The purpose of the field frames when coupled with washers is to house the torsion springs. In the cheiroballistra the field frames were made from steel, but the Sala field frames were fashioned from bronze (Boube-Piccot 1994) and the Hatra ballista (Baatz 1978) utilized wooden framework for the same purpose. For details see the list of archaeological finds.
Wescher's (1867) edition of the cheiroballistra does not include pictures of the field frames, but Schneider's edition (1906: 154) does. The clearest illustrations are in Wilkins' edition (1995: 18), but those have been traced from colored originals and thus are an interpretation, not the real deal.
There are two field frames in each cheiroballistra, each consisting of one curved bar (ΔΒ and ΗΘ) and one straight bar (ΓΑ and ΕΖ). To the end of these bars two rings are attached at ΚΛ, ΜΝ, ΞΟ and ΠΡ. In codex M's diagram (e.g. Wilkins 1995: 18) the Ζ is clearly in the wrong place - it should be next to Π. This does not affect the interpretation in any way, though.
The center parts of cheiroballistra's outer field frame bars curve outwards. The curved part needs to be made much wider than the rest of the bars or it will bend under the pressure of the torsion spring. The purpose of the curve is to allow the arm to travel a slightly greater arc than it otherwise could. Without the curve the arm could also not point forward, let alone slightly outwards, at rest. This would shorten the length of the bowstring, which would reduce the draw length and thus adversely affect the weapon's energy storage potential. All the archaeologically attested field-frames share these features. The Gornea field frames seem to be unfinished, because the curves are incomplete.
Many of the measurements in the cheiroballistra manuscript are directly derived from the thickness of the field frame bars. Iriarte (2000: 56) argues very convincingly that that vital measurement can be derived from the text itself, and that the thickness of the bars should be between 1/4 and 1/3 dactyls. In my reconstruction where one dactyl is converted to 20 millimeters for simplicity, the bars would end up being 5-6.66mm thick.
Archaeological finds can also be used as a starting point for guessing the thickness of the bars. Baatz (1978: 15) tells us that the Gornea field frames have about 7mm thick bars, and that the rings are about 4mm thick, and that the field frame rings had an inner diameter of 59mm. In the cheiroballistra the inner diameter of the rings is only 37mm. The spring volume of these two weapons is, however, almost identical, because the Gornea field frames are much squatter - for details have a look at the Calibration formulae and archaeological finds article. Of course we can't be sure of exact, original thickness of the Gornea field frames, as the measurements had been taken after derusting. In any case the thickness of the field frame bars in the Gornea field frame seem to indicate that the field frame bar thickness in the cheiroballistra can be in the range outlined by Iriarte (2000: 56).
Practical tests with highly pretensioned nylon springs have shown that 5mm is too little, and that 6.66mm is on the limit of being too weak. When springs are pretensioned very conservatively - in my stretcher that means 250hz - 5mm bars are probably strong enough. However, when the pressure of the springs is doubled (350hz) even 8mm thick, unhardened spring steel bars start bending outwards quite visibly even before the washers are rotated to increase the tension. Not surpringly the curved field frame bars are stiffer and can take the load better, as they're much wider at the center. The best way to combat excessive bending is to make the straight field frame bar thicker, in particular in the middle. In any case a small amount of bend is to be expected and is not detrimental.
What remains to be seen whether sinew cord can store more energy per unit of volume than nylon. If so, then perhaps the same energy levels can be reached with less pretension than with nylon, and the field frames can be slightly weaker with no ill effect.
NOTE: Obsolete diagram removed for now. New one is coming up.
NOTE: Obsolete diagram removed for now. New one is coming up.
Assembling the components Edit
We know the basic idea behind assembling the field-frames from archaeological finds. In the case of cheiroballistra, the dimensions given in the manuscript for the little ladder and the little arch determine in part how the field frames are assembled. Conversely the structure of the field frames affects the measurements and design of the little ladder and its tenons in particular.
The biggest issue with field frame assembly is the exact layout of field frame bars in relation to the field frame rings. The main problem is how the distance between the beams (3.5d) is measured. Here are the two main options:
There are several other variants of the above two options, but none of them really change the big picture.
Without practical tests or experience either one of the above layouts could work equally well. In fact, the one on the right would seem better, because it matches the dimensions for little ladder and the little arch quite well. It also does not assume there is a missing dimension in the manuscript. That said, the manuscript explicitly states that the 3.5d refers to the distance (or interval) between the field frame bars.
However, analyzing the manuscript does not get us very far, because the option on the right does not work in practice. This is because the distance from the inside edges of the field frame bars to the torsion spring is too short at about 0.52d. When the arm is inserted, each half of the spring bundle is pushed outwards by about 0.5d. This in itself could cause the spring cords to get stuck to the field frame bars when the arm is rotated. The real problem manifests itself when the washers are rotated to increase tension: the two halves of the bundle move towards the hoop and end of the cone, increasing the effective diameter of the torsion spring from 1.33d to about 2d. On top of this the back end of the cone needs some extra space (~0.25d) to prevent slippage of spring cords. A bit more (~0.25d) is needed to ensure that the base of the cone does not hit the field frame bars. All this adds up to about 2.5d to ensure correct operation. If the field frame bars were really thin, then there might be just enough space, but practical tests have shown that the bars need to be about 0.4-0.5d thick to withstand the pressure of the torsion spring. An additional issue with this layout is that the field frame bars are placed under the rim of the washer, meaning that the bars would have to be forge welded, or their rivet heads sunk below the surface of the field frame ring.
So, the only real alternative is to place the bars as shown on the diagram on the left. This requires making the field frame ring oval, but oval washers are present in the cheiroballistra manuscript diagrams and in most archaeological finds. This results in a elliptic field frame ring with one axis 2d wide and the other about 2.5d wide.
There seems to be slight error in the manuscript when it comes to the length of the little ladder beams on one hand (26d/24d), and the length of little arch's forked ends (4d/2d) on the other. If we preferred the little arch's dimensions we'd logically have to place the field frame bars at a 2d offset, not 1d like above. Two dactyls offset plus a curve in the field frame bar seems a bit excessive, because it would mean the arms would point significantly to the sides at rest, and practical tests have shown that there's not much, if any, benefit in doing this.