From The Barcarii Facebook page
In terms of the equipment used by a sagittarius, the main bow is generally understood to be the composite bow, sometimes referred to as the Hunnic bow, which used asymmetric limbs. The lower limb was the shorter of the two. This composite bow was reinforced with bone or antler laths - 2 on each limb and 2 more in the central grip. Few remains have survived in the archaeological finds and what has is invariably the laths - the wood and other perishable remains having long since rotted away.
These bows were common in the eastern portions of the Roman Empire and had a long tradition in the Nomadic, Syrian, Arab and other cultures. In the west, which did not have a strong bow tradition, the self-bow or ‘arcubus ligneis’ predominated, examples of which survive in the Nydam finds.
These were long bows with asymmetric nocks, sometimes with horn or iron tips or cordage bound about the limb - but not always. These bows were used in the Late Roman army as mainly training bows for the recruits until they were competent enough to progress onto the main composite bow as describe by Vegetius (Book 1, 15).
In this context, it is worth noting that the only Imperial or State fabrica dedicated to the manufacture of bows was located in the west at Pavia, indicating that the eastern portions of the Empire were manufacturing composite bows using local craftsmen in sufficient numbers that the army suppliers could purchase them in volume out with a fabrica need.
The quiver is represented on a number of illustrations, mosaics, carvings and statuary, often in mythological contexts. It is usually a leather tube with perhaps a wooden core, with sometimes a fringed decoration and a cap to protect the arrows from the weather.
The Nydam finds contain a wooden tube made of maple grooved to allow straps or fabric to be wound around for carrying. The position of the quiver is a matter of some debate.
Conventional opinion usually states that quivers were carried hung from the belt whereas back-slung quivers are relegated to mythology and not actually worn as such in warfare. However, various manuscript illustrations and carvings do show back-slung quivers which hint at a more complex situation.
As always, in reality, the needs and demands of the moment will outweigh convention or training and it might be that in a main battle-line or siege quivers were hung from the belt to aid drawing while skirmishing or hunting or on an extended march, they were then slung across the back so as to not impede movement.
The advantages of sagittarii in a naval or riverine context do not need to be overlaboured here. As Vegetius makes clear in his section on naval warfare, missile weapons predominate over ramming prior to boarding.
The images below show a variety of Late Roman archers using the composite bow and self-bow after the Nydam model. The composite bows on show here lack the bone or antler laths and are therefore anachronistic as far as is known. It is possible that the ‘arcubus ligneis’ referred to by Vegetius is not actually a self-bow after the Celtic or Germannic models but instead a lesser composite by lacking the laths and hence merely a wooden bow as understood by the Romans.
Recruits therefore train on this model before progressing to the more powerful Hunnic version. This, however, is speculation. Such a bow would not survive in the archaeological finds and so remains unknown.