Next Page »


There is a certain sullen satisfaction in the knowledge that another job in the machine shop is drawing to a close. The floor will be swept clean. Its metal chips and spilled cutting fluid fed to the patient garbages cans that sit outside.  Lathes and milling machines and grinders will be wiped down and readied for the next project.  Sometimes I think of all the molds and prototypes, production runs and repairs that this small shop has put out over the last twenty five years, and wonder what they would look like if they could all be piled into a big heap.  Would they crowd out the door and spill into the driveway?  Or would they actually take up a more modest volume, something a few dumpsters could handle?  Either way, there is a lifetime of work tied up in that imaginary heap.  A million details fretted over and resolved.  Customers made happy, and a few who could never be made happy.  Five years ago some heavy metal plates overbalanced the table-saw I was using for a temporary cart.  The whole thing went sideways, and 300 lbs. of steel smashed deep scars into the concrete floor.  Angry at myself, I vowed to patch the concrete.  Of course, I never did.  Now those scars pinpoint a time and project that would otherwise be long forgotten.  Is it only the things that go wrong amid the general clutter of things that go right, that gives our memories form and structure?  I haven’t needed a band aid in quite some while now.  Does this mean my fingers know something that I don’t?  In this game nothing can be taken for granted.  Sometimes there is no preparation or planning or state of janitorial excellence that can prevent a good thing from going bad.  When that happens, the undoing of the thing gone wrong is always more carefully considered the second time around.  By then we are painfully aware of something we knew all along,  the molecules will not tolerate our disrespect.

…….. And so, with these somber reflections in mind, I have decided to use some stainless PH15-5  H-900 for the hooks on the new limbs.  It is not authentic for the Romans, of course, but with a tensile strength of nearly 300 psi, it is 200% stronger than T-1.  Unpleasant punctures from fractured hooks are guaranteed to spoil the data stream, and I am fully committed to getting a long, uninterrupted series of shots before taking this contraption on a field trip.  Choosing the right molecules in the first place is an important survival tactic when you are making high power catapults.  And besides, it is hard to have any fun if you are continually wincing.

I came across this old Anglo-Saxon riddle the other day.  It amuses me to think that this arcane subject has such a visceral root.


  •          ‘I am the defender of my people. 
  • Strengthened with wires and filled with gifts.
  • During the day I spit them forth.
  • The fuller I am the better I am.
  • I swallow dark weapons of war.
  • Bitter arrows and poisonious spears.
  • I have a good stomach.
  • Men seldom forget what passes through my mouth.’


Answer:  Ballista.


Meanwhile, back at the little catapult factory:

We have been struggling with the design for the new set of limbs.  The design shown in the last posting had the entire butt of the limb encased in steel tubing.  Perhaps a tad bit of the old overkill going on there.  It is a personal tendency I have.  Probably I would make the world’s worst airplane designer.  Nothing would ever get off the ground for all the armour plate.  And so,  after much contemplation, I have come to the conclusion that the plan laid out by Mr. Aitor Iriarte in his paper that appears in, The Journal of Roman Military Equipment Studies 11, 2000, is the best approach for the new limbs on our reconstruction of the Orsova machine.  Not only does this design benefit from Mr. Iriarte’s detailed and thoughtful scholarship, but it seems to be a practical and workmanlike solution to a tricky problem.  The drawings Mr. Iriarte presented in that paper are seen below.


The thickest section from our hickory peavy handle, corresponds very closely to the proportions of the tapered cone seen in the woodwork here.  We will use 5/8″ T-1 steel, to form the full length brace and hook.  Our collar will be a bit bigger than the one shown here, and will be situated on the limb to protect it from the impact from the stanchion.  Probably we will also include another lighter collar in the middle of the limb for added strength. The cord whipping at the hook seems like a good idea.  Epoxy soaked, Dacron bowstring thread should do the job nicely.  With a bit of luck, we will be up and shooting again by mid-week. 


The snow has finally started to melt off.  If the mud out on the flats isn’t too bad, it might be time to take a field trip.  Shooting through chronographs and twiddling around with these design issues is all very well, but nothing can match the pure exhilaration of watching a heavy bolt streak out of sight at a high angle.  It is just about time to have some fun.



