Image of a Legionary

Some ramblings on unit transfer in the Roman army

“The order has come” – Some Ramblings On Unit Transfer In The Roman Army

Jorit Wintjes, Würzburg –

I. The villa in Vectis

In 1887, Alfred John Church, professor of Latin at University College London, first published “The Count of the Saxon Shore or The Villa in Vectis”, a historical novel covering the “end” of Roman Britain at the beginning of the 5th c. –or, as he put it in the preface:

“In this story a novel, but, it is hoped, not an improbable, view is taken of an important event—the withdrawal of the legions. This is commonly assigned to the year 410, when the Emperor Honorius formally withdrew the Imperial protection from Britain. But the usurper Constantine had actually removed the British army two years before; and, as he was busied with the conquest of Gaul and Spain for a considerable time after, it is not likely that they were ever sent back.” (Church 1909, iii)

The book turned out to be fairly successful, being reprinted at least three times in 1890, 1900 and 1909 (all quotations in this paper are from the 1909 edition). While Church was a scholar of considerable repute – among other ancient authors he, together with Cambridge scholar Rev. William Jackson Brodribb (1829 – 1905), published translations of the works of Tacitus, which were republished as late as the 1960s – his novel is obviously a product of the age in which it was written, and many descriptions and scenes reflect mid-19th c. thinking about history in general and Roman history in particular. Yet it is in the eighth chapter that one can find a scene that is quite remarkable – and, while unsupported by what one might call ‘solid evidence’, nevertheless not only plausible but in fact quite suitable as a starting point into a small thought experiment about some of the realities of soldierly life in the Roman Empire.
Up to that chapter the novel follows two plot lines, one describing how the usurper Constantine (also known as Constantine III) gained the purple and the other focussing on how a rather amicable chap called Aelius, who happens to have had the position of comes of the Saxon Shore, tried to keep the shores of both Britain and Northern Gaul safe from various Saxon piratical activities. He is described as

“a man of the best Roman type, a man of ‘primitive virtue,’ as the classical writers would have put it, though this virtue had been softened, refined, and purified by civilizing and instructing influences, of which the old Roman heroes—the Fabiuses, the Catos, the Scipios—had known nothing.” (Church 1909, 32)

While Aelius is thus engaged in employing his ‘primitive virtue’ to keep the Saxons at bay, Constantine decides to take his army to the continent – motivated mainly by the need to find something to do for his restless soldiery and by a desire for filling his nearly empty coffers again. Chapter eight then follows one of the centurions to his home, a house built around a small courtyard, where he arrives after a festivity during which the decision to leave Britain is communicated to the unit officers and their centurions. The centurion tries to break the news of the new order to his family – consisting at the time of his wife, two daughters and his mother-in-law – instigating a rather remarkable dialogue:

“‘Our happy days here are over, my dearest,’ said the centurion, drawing his wife to him, and tenderly kissing her, as soon as they were within doors.
‘You mean,’ said she, ‘that the order has come.’
‘Yes,’ he answered, ‘we are to leave as soon as the transports can be collected. The resolution was made to-day and will be announced to the army to-morrow. It is no secret, I suppose, or will not be for long.’
‘And where are we to go?’ cried the elder of the girls […]
The poor centurion changed colour. The girl’s question brought up the difficulty which he knew had to be faced, but which he would gladly have put off as long as he could. ‘We shall go to Gaul, certainly; where I cannot say,’ he answered, after a long pause, and in a hesitating voice.
‘Oh, how delightful!’ cried the girl; ‘exactly the thing that Lucia and I have been longing for. And Rome? Surely we shall go to Rome, father? Are you not glad to hear it, mother? I am sure that we are all tired of this cold, foggy place.’ The mother said nothing. If she did not exactly see the whole of the situation, she had at least an housewife’s horror of a move. The poor father moved uneasily upon his chair.
‘The legion will go,’ he said, ‘but your mother and you——‘ ‘Oh, Lucius,’ cried the poor wife, ‘you do not, cannot mean that we are not to go with you!’ ‘Nothing is settled,’ he replied, ‘it is true; but I am much troubled about it. You might go, though I do not like the idea of your following the camp; but these dear girls—and yet they cannot be separated from you.’”
(Church 1909, 87-88)

