Jorit Wintjes – Julius-Maximilians-Universität Würzburg
When thinking about the spatial impact of the Roman army on the countryside – or to put it more plainly, how the permanent garrison of a Roman army unit looked like – we are used to think in terms of a fort, be it of turf and timber or of stone, and perhaps also a vicus. This is understandable, as buildings, walls and ditches can feature quite prominently in the archaeological record, and as there exists at least in the case of forts – if somewhat scanty – literary evidence to back up that record. As a consequence, the classic “model”, for want of a better word, of a Roman garrison consists of a fort and an extramural settlement, if there is one.
While these two elements (the fort and the settlement) undoubtedly are key components of a Roman garrison, a third component is missing in that model which however was as important as the other two – the extramural space assigned to the keeping of the animals belonging to the unit, a space that must have been quite extensive and was at the same time both properly defined and guarded. The purpose of this brief text is to conduct a small thought experiment based on the cohors I Aelia Dacorum, a unit roughly 800 men strong which for most of its known existence was stationed in Birdoswald/Banna (most of the available evidence for this cohors has been collected by John Spaul, cohors². The evidence for and a short history of the auxiliary infantry units of the Imperial Roman Army (BAR International Series 841), Oxford 2000, 344-347, to which RIB 3438, a dedication to Jupiter datable to the reign of emperor Probus, can be added).
Before taking a closer look at the cohors however it is necessary to stress a point that is as important as it is self-evident (well, or should be): the primary purpose of Roman army units was neither to litter the countryside with impressive buildings so that 2,000 years in the future archaeologists could have a great time, nor causing the emergence of civilian settlements so that barbarians still dwelling miserably in mudholes could eventually see the light of civilization. Of course Roman army units did both, but their primary purpose was to project Roman power to wherever it was needed – in other words, the cohortes and alae dotting the frontiers of the empire had to function as military units, they had to be capable of a fairly wide array of military operations from decidedly low-key policing operations to full-blown, highly kinetic campaigns leading deep into barbarian territory. And while many operational details lie in the dark due to a rather obstinate silence displayed by many of our sources, it is clear that Roman army units, at least for much of the time, did function rather well.
So, although this author firmly believes that the life of the average Roman soldier was far from being one of constant conflict, there can be no doubt that every now and then Roman army units actually did go on operations (and more often than not quite successfully so) – which as a consequence makes the question of how these operations looked like and what impact they had on the life of a garrison a rather important one. Unfortunately, available sources are too few to allow any firm conclusions; however, based on the little that is known about Roman army operations, at least a small number of observations can be made, even if they are somewhat speculative. The present brief article will take a closer look at one aspect of these operations and make some suggestions as to its consequences for the outward appearance of a Roman garrison.
Among the key characteristics of Roman army operations was the mobility of Roman army units. In the case of frontier systems like Hadrian’s Wall or the limites elsewhere throughout the empire, the army did not man the walls to repel any invader, but rather went out of their forts to meet him in the field. Frontier systems, in which much of the exercitus of any province was located, furthermore could also serve as natural springboards for campaigns deep into barbarian territory. In both cases mobility was required on two different levels: tactical mobility included both the capability of moving quickly on the battlefield itself and of outmanoeuvring an enemy in the run-up to a battle, while operational mobility was what gave an army on campaign freedom of movement. Tactical mobility depended mostly on the mobility of the soldiers, while operational mobility depended directly on the flexibility of the logistical support services of a campaign army.
In addition, Roman army units not only needed to have a certain degree of mobility, they also had to be able to march out of the gates of their garrisons at comparatively short notice, carrying supplies for at least a few days with them. And indeed, when looking at the standard plan of a Roman auxiliary fort, the horrea can be interpreted as a testimony to exactly that capability – of carrying out movement without the need for time-consuming preparation (for a dated but still useful overview over form and function of horrea in forts in Roman Britain see Geoffrey Rickman, Roman granaries and store buildings, Cambridge 1971, 213-238). Yet having supplies and ammunition readily available in case the unit was called out was one thing; mobility also meant to be able to move supplies, ammunition etc on short notice. After all, having access to a large amount of supplies was only one half of the problem: if the soldiers had to move into the field, the supplies had to go with them. Which means that, to put it in a slightly theoretical way, any mobility asset required for moving supplies had to be available at short notice as well, as only starting to collect these assets once the order to move out was given would have led to serious delays.
As for the nature of these mobility assets, there is not much information out there; apart from the mules which find mentioning every now and then one could think of other pack animals as well. For the purpose of the present thought experiment, only mules are used for simplicity’s sake; while the employment of other pack animals could certainly lead to changes in detail, it would not have a significant impact on the overall outcome of the experiment.