The full length steel braces and high tensile steel hooks for the bowstring will be installed on the new style limbs in the next few days.    I have got to admit, the curved pocket in the stanchion certainly looks like it was designed to be used with a round section limb.   Because the shape of the pocket was a given, from the original artifact,  designing a limb that fits into it snugly, is a rough indication of the limb diameter they probaby were using.

One thing is for sure, without buffers, anything less than an up-armoured limb like this one would just beat itself to pieces against the narrow impact area of the stanchion.  Those sneaky old Romans must have done something similar. 






















It seems that all of our problems with strings snapping started when the buffers began to break down.  When the buffers become thin or severed like those above, it allows the limb to rotate further into the pocket of the curved stanchion.  The loss of a 1/4″ in the thickness of the buffer can translate into over 2″ of extra movement at the limb tips.  As mentioned in a previous posting, the force vector on an inswinger is in a direct line with the bowstring as the limbs complete their power stroke.  If the limbs over rotate because the buffers are missing, even the strongest bowstring will invariably snap or be severely compromised.  This begs the question, why do we have buffers at all?  I had included them originally out of an excess of caution, it was an attempt to reduce the shock load on the precious field frames.  Perhaps they are not necessary after all.  In a way the kamarion is a natural shock absorber and probably has enough give in the arch to mitigate any excessive shock to the field frames.  Also, the bundles themselves must have some give in them at the end of the power stroke; the limb hits the curved stanchion and rolls on around it as it acts like a  fulcrum.  At this point it seems reasonable to omit the buffers and have the steel brace on the limb contact the curved stanchion directly.  If this allows for a more precise and predictable movement of the limb tips, our loose string strategy will hopefully prevent any further string breakage.

Just for kicks, I tried a shot with a ridiculously loose string.  The string survived just fine, although the bolt did not.  It passed through the corner of the backstop and twirled away into the rocks at a high rate of speed.  The shock absorbent nature of the kamarion’s arch can be seen in the vibrations visible here in 4x slow motion.  The field frames do not appear particularly stressed out by their role as a stopping point for the speeding limbs.  Click for slomo:  ultra-loose

Could it be that the Romans included the arch not just because it provided better visibility, but also because it meant they didn’t have to fuss with pesky buffers?  It seems clear that a reliable and predictable length of travel for the limb tips is essential for gauging how long to make the bowstring.  Without buffers that break down, this should be easy enough to figure out.  Therefore, it may well be that the famous iron arch of  these late model Roman ballista’s, was as much about shock absorbency and protecting the field frames, and ultimately dispensing with fungible buffers that lead to unpredictable stresses on the bowstring, as it was anything else.

In our quest for a suitable backstop material, the rubber mulch used as a safe surface underneath playground equipment, seemed like a reasonable thing to try. 

Click here for some 4X slomo:  rubber-mulch 

Seeing this video we might conclude, perhaps not so much. 

However, several things were flawed in this test.  The lid on the bucket popped off and compression inside the target medium was lost.  The bolt seemed to rise a little higher than expected and did not hit the bucket as squarely as intended.  (In fact it exited the bucket, skimmed along the top of the bales of newspaper and left a healthy divet in the aluminum plate on the back of the cart.)  Also, it again seems I have the bowstring rigged too tightly.  The machine has performed 42 shots in total at the moment.  It has only been during the last dozen of these that strings have been breaking.  It may be that the dacron has been compromised by prior use.  It was left over from the Gallwey project a decade ago and has seen much use lifting any number of things.  New  5/8″ XLS double braid has been ordered and eventually we will get to the bottom of this.

Here we see steel being used as a tension backing on the bottom side of this axle, with enough wood on the top side to adequately resist compression.  If they are balanced one to the other in the correct ratio, this composite strategy yields an immensely strong, yet supple beam.  The above artifacts are from the Molson Museum in Washington State.  They have a fascinating array, of late 19th. and early 20th. century, farm machinery  out there.

Thank you, old hack buggy that is stored under the leaky shed right behind the old Molson bank building.  You have been very useful in suggesting appropriate ratios of wood and iron for our new limbs.