Setting aside the rather shocking lack of enthusiasm for country life in southern Britain displayed by the centurion’s younger daughter, as well as the literary quality of the whole scene – which probably had more appeal to a Victorian readership of the late 19th c. than it has to an ‘Elizabethan’ one of the early 21st – this dialogue is striking because it deals with one of those issues that rarely surfaces in the available literary sources, an issue that is still difficult to identify in the archaeological record and consequentially not as well-covered by contemporary scholarship as other aspects of Roman army history: the family life of Roman soldiers and, in this particular example, the impact of unit transfer on the families and other dependants of Roman soldiers. That modern scholarship appears to be about as eager to look closer at the families of Roman soldiers as the centurion’s younger daughter was to enjoy life out in the country – Walter Scheidel once wondered whether thinking about the demographic implications of having the Roman army stationed in a province was a “mission impossible” (Scheidel 2007, 425) – is rather unfortunate, as it is only by taking these families into account that a complete picture of the Roman army can emerge.

II. Troublesome definitions.

Here an important issue arises – what exactly is a Roman army unit? On the face of it, the question is easily answered. The current US DoD Dictionary of Military and Associated Terms defines ‘unit’ as “any military element whose structure is prescribed by competent authority” (DMAT 15-12-2016, 244; a military historian can’t help but have second thoughts at the reference to “competent authority” …); taking this definition to the Roman army produces the obvious result: a Roman army unit is a body of soldiers with a command structure and at least some semblance of organisation – which not only includes legiones, cohortes and alae, but also, in many cases at least, vexillationes. Yet while this definition is without doubt very useful, and while on campaign or on the battlefield one would understand a unit to be exactly that, looking at the problem from a different angle reveals that there is more to the unit than merely a chain of command, a table of organisation and cartloads of equipment.
Indeed, if one goes beyond the definition given above which essentially is one of military function and instead tries to understand a ‘unit’ as a social phenomenon, for want of a better word, the picture changes rather dramatically. While on a battlefield a ‘unit’ is indeed a group of – in antiquity – men burdened with military equipment and led by someone with authority and – hopefully – at least a vague idea of what there was to do, outside combat situations a ‘unit’ should be more broadly defined as a body of soldiers closely connected to each other both administratively and socially. Thus, apart from the soldiers and non-combattants tasked with supporting the soldiers in their day-to-day work, the families and dependants of the unit members invariably come into view.
Now, it is important to stress at this point that this is not to say that civilians ever belonged to Roman army units in any organisational or legal sense. They did not, and all the legal evidence that is out there is very clear about that – Roman military law was for soldiers, and for soldiers only, as made abundantly clear by Isidor of Sevilla’s famous definition of what the ius militare was:

“(1) ius militare est belli inferendi sollemnitas, foederis faciendi nexus, signo dato egressio in hostem vel commissio. Item signo dato receptio; item flagitii militaris disciplina, si locus deseratur; item stipendiorum modus, dignitatum gradus, praemiorum honor, veluti cum corona vel torques donantur. (2) Item praedae decisio, et pro personarum qualitatibus et labori iusta divisio; item principis portio.”
“(1) ius militare is the formalised practice of waging war, the bond of making treaties, the marching against or engagement with the foe at a given signal. Also the cessation of hostilities at a given signal, and the military discipline for the disgrace of deserting one’s post; also the method of distributing military pay, the hierarchy of ranks, the honour of rewards, as when a crown or torques are given. (2) Also the distribution of booty, and its equitable division based on the status and deeds of the individuals; also the leader’s share.”
(Isid. 5.7.1-2; tr. Barney et al.)