Now, going back to cohors I Aelia Dacorum in Birdoswald, the first question we have to ask is how large the unit, a miliary cohort of in theory around 800 men, was in reality. If the Vindolanda tablets are any indicator, then we have to assume that not only at least a tenth of the actual strength was unfit for service at any given period, but also that significant parts of the unit might have been posted elsewhere, while the unit itself was unlikely to be at full strength in the first place (See Tab. Vindol. 154 for the strength report of the first cohort of Tungrians, another cohors miliaria). It should be noted, however, that from a mobility point of view the outposting of a large detachment simply meant that the mobility assets which must have been initially available went with that detachment – unlike soldiers detached for example to the personal guard of the provincial legate, which will not have required access to the mobility assets of the unit.
Assuming the cohors I Aelia Dacorum of around 800 men nominal strength to be at only 720 men actual strength, and then removing a further 10% of that actual strength as temporarily unfit for service, and yet another 10% for the outposting of soldiers either individually or in small groups, we arrive at a number of 576 men who were actually present in the fort and for whom suitable mobility assets had to be available. As for the number of those assets, it is usually assumed that each contubernium, or group of eight men, was assigned one mule for carrying the tent as well as the heavier pieces of equipment, while in case of longer operations a second mule carried further supplies. It is unclear how illness and outposting affected the composition of the contubernia, and whether for example tents were shared around as the need arose, or whether contubernia stuck together even if there were only four or five men left; logic would suggest – and economy possibly dictate – that only as many tents were carried on operations as were necessary. For the purpose of the present thought experiment it is therefore assumed that for an operation taking longer than one or two days, sufficient tents were carried to house the soldiers in groups of eight men each. Even in the case of a major campaign, however, the unit is unlikely to have taken everybody able to carry a sword into the field; in order to keep up day-to-day operations, some soldiers will have remained in the fort, thereby slightly reducing the requirement for mobility assets. Assuming that of the 576 men actually present in the fort only 520 actually went out, one would arrive at 65 eight-men-groups, requiring 65 tents to be carried. Adding tents for the unit commander, the centurions and other command personnel one might arrive at around 70 tents in all, requiring 70 mules.
Of course, the transport requirement did not stop at carrying tents and other contubernia-related equipment. While there is not much evidence for the presence of artillery with auxiliary units, what little material exists does suggest that at least some small artillery pieces will have been available (there is for example some inscriptional evidence from the Severan period for ballistarii or artillery workshops in auxiliary forts, see RIB 1280 and 1281 from the Roman fort of High Rochester); even the need of carrying only a very small number together with the necessary ammunition could easily have meant another five mules, resulting in an overall total of 75 mules necessary for getting the unit – or rather 72 percent of its actual strength – on the move.
It should be added that, while the following calculations are based on 75 mules, the number is likely to be too small by quite a margin, as the cohort not only had to carry other equipment, but also supplies for the non-combatant personnel accompanying the soldiers which included but was possibly not limited to the muleteers. Also, in the case of longer campaigns one would expect higher-ranking command personnel to carry at least some sort of creature comfort with them around, further increasing the overall transport requirement. In fact, there are a few notes scattered across the literary record shedding some light on these creature comforts. Thus, Cato the Younger is reported to have been on campaign as a military tribune in 67BC supported by an entourage of 15 slaves including a baker as well as a cook, two freedmen and four friends – which was regarded as laudable example of moderation (see Plu. Cat. Mi. 9.2, see also 12.2). While this is not to suggest that a prefect like the commanding officer of cohors I Aelia Dacorum required similar amenities on campaign to someone commanding a legion, it at least shows that a unit on the move required transportation beyond tents, supplies and ammunition.
Assuming a requirement of 75 mules for moving the unit out into the field at short notice, these mules had to be readily available. While that appears to be – and indeed is – a clear case of stating the obvious, it actually has a number of important implications. First of all, just as the unit itself had a number of soldiers unfit for service, we have to assume that among the mules available to the cohort there were a number of animals temporarily unable to perform their duties as well; if around a tenth of the available mules was unfit for service at any given time, the actual requirement was not for 75 mules, but for 83 mules to which the unit had to have more or less direct access.
83 mules – and that may again seem like stating the obvious – is not exactly a small number. In fact, the number raises several questions, like how were they supplied, how much manure did they produce on a daily basis, what was done with that manure (was it possible to buy something like “army grade” fertilizer based on manure in the extramural settlement?), and in particular, where did they come from, questions which are outside the scope of this small article.
