Arcuballista: A Late Roman Crossbow

The sole mention of the arcuballista occurs in the work of Vegetius (Epitoma rei militaris) [1]. In book II.15, when describing how the ancient legion was drawn up in battle order, he refers to soldiers who shoot bolts with manuballistae and arcuballistae, terms which are typically translated as catapults and crossbows respectively.

In etymological terms, Latin ballista is taken from the Greek βαλλίστρα ‎(ballístra), itself derived from βάλλω (bállō, “I throw”). So, manuballista (the Latin variant of the Greek cheiroballistra) has the sense of “hand projector”. Most scholars [2][3][4][5] seem to agree that this weapon was a torsion-powered bolt-shooter, but debate on its configuration is effectively divided into two camps: those who accept the translation literally and believe the manuballista was hand-held and, conversely, those who see it as a continuation and technological development of earlier stand-mounted catapults as seen, for example, on Trajan’s Column.

As for the arcuballista, its name suggests it incorporated an arcus (“arch”), but whether this refers to the bow of a non-torsion weapon or to the arched-strut (Latin: arcus ferreus) design of torsion weapons from the 2nd-century AD onwards cannot be unequivocally stated.

However, in Book IV of Epitoma Rei Militaris (op. cit. IV.22), while discussing siege and naval warfare, Vegetius continues to make a clear distinction between the weapons “they used to call ‘scorpions’ what are now called manuballistae” and those such as fustibali (“staff-slings”), arcuballistae (crossbows) and slings. Unfortunately, Vegetius does not describe the latter any further assuming that his contemporary readers would be familiar with their form and function. Weapons such as slings and staff-slings are clearly hand-held weapons however and, by implication, so must arcuballistae. Moreover, by repeatedly referring to manuballistae and arcuballistae separately Vegetius, and other authors such as Arrian, is clear that the latter are different from torsion powered bolt-shooters and stone-throwers. So, if arcuballistae are not torsion-powered, then it seems logical that they were very similar to Mediæval crossbows using flexion bows.

We turn, therefore, to the crossbows depicted on the Gallo-Roman carved stone reliefs (pictured) from Solignac and Saint Marcel. These are obviously not a gastraphetes (“stomach bow”) as they lack the distinctive crescent-shaped stomach rests characteristic of such weapons. Nor is there any sign of winching mechanism for spanning the bows. Are these then the elusive arcuballistae?

From a photograph of the Solignac relief in his article [6], Baatz suggests a plausible reconstruction of an arcuballista:

It is unclear whether the reliefs depict self-bows or composites; both were known and used by the Romans. Any reconstruction of a arcuballista could, quite reasonably, use either. It is unlikely that a steel prod was used as in later medieval crossbows, however.

No known contemporary spanning devices have been discovered, and none are shown on the surviving French reliefs. To span an arcuballista, a sketch in Baatz shows the arcuballistarius placing a foot on the belly of the bow, either side of the stock, and drawing the bowstring by hand. If correct, the draw weight of the arcuballista cannot be so great that the bow cannot be spanned by hand.

A visual assessment of the Saint Marcel relief suggests the stock was between 600 mm and 700 mm in length using the forearm of the arcuballistarius as a cubit measure; the average cubit being 480 mm (0.5 m). Likewise, using the same method of measurement, a bow length of c. 1300 mm (tip to tip), as Baatz' suggests, is plausible.

Using the length of the depicted quiver on the relief from Saint Marcel, Baatz makes the assumption that missiles were of similar length to arrows shot from standard bows.  Furthermore, with the nut placed toward the end of the stock, Baatz also assumes that the draw length was longer than that of later crossbows, and thus longer arrows could be used rather than shorter bolts.

The Solignac relief depicts a similar hunting crossbow from above. This clearly shows a revolving-nut release for the bowstring, together with a groove for the bolt. It proves the arcuballista is the ancestor of all Mediæval crossbows that use the same revolving nut release. Finds of objects resembling revolving nuts have been discovered in Britain and dated to the 5th or 6th century AD. Unsurprisingly, these have been attributed to a form of late Roman crossbow. The author’s reconstruction uses a Mediæval-style trigger bar set beneath the stock to release the nut which freely rotates for the bowstring to begin its power stoke along the stock accelerating a bolt toward its target.

Unlike Mediæval crossbows which tend to have a wholly rectilinear stock or “tiller”, the arcuballista has a unique carved handgrip at the rear. Being so distinctive, it has been conjectured that pulling back on this handgrip, if connected to a straight trigger bar, might have been means to release the revolving nut. Simply pushing the handgrip forward may have engaged said trigger bar into the locking notch cut in the nut. Precisely how this type of mechanism might have operated remains speculative, but one solution is pictured below:

References:

1.  Milner, N.P. (1993), Vegetius: Epitome of Military Science, Liverpool University Press.

2.  Baatz, D. (1978), “Recent finds of ancient artillery”, Britannia 9, pp. 1-17.

3.  Gudea, N., and Baatz, D. (1974) “Teile spätrömischer ballisten aus Gornea und Orşova”, Saalburg Jahrbuch 31, pp. 50-72.

4.  Marsden, E.W. (1971), Greek and Roman Artillery: Technical Treatises, Oxford.

5.  Wilkins, A. (2024), Roman Imperial Artillery, Oxford: Archaeopress, pp. 202-204.

6.  Baatz, D. (1991) “Die Römische Jagdarmbrust”, Archäoligisches Korrospondenzblatt 21, pp. 283-290.