Military law was about fighting wars, preserving discipline by meting out harsh justice, promotions and rewards, and pay. In all this, civilians clearly had no place. And on campaign – and even more obvious on the battlefield – a “unit” was a body of soldiers.
Yet, the realities of life being what they were, soldiers spent most of their time neither on campaign nor on the battlefield; in fact, one could argue that many Roman soldiers in the early or high principate rarely saw “real” combat at all – discharge inscriptions like CIL 3 8110 listing comparatively large numbers of veterans have been suggested to support the notion of a quiet life on the frontier (see eg Scheidel 2007, 428; for a slightly different interpretation of what at the end of the day is a very small amount of evidence see eg Alston 1995, 44-48), but in and around their garrisons. And there, family members and other dependants not only played a role in everyday army life, everyday army life was in fact shaped to a large extent by them. At the same time, any significant change in the routines of everyday life affected not only the soldiers, but also their dependants; this is most obvious in cases when whole units got transferred from one garrison to another, something that occurred not infrequently.

III. All the emperor’s horses and all the emperor’s men – and women, and children, too…

So what exactly, then, did a permanent transfer of a unit from, say, a legionary camp in Britain to one on the Rhine or Danube frontier, actually mean? For an answer to this question, it is necessary to look at the numbers – how many people were actually affected by such a transfer? Or, in other words, when talking about a Roman army unit, it is still rather common to use its fighting strength as the basis for all calculations. While this approach may be useful for calculating the strength of an army on the battlefield, it rapidly loses usefulness when it comes to thinking about permanently moving the unit to a new garrison. Instead, when thinking about the overall number of people involved with a Roman army unit – something the fancifully-minded might call the “demographic footprint” of said unit – those men standing on the battlefield when the balloon goes up, those comprising the real fighting strength of the unit constitute but one part of an overall total that included several other sizeable parts. Thus, in addition to those fit for fighting, any unit will have had soldiers on strength which would have to be considered not part of the fighting strength either temporarily or permanently; one might think of office personnel in the latter case and sick and wounded soldiers in the former. Then any unit would have had a significant amount of non-combatants directly attached to it; these, while not soldiers and – presumably – not falling under the ius militare (the ancient legal literature, as far as it survives, is silent on them), were nevertheless essential to the proper functioning of any unit (for a detailed discussion see eg Phang 2005). They included the muliones caring for the mules used everywhere in the Roman army, as well as the grooms and stable boys found in every unit with horses on strength. A third important group included the non-combatant relatives of the soldiers, ie their family members. And finally, any other dependants like personal slaves belonging to the households of Roman soldiers formed a fourth important group which has to be counted in as well.
Which means that the concerned observer realizes how all of a sudden any confidence about the actual numbers involved is beating a hasty retreat out of the nearest window – while the theoretical fighting strength of Roman army units can be determined with some precision, it is but one of five elements (fighting soldiers, soldiers on non-combatant duty, non-combatants, family members, other dependants) that together form the “demographic footprint” of a unit, and for the other four elements hard evidence is woefully absent. Even so, however, it might be useful to at least try to estimate the overall number of personnel directly involved with a Roman formation in order to get a vague idea of the dimension of the problem of permanently moving Roman army units from one province of the empire to another.
Taking a Roman legion of around 5,000 men as an example one might first of all assume that of its total strength, only around 80% were actually fighting men, while the rest was either unfit for service or involved in running the camp, employed on office duties and the like; while Rorke’s-Drift-esque circumstances might make it necessary for everybody capable of holding a gladius to throw himself at the enemy – or propping himself up against a wall and waiting for the enemy to throw themselves against him – one can safely assume that usually neither the sick and wounded nor the office personnel went out to join the fighting (for an in-depth discussion of the nominal and actual strength of a Roman legion see Roth 1994).
