Diocletian and Constantine presided over a major restructuring of the army, alongside their reform of civil administration. Troops were now divided into two distinct grades, the comitatenses and limitanei. The latter received less pay and had lower physical requirements, but were still full-time professional soldiers. The root of the name was the word for military road (limes), although there is some dispute as to the precise significance of this. The limitanei were also sometimes called riparienses, which came from ripa (the banks of a river). They were stationed in provinces, usually in frontier areas, and were commanded by duces. Their role was to patrol and police the area around their garrisons and deal with relatively small-scale attacks, such as raiding bands numbering a few dozen or at most a couple of hundred warriors. One study has noted that we never hear of raiding bands of less than 400 men and suggested that smaller groups were routinely stopped by the limitanei.
The comitatenses had an entirely separate command structure and were more likely to be moved from one region to another. The name was derived from comitatus (the imperial household), and the original idea was clearly that they should be at the emperor’s disposal. It is possible that under Constantine they were organised as a single army, ready to follow on campaign wherever he went. When his three sons divided up the empire in 337, the comitatenses were divided into three separate armies. Over time, more distinct armies stationed in specific regions would be created. In practice, the comitatenses were usually commanded by generals with the title `Master of Soldiers’ (MagisterMilitum), such as Silvanus and Ursicinus. Variants of the title included `Master of Horse’ (Magister Equitum) and `Master of Infantry’ (Magister Peditum), neither of which commanded exclusively infantry or cavalry, but a mixture of both. As subordinates, the Masters of Soldiers had officers with the rank of comes (count, pl. comites). Again, the word had its root in the personal companions who had traditionally accompanied an emperor on a journey or campaign. In some cases counts were given small-scale independent commands.
The Masters of Soldiers regularly commanded substantial numbers of troops, although it is unlikely that these forces were bigger than the army controlled by a senatorial legate in a major military province in the first or second century. There were only three of them – the number would double by the end of the century, but never increase beyond that – and this made it easier for the emperor to keep a close watch on them. Yet the Silvanus episode had shown that this might not be enough. More important in ensuring the emperor’s security were the complex divisions of responsibility and power throughout the provinces. The army was divided into two, with comfortably over half of its units being limitanei. However, co-operation between limitanei and cornitatenses does appear to have been common, and senior officers could in practice find themselves with troops of both types under their command during a campaign. A much bigger division was maintained between the army hierarchy and the civil administration.
The army depended on the civilian bureaucracies to supply it with pay, food and clothing. Even weapons and other equipment, which in the early empire had been made in the legions’ own massive workshops, were now provided by state-run arms factories under the supervision of the praetorian prefects. A Master of Soldiers planning to challenge the emperor had to secure not just the support of his soldiers, but the cooperation or replacement of large numbers of bureaucrats. It was much harder for one man to know, and win over, everyone who was important. At the same time, there were many people serving in independent hierarchies able to send real or fabricated reports of the disloyal behaviour of others. The system offered some protection to the emperor, at the cost of making it more difficult to get things done. Campaigns could be delayed or hindered by lack of supplies over which the commanders had no control.’
The Roman army in the fourth century was large, its manpower still dwarfing that of the swollen bureaucracy. The vast majority of the men paid and under the control of the emperors were soldiers. Yet we do not know how big the army was. Most scholars assume that it was larger than the second-century army, perhaps 50 per cent or even 100 per cent bigger, but the evidence is inadequate and inevitably the calculations involve a good deal of conjecture. We do know that the fourth-century army contained many more units and we have a complete list of those in existence at the very end of the fourth and the beginning of the fifth century. Some units had disappeared since Constantius II’s time, others had been created, but this list gives us a fair idea of the overall shape of the army in his day. Unfortunately, we have no certain evidence for the size of each unit so cannot calculate the army’s theoretical total size from this.
The limitanei included a very broad range of unit types. Some were survivals from the time of Marcus Aurelius and even earlier. There were legions, as well as auxiliary cohorts and cavalry alae. Others were new creations. Overall, there was a higher proportion of cavalry units in the limitanei than in the comitatenses, no doubt because they were useful for patrolling. There were no units including both infantry and cavalry, equivalent to the mixed cohorts of the early empire. Legions were sometimes stationed in several outposts. Several were split amongst five garrisons, while in the early fifth century the Legio XIII Gemina provided five garrisons on the Danube, another in Egypt and also had a unit amongst the comitatenses. Even so, they were certainly far smaller than the 5,000-man legions of the early empire. It is also more than possible that most units were smaller than the roughly 500-man cohorts and alae of the auxiliaries in the second century. Very many of the forts occupied by the limitanei were tiny, most a small fraction of the size of earlier auxiliary forts. An Egyptian papryus dating to the start of 300 also suggests some very small units. It mentions a cavalry ala with 116 – just over a year later this had risen to 118 – a vexillatio of legionary cavalry with 77 men and a unit of mounted archers with 121. A number of units of legionary infantry averaged around the 500 mark, but a camel troop seems to have had just a couple of dozen men.
The situation with the cornitatenses is no clearer, although there was less variety of unit types. Infantry consisted of legions and a new type of unit known as auxilia. All of the latter and some of the former were rated as palatina, a title that carried much prestige and some tangible advantages in pay and bonuses, but no difference in function. Cavalry units were all called vexillationes and were smaller than the infantry regiments. A common estimate is to give cavalry units a strength of boo and the infantry somewhere between 1,000 and 1,200. However, the few mentions of regimental strengths in our sources suggest smaller numbers, averaging around 350-400 and 800 respectively. What we do not know is whether such lower figures represent actual campaign strengths, reduced by disease and casualties, or theoretical sizes. A couple of sources mention commanders who kept non-existent men on their unit’s role so that they could draw their pay and rations. Infantry units of colnitatenses were brigaded in pairs, and seem to have permanently operated together. This obviously adds an extra complication since we cannot be sure whether numbers in our sources refer to a single regiment or a pair that the author naturally assumed would be together!
Overall, it is fair to say that the units of the fourth-century army were smaller than their first- and second-century predecessors. Going much further than this quickly becomes conjectural. We cannot even be sure that all units of a particular type had the same theoretical size. Infantry regiments of all types in the field army were probably somewhere between 500 and 1,000 men strong, cavalry units about half the size of infantry. Many units of limitanei may have been a good deal smaller, and perhaps we should think in terms of 50 to 200 men, which would make them more like companies than battalions by modern standards. It is possible that on paper the total strength of the army was larger than in Marcus Aurelius’ day, but we cannot be sure of this. Its actual strength on a day-to-day basis is even harder to assess.’
The evidence is equally poor for the army’s recruitment. Some men were volunteers and others conscripts, but the balance between the two is unknown. Sons of soldiers were legally obliged to join the army. Landowners had to supply a set number of recruits as part of the taxation system, but were often able to commute this duty to a money payment or avoid it altogether. There was also a steady supply of drafts for the army from the groups of barbarians settled by treaty within the empire and obliged to provide soldiers for the army. Tribesmen from outside the empire also came as individuals or as groups to enlist in the army. The old idea that this influx of Germans `barbarised’ the Roman army and over time reduced its efficiency has been discredited. There seems to have been no real difference in reliability and performance between recruits from inside or outside the empire. As we have seen, the senior ranks included many men of barbarian descent who behaved exactly like colleagues of more traditionally Roman stock.