Instead, the following calculations concentrate on another question: where were these 83 mules kept? Of course, it should be stressed again that much of what comes below is speculation, as practically nothing is known about the way the Roman army kept its mules; while two major veterinary texts dealing with mules have survived from antiquity, one by no other than Vegetius, no information on the actual circumstances of keeping them is extant. For the present purpose the author has therefore based his calculations on the Empfehlungen zur Haltung von Eseln (“recommendations on keeping donkeys”) published in 2015 by the Ministry of Agriculture of the state of Lower Saxony, a German state with a long tradition of horse-breeding. While these were written with the modern concept of animal welfare in mind, the Roman army did have a genuine interest in keeping their mules and other pack animals well, as these formed the only mobility asset available to them. So whereas animal welfare may have been alien to the Romans as a general concept, they certainly had a distinct interest in keeping their animals healthy.
Going back to the question where the animals were, there had to be both a block of stables and an open pasture available. While mules could in principle be kept on the latter, bad weather (not entirely unknown in Roman Britain) or illness would have made it necessary to have access to dry shelter. Such a shelter did not have to be a massive building; however, it must have provided both a dry floor and a roof, the requirement of a dry floor making walls on at least three sides necessary as well. Assuming a minimum of 5m² per animal (Empfehlung 2015, 20), stables suitable for 83 mules would have covered an area of 415m² which was roofed over and an additional area of at least two or three times that size in front of the stables where the ground was kept as dry and compact as possible (Empfehlung 2015, 21); in comparison, each of the Birdoswald horrea had a floor size of around 225m², so we are already talking about a significant area reserved for the mules’ stables. It should be added that these stables were most probably quite different to the horrea just mentioned – whereas the latter were stone buildings suitably massive to lend themselves to being turned into a great hall in the sub- or post-Roman world, the former will have been wooden constructions of a much more temporary character. Indeed, it is perfectly possible that instead of one big block of stables accommodation for the mules of the cohors was provided by a row of shacks which could be quickly erected or knocked down depending on the number of animals available to the unit.
These stables were most likely located outside the fort for two reasons. On the one hand the space required for them does not seem to fit into the overall layout of Roman auxiliary forts as we currently believe to know them; one should add the caveat however that our understanding of how a Roman fort actually functioned on a day-to-day basis is still extremely limited. On the other hand, mules also required open space for grazing, on which more below. Even assuming there was space to be found somewhere inside the fort for stabling the mules – perhaps in a “decentralized” way in several smaller sheds than in one big block of stables – there clearly was no space at all for an open pasture. As a result, keeping the mules inside the fort would have made it necessary to drive them regularly to and from the fort, a nuisance which could be easily avoided by locating all the space required for keeping the animals outside the fort.
If we, then, assume that wooden buildings of a semi-permanent character were indeed used to stable the unit’s mules outside the walls of the fort, it should come as no surprise that there is basically no archaeological evidence supporting this assumption. Not only is the extramural space around Roman forts in general much less explored than its intramural counterpart, non-permanent buildings are also much less likely to leave tangible traces in the first place.
If the space taken up by shelters for the unit’s mules was already far from inconsiderable, the area required for the open pasture was even larger. It is nowadays assumed that 500m² are a minimum requirement for a maximum of nine animals, with each additional animal requiring further 50m² (Empfehlung 2015, 22). For 83 mules this results in an overall requirement of 4,200m² open pasture. Together with around more than 800m² for the stables and adjacent areas of hardened ground one arrives at 5,000m² or more – or roughly a quarter of the whole area of the fort of Birdoswald – permanently assigned to the keeping of mules.
While it is in many cases probably easy to find sufficient open space outside a fort to accommodate both stables and an open pasture (in the case of Birdoswald, a recent geophysical survey has shown areas both to the West and to the East of the fort which may have been suitable, see J. A. Biggins/D. J. A. Taylor, Geophysical Survey of the Vicus at Birdoswald Roman Fort, Cumbria, in: Britannia 35, 2004, 159-178), locating the mules and with them a key element of a unit’s operational mobility outside the confines of the fort must have had a number of important implications. First of all, keeping animals which belonged to the army anywhere will invariably have meant putting them under guard in order to prevent them from being stolen. Also, access to then will have been limited to authorized members of the garrison – in an army requiring the production of receipts in quadruplicate it is hardly conceivable that any soldier could turn up at the unit’s stables and simply grab a mule unless he could produce a piece of writing ordering those in charge of the stables to hand over one of the animals. Indeed, given the degree of close control the army apparently exerted over their soldiers’ equipment as evidenced for example by the Carlisle tablets (see eg Tab. Carl. 16), one has to assume that the management of unit mules was done with the same administrative and bureaucratic zeal as that of arms, armour and supplies.
