missing the point here?

Anyway, thanks for all the help so far. I'll inform my co-Roman with this information and I'll report in later when we made any progress. If it would work you might expect to see a video of some shots :cool: .

Especially since Dutch is drastically different from English, isn't it?
Since it's a Germanic language, it is :biggrin:. And no, this does not mean it's the same as German, if you wish to see how much it differs, try looking at English and French :tongue:

Suggested for: Construction of the Roman Onager (catapult)

  • Last Post
  • Last Post
  • Last Post
  • Last Post