I tried to utilize the web to figure the optimum size ratio for an ancient Roman torsion spring, (e.g. it’s length vs. it’s diameter).   The varying opinions quickly got me disoriented.  It was all a bit of a muddle.

So I’ve decided to use a “cut the chatter” kind of strategy and act from what I know already.  Because Firefly’s springs have proven themselves highly successful over the years, especially with the short limb rotations we are currently exploring, I will use that same spring ratio on Phoenix.  That is:  6.6 times as long as they are wide in diameter.  6.6 to 1, if you will.

Firefly has not had her springs adjusted in the last 4 years and has always shot a very consistent 316 fps, plus or minus 10 fps.  Her power is enough to shoot a 520 gram bolt 800 yards.  I believe that if there are gains to be made in performance they will be found somewhere other than the spring ratio.  “The spring ratio is not to be monkeyed with!”,  says my executive function.  A voice that demands obedience from the other functions that keep this project puttering along.

Developing Firefly’s torsion springs was not an easy process.  When gains are hard-won, one tends to be less flippant in considering potential “improvements”.  Counting one’s blessings can make a lot of sense.

Here is a preview of one of Phoenix’s washers, shown in cross section.



The new washers will be 2  3/4″ inches long.  This will make Phoenix’s springs about 16 1/2 ” long, and with their 2 1/2″ diameter, they will be 6.6 times longer than they are wide.

And so, in a gesture of mathematically precise conservatism, Phoenix’s spring ratio will be exactly the same as her older sibling, because, well … why not?  That’s how pleased I am with the performance generated by Firefly’s spring ratio.  No changes needed for Phoenix in this area.

To the best of my knowledge this catapult project is different from any others you may have heard about.  For a historically defined machine of this class and size, the one you see below is the most comprehensively tested, powerful, long-ranged, and accurate shooting ballista made since ancient times.  No brag, just fact.* Her name is Firefly. This blog is a journal of how she came to be. 

Capture XXA

Click for video: 000142 (2)       Click for another vid:  00017(1)

The impressive work of the Pumpkin Chunkers may seem like an obvious counter to this bold claim.  However,  they are working with an unlimited class of machine that is not concerned with matters of historical fidelity.  This journal details the day to day efforts it took to make a formidable, Fourth Century Roman ballista that did not stray outside the dimensional constraints of the artifacts it is based on.

Like any modern catapult reconstruction, Firefly is not a perfect duplication of the original machine.  Which is hardly possible as there are no intact ancient catapults,  just rusty old bits and pieces, and a few original texts that are inadequate in explaining how these machines achieved their reputedly high performance.

Firefly is in a ongoing state of testing and refinement. Many of you with a bent towards reenactment will no doubt find areas where she is not particularly convincing.  That is to be expected as Firefly is not a machine built primarily for display purposes.  As much as anything, she should be considerd a “proof of concept” shooting machine.  A historical test bed, if you will.   Perhaps polished a little more than is strictly necessary, but that’s a personal issue……

A word about my sometimes flippant tone — I’ve only got so much happy in me for this kind of trial and error marathon.  I find a certain amount of glibness necessary to keep buoyant after so many setbacks. Hopefully it doesn’t belie too much of the actual work.

And with that, here is the continuing saga of Firefly and our new machine, Phoenix…..

Nick Watts,  September 2015


* Believe it or not,  I really would appreciate it if someone would refute this claim.  Numbers, please.  Comments section is open.


What follows is essentially a set of field notes and facetious contemplations.  You will need a sieve.  If this wordy, highly detailed, maze of a blog is not to your taste, please try the following website for a more concise presentation:


The two postings below will provide newcomers a glimpse into what this project is all about. They are a little older and are presented out of sequence. For the latest postings please go to the menu on the right and check out our other categories. Day one of the project defines the historical origins of this device and is easily accessible by clicking in the archives.   Actually reading through the rest of the days — that will take some stamina….



My old friends, Brian and Frank, were on hand for our first successful attempt at shooting a twelve shot group at Firefly’s maximum range.   They took some nice photos of our set up.  Here we see the newly installed Mk. X limbs just before being put into action for the first time.  (Click to enlarge.)

And next, here is a shot of that space blanket reflector we needed for the laser range finder. (Which worked extremely well, we are pleased to report.)