The number of non-combatants directly involved with the unit is much more difficult to estimate. For the army of the principate, one could fairly safely assume that for each contubernium, there was one mulio, and then add further muliones and other non-combatants for duties like horse-keeping duties. Numbers are notoriously difficult to come by, particularly as the composition of the artillery arm of the Roman legion is poorly understood, but it has been variously suggested that around 1,000 non-combatants were part of a legion of the early and high empire (see eg Junkelmann 2003, 94-95). Together with the legionaries the overall number of combatant and non-combatant personnel of the legion would therefore total at around 6,000 men.
Also difficult – if not impossible – to estimate with any precision is the number of the soldiers’ family members. While it is reasonably clear that late Roman soldiers could and did marry, the marriage ban originally introduced by Augustus being lifted by Septimius Severus (on the ban see Phang 2001, 22-85), in legal theory soldiers of the principate were not allowed to have families, or rather legal offsprings. Going by the reality of the auxiliary units however one has to assume that a significant number of the soldiers actually did have families and dependants, for which there is a fair amount of epigraphic evidence surviving (see Phang 2001, 142-196). It is nearly impossible to estimate how many of the soldiers of the legions actually had families; it seems plausible though that young recruits first served for several years before starting to raise a family, and there may have been huge differences between individual units. For the purpose of the present calculation it is – somwhat arbitrarily, it has to be admitted – assumed that one fifth of the legionaries, or around 1,000 men, actually had families; it may well have been more than that.
Of course, identifying the soldiers having families – or quasi-familial relationships – is one thing, the actual numbers involved are quite another matter. Again lacking useful source material of any kind, gauging the number of spouses, children and other family members living with a soldier of Roman army is a matter of pure guesswork. In the case of Church’s late Roman centurion, his household had four other family members – his wife, two children and his mother-in-law. Taking these numbers as a starting point, a ratio of four family members for each soldier does not seem to be implausible – if the soldiers had a family in the first place of course. For a legion of 5,000 soldiers, of which 1,000 men had a family, this would have resulted in around 4,000 civilians directly ‘attached’ to the unit; the unit’s non-combatants may have had personal dependants as well, but these are at present left out of the calculation.
Finally there are other dependants of the soldiers and their families. In his novel, Church is initially silent on any slaves or servants in the household of the centurion – although later at least one other old servant is mentioned –, yet obviously they must have existed. Getting anything remotely resembling certainty into estimating numbers here is impossible. Even so, numbers must have been considerable; assuming there were on average two slaves or servants per household – while quite a few households will have had none, centurions and officers will have had several –, another 2,000 persons would have to be added to the overall total (again, any dependants they may have had are left out of the calculation).
Taking these numbers together, a picture emerges that differs considerably from the general idea of a 5,000-strong Roman legion: where there is a unit of 5,000 men fighting strength, there are in fact an additional 7,000 persons directly and permanently attached to it, resulting in an overall total of no less than around 12,000 – or well over twice as many persons as the “nominal strength” normally associated with a Roman legion (see also table 1). Now, it has to be emphasised that this is mostly based on guesswork and that it may therefore be off the mark quite a bit; and, as noted above, it will most certainly have differed considerably from unit to unit. Yet however imprecise the numbers actually are, it is clear that the overall number of persons constituting what has been termed above the “demographic footprint” of a Roman legion must have been considerably larger than its fighting or even its nominal strength.