Our newly spliced, double braid Dacron string managed a couple of shots (351.2 fps, 347.5 fps) before  deciding to part from this earth. It was a sad end for such a pretty string. However, much has been learned.  Having practised twenty odd eye splices, I find myself bordering on proficiency at the whole miserable business. It is ironic that it now seems these terrors from the world of cordage are not really all that ideal for our purposes anyway. The strings are not breaking near the ends. They are snapping more towards the middle. The reason for their failure popped into my head in the middle of the night.  Without exposing the reader to the horrors of my dreamworld (populated as it is with writhing eye splices), suffice it to say, the following was revealed with a bit of the old lucid dreaming :

“………The inswinger design puts much more stress on the bowstring than a conventional outswinger. Now, this is only true if you rig the string too tight. Put on a slack string and the blow from the limbs is transferred completely into the frame of the machine. A string that is loose enough, need never experience that awful snapping action that occurs by jerking a fast moving pair of limbs to a complete stop.  With an inswinger the limbs end their power stroke at a force vector that is in a direct line with the bowstring.  This puts much more tensile stress on a  tight string than the acute vector coming from an outswinger. Therefore, loose bowstrings are essential for inswingers………

Armed with this vital intelligence, I found myself embarked on an afternoon filled with bowlines.  Bowlines are far more friendly with this sort of experimentation because their length is so easily altered.  (Timber hitches are out because they would never fit into the nocks.)  After a dozen or so tries I managed to find the right length string and quickly installed it on the machine.  Somehow though, it didn’t feel quite right. The Gallwey string was always nice and twangy, even at rest.  Now the string felt like a slack dog leash rather than a well tuned musical instrument.  Still, I remembered the premonition from my dream and tried to keep an open mind about it all.   We were only able to manage one shot with the loose string because it was our last bolt and it shattered into pieces trying to plunge through the local phone book.  A decisive end for such a worthy projectile.  Before dying,  it generated a comforting 350.9 fps.  This is very close to the other two shots fired with the tight fitting string that had broken earlier in the morning. I had been careful to increase the draw length with the new loose string by a couple of inches to help compensate for its slackness. Doing this gave us 3400 lbs of draw weight, exactly the same as used with the first two shots. All of which seems to imply that a loose string does not much effect the power of the machine, other than perhaps by losing a couple of inches of draw length. Not much problem there, we have plenty to spare with an inswinger.

It is encouraging to note that we now have three more data points that seem to indicate a fairly constant velocity, (351.2 fps, 347.5 fps, and 350.9 fps). It is still not the set of twelve uninterrupted shots I want to see, but I am starting to sense an underlying stability to the machine’s behavior. Better limbs, better strings, more bolts. The answer is in there somewhere.

Seen against the white background and slowed down 4 times, it is possible to just make out the bolt’s flight path as it wallops into a few stacks of the Daily Chronicle:   streak2

The English longbow notwithstanding, bowyers have long known that heavy nocks retard cast.  However, I am starting to think that the idea of keeping the bow tips as light as possible may not be as much of a potential detriment on a 2 ton ballista as it is with the simple hand bow. There is a vital advantage to having a  weapon of war built as ruggedly as possible.  If adding a bit of weight to our ballista limbs has “X” amount of negative effect on the velocity, perhaps the extra strength of new, heavier limbs will allow higher pull weights that would easily eclipse the aforementioned “X”.  Therefore, acting upon this premise, I have ordered a pair of hickory peavey handles. Yes, I know, hickory isn’t period authentic for the Romans. And it is heavy. But the extra margin of safety it will yield seems worth the sacrifice.   In rethinking my dogma on limb design, I am forced to conclude that the more conventional cone shaped limbs with full length steel braces and well formed high tensile hooks, may be the best design alternative for this machine.  By simple good luck, the 5′ peavey handles seem perfectly proportioned for this purpose.   And so, a new set of limbs are in the works.  Inauthentic hickory and all.

………..Meanwhile back at the ranch,

the terror of the double braid eye splice has reemerged from its lair……..

Next Page »