There happens to be a rock outcropping overlooking our firing line and Brian was able to get a good vantage point for this next video.  It takes some sharp eyes to catch the trace of that bolt across the screen, but it is there.  Click for vid:  20120918143324(2)(1)

From atop the rock, Brian and Frank were able to see the entire flight of the bolt out to it’s 800 yard endpoint.  Apparently, witnessing the scale of that arching trajectory in person is quite impressive.  It leaves me a bit jealous as I’m the muggin’s that always has to operate the machinery.  The most I get to see looks something like this:  Click for vid.   20120918144301(1)

But I ain’t complain’.  At least we got some data this time.  These next two charts show how the dispersion of the shots appeared when we finally went downrange to take some measurements.

This first one shows the overall range of the shots in yards and their dispersion to scale.  Bolt weight is 521 grams.  (Click to enlarge)



And this second image shows the dispersion of the shots up close and to scale, with the shots numbered in the sequence in which they were fired, their distance from the firing line, their muzzle velocity,  and the angle that they were found sticking in the ground. (Click to enlarge)



A quick analysis of these 12 shots reveals the following:  Shots numbers 1, 2 & 3 can be discarded when determining the overall group size for the simple reason that the wooden wedge I used to secure the tail of the machine had been inserted from the wrong side and fell out during use.  (Although 2 & 3 do seem to have all fallen very close to one another.  Something is going on here, so maybe the machine didn’t shift position after all.)   Anyway, the wedge issue was easily fixed and shooting resumed without further incident.   Shot number 7 we can call a flyer due to gusting winds at the time of firing.  The remaining 8 shots can be considered valid from a shooting standpoint, and basically fell inside a 60 foot radius.    That is just about enough precision to hit your average three story Mc’Mansion  at almost half a mile.  Not bad, but I know we can do better.

Before going on this field trip I only had time to put four shots through Firefly after the new limb installation.  I suspect the bundles are still moving towards a stable equilibrium and that this is causing some of the velocity variation we are seeing in the above chart. Looking at the shots in numerical order it does seem as if there is a gradual decrease in velocity and range that tends to stabilize at the end of the string.    Unfortunately the chrono showed error readings with the last few shots, but if we consider only the last five consecutive shots it is clear from the chart coordinates that the group size shrinks to something like a 30 foot radius, which seems pretty darn good to me.  I certainly wouldn’t want to be standing inside that circle if some hotshot ballista crew were taking potshots at me from half a mile away.

Today’s tests show that this level of precision would make machines of this type effective against massed adversaries to the full extent of their range.  I have no doubt that with lighter bolts of the Dura pattern,  Firefly’s  maximum range could exceed 1,000 yards. With a bit more tuning the heavier 521 gram bolts we are shooting here should clear 850 yards with ease.  Given how hard it was to extract the bolts from the ground, and considering their average 10 1/2″ of penetration into the packed and dry topsoil, it seems clear they would have been brutally effective on cavalry and lightly armoured troops.

Also of interest:  The shooting session had varying winds from zero to eight knots and was less than ideal for our purposes.  Because time was limited for my two compatriots, we took to the field with a minimum level of tuning on the new Mk. X limbs. There appears to be some asymmetry in the way these limbs are tracking in their planes relative to the deck of the machine, the starboard limb rising perhaps an inch high and the port limb a similar amount low. All fixable by more tuning.  Apart from that, the limbs behaved brilliantly, with no observable curvature at full draw, and sacrificing only perhaps 15 fps from the 1/2 pound increase in their weight from the spreaders and extra bindings etc.  This we should be able to make up, and more, when we final tune the machine.  The average velocity with the 521 gram bolts we are using in this test is 308 feet per second.  All shots were fired with a launch angle of 44 degrees.

And to give some more scale to the proceedings, here is a video with Brian acting as a human laser reflector as we map out the position of the bolts in our group.  Click for vid:  20120918151232(1).   Frank was kind enough to drive back to the truck and retrieve my binoculars that have an internal compass in them.  This made measuring the angles out to each shot a snap, and combined with the laser range readings off Brian’s chest, plotting the locations of these twelve shots only took a few minutes.

In this next video we can see the laser range finder in action as it takes a reading back to the reflector at the firing line.   Click for vid:  20120918150853(1).    The rock outcropping mentioned before is visible on the horizon line of the field we are standing in, just to the right of center screen and at the end of the video.  For reference, that tiny dot to the left of the outcropping  and in the middle of the screen, is the truck and radar reflector.  Can’t see it?  Well it looks a very long way away when seen in person too.

Here is a quick pan across the field of carnage.  click for vid:  20120918153451

And finally, a couple of photos of the new Mk. X limbs sharing the full draw weight of 5,000 lbs.

Note the lack of  bending as compared to those snapshots of the Mk IX’s from a few days ago.  I wonder if it is a good thing that shooting Firefly is not quite as terrifying as it used to be.   No doubt the Catapult Gods will keep me apprised about all that.