IV. Moving the unit – and everybody associated with it.

This raises several issues of varying degrees nastiness, so to speak – because they all demonstrate that precious little information survives about how everyday life in a Roman garrison actually looked like, and how it was organised. To begin with, at least a significant part of the non-combatants attached to a unit will have lived near the garrison, even if some may have lived elsewhere in small towns or rural communities; another possibility is the presence of at least some of the soldiers’ families inside the forts, where the commanding officers had their households, though evidence is at present limited to archaeological material surfacing in several sites. Assuming the majority of the soldiers’ dependants lived in the vicus – which is usually done, though the actual evidence for large-scale vici (assuming that only half of the soldier households estimated above were located in the vici would result in around 500 dwellings for a legionary vicus and around 50 for one of an average auxiliary cohort) is not exactly plentiful (for a comprehensive discussion see Sommer 1988) –, leaving them in place while the unit was moved to a different part of the empire does not appear to be in any way practical, particularly not if a new garrison was to arrive. The latter would have to live with the strange situation that the vicus immediately outside its gates was populated by the families and dependants of its predecessor, while the old garrison would have to put up with its families stuck in a different part of the empire; one could think of arrangements more conducive to the preservation of troop morale.
Common sense suggests that, particularly in cases when new provinces were garrisoned or large-scale reorganisations of provincial exercitus took place, families and dependants clearly had to move as well, which raises several key questions. The move had to be organised, and while it is difficult to imagine the Roman army being officially involved in moving the households of men who were legally supposed to be unmarried, it is likewise difficult to imagine it would have turned a totally blind eye to the plight the soldiers’ families as that could have had a significant impact on the moral of the soldiers. And even if ordinary soldiers may have been left to their own devices when it came to moving their families, it is hardly conceivable that the same was the case with centurions and other key personnel. Accordingly, even if moving the families and dependants of the soldiers may not have been an ‘official’ operation, it is very likely that army logistics had a hand in it.
Of course, simply moving the units was just one part of the problem. The next key issue was finding suitable accommodation for at least some of the households near the new garrison. Again nothing specific is known about how this was done, though houses in vici outside Roman forts usually show little difference to fort buildings with regard to the general quality of the construction, which could suggest that the same builders were at work. It is easily conceivable that in the case of newly established garrisons, the soldiers responsible for building the fort simply extended the operation to erecting buildings in the vicus as well, though hard evidence for this is sadly lacking. One would also like to know how the distribution of housing was organised – particularly if for some reason an existing vicus could not accommodate all households attached to the newly arriving unit –, but again no information whatsoever survives.
Assuming that the army was involved in providing accommodation directly leads to yet another issue which lies in the dark due to the utter lack of information. Assuming the houses in the vici were indeed built by the units and used for accommodating the families of those stationed inside the fort gates, then who was the legal owner of the houses? If the army had originally built them, would it have been possible to acquire them from the army, or would the army have retained ownership of the houses, renting them out to the families of soldiers? Were the vici true civilian settlements where house owners could make changes to their property as they saw fit, or were the vici more on the lines of civilian quarters inside a military cantonment? Behind this looms the larger issue of how (and whether at all) the Roman army interacted as a provider of goods and services to civilian society; thus it is unknown, to take but one example, what happened to surplus equipment and provisions left over after the end of major campaigns; while perishable goods will probably have been used up, several scenarios are possible for larger pieces of equipment like transport ships, including selling them off to civilians. Other examples like tiles produced by the Roman army and found in civilian contexts may likewise suggest that surplus production was sold off to civilian users, though any positive evidence for such transactions is again lacking.

V. Conclusion.

The results of the small thought experiment sketch out above are somewhat disheartening as they raises several questions which cannot be answered at the moment for want of suitable evidence. At the same time they show that at present, the nature of every day life in the Roman army lies still largely in the dark.
Even so, the calculations above, which, it has to be stressed again, are mere estimates and may be off by quite a margin, allow at least for an important observation in principle – a surprisingly large percentage of the total population of a Roman frontier province was in one way or another directly connected to the army. Taking Roman Britain as an example, one arrives – depending on the estimates for the overall population, which vary considerably – at up to one seventh of the overall population which was at any time directly attached to the army. Every new generation of soldiers further added to this number as did every round of discharges from the army, which is usually assumed at around 120 veterans per legion per year (see eg Scheidel 206, 426-427); for Roman Britain this would have meant an influx of ex-army personnel of close to 500 men per year during the early principate, most of whom will have had families and dependants – though it is unclear how many of them actually stayed in Britain once discharged from service. The numbers calculated above also exclude all those who, while not being directly attached to a Roman army unit, were economically dependant in one way or another from the army. Whether they would have followed an individual unit when it moved is debatable; they would however likely follow the army if the whole exercitus of a province was removed. Thus taking all persons attached to the army in their various ways together one would probably arrive at a sigificant part of the provincial population.
Moving them was a major exercise then, and going back to the supposed end of Roman Britain covered by Church’s novel, one might end on another thought experiment – a total evacuation of the British provinces would probably have included moving a large number of civilians as well, something about which all available sources are utterly silent. In fatc, the only comparable process would have been the evacuation of Dacia under Aurelian – where there is in fact evidence for Roman civilians leaving the province:

“Provinciam Daciam, quam Traianus ultra Danubium fecerat, intermisit, vastato omni Illyrico et Moesia, desperans eam posse retinere, abductosque Romanos ex urbibus et agris Daciae in media Moesia collocavit appellavitque eam Daciam, quae nunc duas Moesias dividit et est in dextra Danubio in mare fluenti, cum antea fuerit in laeva.”
“The province of Dacia, which Trajan had formed beyond the Danube, he gave up, despairing, after all Illyricum and Moesia had been depopulated, of being able to retain it. The Roman citizens, removed from the town and lands of Dacia, he settled in the interior of Moesia, calling that Dacia which now divides the two Moesiae, and which is on the right hand of the Danube as it runs to the sea, whereas Dacia was previously on the left.
(Eutr. 9.15, tr. Watson).”

It is quite likely that the majority of the abducti Romani mentioned by Eutropius were in fact the families and dependants of the Roman soldiers of the Dacian exercitus.


Table 1

“demographic footprint” of a Roman legion (2nd c. AD).
fighting strength5,000
of which with families1,000
non-combatants (unit)1,000
family members4,000
other dependants2,000
nominal strength5,000

Table 2

“demographic footprint” of the British exercitus (2nd c. AD).
fighting strength55,000
of which with families11,000
non-combatants (unit)11,000
family members44,000
other dependants22,000
nominal strength55,000

Table 3

The British exercitus and the population of the province (2nd c. AD).
fighting strength55,000
demographic footprint132,000
overall population (low estimate)1,000,000
Overall population (high estimate)4,000,000
percentage of military personnel
(low to high population estimate)
13.2% to 3.3%


Works cited.

Richard Alston. 1995. Soldier and Society in Roman Egypt: A Social History. London: Routledge.
Alfred J. Church. 1909. The Count of the Saxon Shore. London: Seeley & Co.
Department of Defence. 2016. DOD Dictionary of Military and Associated Terms as of 15 December 2016. Washington.
Marcus Junkelmann. 2003. Die Legionen des Augustus. Der römische Soldat im archäologischen Experiment. 9. erweiterte Auflage. Mainz: Zabern.
Sara Elise Phang. 2001. The Marriage of Roman Soldiers (13 BC – AD 235): Law and Family in the Imperial Army. Leiden: Brill.
Sara Elise Phang. 2005. Soldiers’ Slaves, ‘Dirty Work’ and the Social Status of Roman Soldiers. In: Jean-Jacques Aubert and Zsuzsanna Várhelyi (eds.). 2005. A Tall Order. Writing the Social History of the Ancient World. Essays in honor of William V. Harris. München: Saur, 203-229.
Jonathan Roth. 1994. The Size and Organization of the Imperial Roman Legion. In: Historia 43, 346–362.
Walter Scheidel. 2007. Marriages, Families, and Survival. In: Paul Erdkamp (ed.). A Companion to the Roman Army. Malden/Oxford: Blackwell, 417-434.
C. Sebastian Sommer. 1988. Kastellvicus und Kastell. In: Fundberichte Baden-Württemberg 13, 457-707.