On a practical level this meant that there was probably quite a lot of activity regularly going on among the stables housing the unit’s mules, though administration will have been only one part of it. As was noted above the Roman army may not have cared about animals in terms of animal welfare, but it certainly cared about animals as key providers of operational capability. While much of that care lies totally in the dark, at least veterinarians are attested both in literary sources and in the epigraphic record (for the latter see eg CIL 3 11215, the gravestone of a legionary veterinarian from Carnuntum, or CIL 6 37194, a gravestone for a veterinarian serving among the Praetorian cohorts), and they do also appear in the Vindolanda tablets (see eg Tab. Vindol. 181). In general, care for animals appears to have been much better organized than in many modern armies right down to the end of the 19th c. (see for example the development of the US Army’s veterinary service; for the period from 1861 to 1899 see Leslie G. Huck et al., Military Veterinary Support before and after 1916, in: Leslie G. Huck and Ronald L. Burke (ed.), Military Veterinary Services, Fort Sam Houston 2019, 4-80, 5-8).
Veterinarians themselves were soldiers and counted among the immunes (Dig. 50.6.1) – common soldiers exempt from normal fatigue duties and eligible for extra pay – while most of the other personnel involved from the assistants to the veterinarians right down to stable boys clearing out the manure were most likely non-combatants, in part possibly slaves. While the veterinarians and their assistants will have cared for all animals attached to the unit caring for the mules must have taken up a significant amount of their time, and it is hardly surprising that with the already mentioned veterinary texts there is evidence surviving of what was presumably once a significant corpus of literature on treating animals in use by the army.
Now, it has to be stressed that the above has just been a thought experiment, and Roman army procedures may well have been very different from that required by present-day regulations. Even so, however, it is obvious that the need to provide shelter, pasture as well as facilities for the veterinary care of the animals must have existed, and that the Romans had a distinct interest in keeping their animals in a healthy state, simply because they depended on them for their operations. And while in certain exceptional situations like large-scale campaigns or emergencies a wasteful use of animals may have happened, this cannot have been the case during the day-to-day operations of any Roman army unit, simply because of the cost involved in constantly buying – and training – new animals. Also, the employment of animals in day-to-day operations and the preservation of mobility as a permanent capability cautions against comparisons with well-known 19th c. practices which seem to point towards a wasteful use of animals. While even during multi-year conflicts like the American Civil War between 1861 and 1865 armies could and did use animals in extremely wasteful ways, the US Army losing more than a million horses in the war (Cheryl D. Sofaly et al., Military equine programs, in: Leslie G. Huck and Ronald L. Burke (ed.), Military Veterinary Services, Fort Sam Houston 2019, 207-225, 210; see also Leslie G. Huck et al., Military Veterinary Support before and after 1916, in: Leslie G. Huck and Ronald L. Burke (ed.), Military Veterinary Services, Fort Sam Houston 2019, 4-80, 5-8), even in the case of longer conflicts these were exceptional circumstances. It is obvious that day-to-day operations over the course of several decades would never have been possible unless care was taken to limit the losses among pack animals. Just as present-day army units are supposed to keep at least most of their vehicles serviceable, so the Roman army will have taken care to keep most of their animals in good shape.
The importance of mules (or other pack animals) as mobility assets to the Roman army leads to two conclusions: one, in all Roman army units, mules must have been available in significant numbers so as to guarantee operational mobility. And two, as a consequence, all Roman garrisons must have had an infrastructure capable of supporting the necessary number of animals – which in the case of a fort like Birdoswald meant that an area of considerable size, with one or more temporary but rather large buildings on one or more sides, was set apart for the mules of the cohort. This area must have been a prominent feature not only because of the acoustic and olfactory impression it will invariably have made on anyone visiting the fort, but also because it was undoubtedly fenced in and guarded, either by soldiers or by the muleteers and other non-combatant personnel charged with taking care of the mules. The area reserved for the animals – which in the case of cohortes equitatae or alae will have been significantly larger – has to be seen as a key element of any Roman garrison, an element that so far has seen very little attention, no doubt because it left practically no traces in the archaeological record so far. To put it perhaps slightly pointedly, there was more to a Roman garrison than the fort itself, even if nowadays the fort may be the most tangible trace of it.