How much can we blame barbarisation of the Roman army in leading to the end of the western empire in AD476?


The ‘barbarisation’ of the Roman army has become a popular explanation for the end of the Roman Empire. While the empire possessed a strong army of professional Roman soldiers, it could not fail; therefore its end in AD476 was the result, directly or indirectly, of the failure of the army. And since by the fifth century AD the army had come to incorporate many non-Romans into its ranks, logic follows that this ‘de-romanisation’ of the army – the deterioration of Roman military discipline, the end of the legions of the Principate – made the army ineffective and weak.¹

     Considering how widely accepted this idea is, it is surprising how little evidence there is for the barbarisation of the fifth-century army. Aside from the tirades of Vegetius regarding the deterioration of Roman discipline² and the letters of Synesius, who seemed to enjoy criticising everyone and everything,³ there is little general, let alone specific, evidence (as it is impossible to quantify ‘barbarisation’), to prove that Roman soldiers of the fifth century were any less capable than their Principate counterparts. More importantly, there is very little evidence for the army of the fifth century in general. While the fourth-century army is quite well attested thanks to the account of Ammianus Marcellinus, the disappearance of more or less reliable historical narrative after Ammianus has created a yawning gap in our knowledge of the Roman army between the late fourth and early sixth centuries. Among what we do have, the Notitia Dignitatum is the most important single source, followed by laws from the Codex Theodosianus regarding military matters. The History of Zosimus is also useful, though it is important to note the bias in his writing. Beyond that we must rely on sources from before or after the fifth century: Ammianus’ account is ever useful, as are anecdotes from several sixth-century writers.

     This essay aims to show that conclusions regarding the fifth-century army in general are very hard to draw. However, this essay will show, through study of the sources, that the army of the fifth century was perhaps not as radically different from the army before as has often been supposed. Barbarisation of the army might have made the army weak, but from what we know, it is hard to say that it did.

The Notitia Dignitatum

The Notitia Dignitatum, or Register of Dignitaries, is for our purposes a list of all the military units of the fifth-century Roman army. It is split into two halves: the Occidens and the Oriens. There are three main areas of the Notitia in which we are interested: the lists of the magistri (Occ. 5-6), which document the rankings of the various military units; the distributio (7), which lists their location; and the lists of the limitanei and their officers (25-42). While the Notitia is arguably the most important source for our period, it has several major flaws. First, it is important to remember that the Notitia does not provide a snapshot of the fifth-century army at any one point. The document was drawn up at an unknown date, and in the copy possessed by the western empire, the western half continued to be revised sporadically and half-heartedly down to perhaps the mid-fifth century.⁴ The eastern half contains no entries or military units datable after the end of the fourth century.

     Secondly one must remember that the Notitia gives paper strengths only. Many of the units listed were duplicated. Jones notes the problems between the lists of units under the magistri and the distributio;⁵ while both seem to share a purpose of listing troops, in fact the magistri list primarily aimed to show precedences, while the main purpose of the distributio was to record actual army strengths and numbers. The problems caused by this will be examined in more depth in the next section. For now let us examine what the Notitia can tell us about the fifth-century army.

Troop Types and Numbers

At first glance, the late Roman army, especially the fifth-century army, was structurally a very different beast from that of the Principate. Up until the days of the Tetrarchy, infantry legions, auxiliary cohorts and alae were the functional components of the Roman army. By the mid third century, however, a new body, the comitatus, was rising in importance. The comitatus, a quick-response cum bodyguard force composed mostly of legionary detachments, was born out of the civil wars of the third century; emperors found it much easier to deal with threats with a personal force continuously at their disposal. By the time the Tetrarchy was established, Diocletian and all his colleagues had their own comitatus. In time it would become the most important component of the forces at the emperor’s disposal.⁶ 

     Though Diocletian maintained and no doubt enlarged the comitatus as an institution, the standard view is that Constantine made it the central focus of the army. Zosimus blames him for denuding the frontiers and weakening Roman defences⁷ – though this was probably pagan bias (as a pagan Zosimus was critical of Constantine’s support for Christianity). In fact Constantine was siphoning detachments (as well as raising new units) to enlarge his comitatus, which now became a central mobile striking force, the core of the Roman army. Constantine also instituted the magister peditum and magister equitum to take command of the new field army troops, the comitatenses. The old frontier troops were designated limitanei, and became subordinate to the comitatenses in rank and prestige.

     Constantine is also credited with the raising of new infantry units called auxilia, usually believed to be have been recruited mostly from conquered ‘barbarians’.⁸ In the Danubian provinces (Pannonia and Valeria, among others), the auxilia and new cavalry units called cunei replaced the old auxiliary cohorts and alae (though these were not totally replaced, as is evident from the Notitia). The measures taken by Diocletian and Constantine therefore created, out of the chaos of the so-called ‘third-century crisis’, the late Roman army, in which distinctions were no longer between citizen and auxiliary, but comitatenses and limitanei.⁹ At the beginning of the fifth century the army cannot be proved to be radically different from this basic formula. As is evident however from the Notitia, precedence was strictly graded and of the utmost importance. Palatine legions, auxilia and cavalry come top of the list of the magistri,¹⁰ followed by field army legions and cavalry, and then the units of the limitanei.¹¹ Demotion to a lower rank could be inflicted on entire units as a punishment.¹² Also, by this period many of the comitatenses had ceased to serve as mobile troops; instead they were dispersed throughout the provinces (those stationed with or around the emperor gained the epithet palatina).¹³ The best elements of the limitanei were also transferred to the now regional comitatenses and styled pseudocomitatenses, arguably weakening the limitanei considerably. After the end of Honorius’ reign, emperors stopped raising new units altogether, according to Jones.¹⁴ Instead they promoted units of limitanei to the pseudocomitatenses, or enlisted the help of foederati (militarily-obligated barbarian settlers).

     The standard infantry unit was the legion, found both in the field armies and the limitanei. Though the infantry legion had theoretically numbered 6000 men (in reality around 5000) till the end of the Tetrarchy, the ‘legion’ of the Notitia Dignitatum clearly was not of the same scale. The change was most probably born out of the military reforms of Diocletian and Constantine; since the dark days of the third century it had become common practice to field detachments called vexillations¹⁵ from the 5000 strong legions instead of the legions themselves. Though there is again no proof of a standard strength for any vexillation (they were detached ad hoc, their size depending on the situation), some of them were styled milliariae,¹⁶ and therefore were probably around 1000 strong. From the time of Diocletian onward, emperors began to station these vexillations permanently away from their parent legions. The descendant formations of these detachments are perhaps what the Notitia calls ‘legions’.¹⁷ On the one hand, legions which were raised by Diocletian’s successors were probably of similar size to these milliariae;¹⁸ on the other, the old-style legions were not initially scaled down to 1000 men as a matter of course. But over time they would have been scaled down through transfers, cross-postings and attrition in battle. Several old legions along the Danube probably retained their full complement for a while, but in time these were spread out into several detachments. Examples found in the eastern lists of the Notitia include the Secundae Herculiae (in Scythia, spread out into three local detachments), the Primae Italicae (in Moesia Secunda, three detachments) and the Tertiadecimae Geminae (on the Dacian frontier, five detachments). Given that old legions were around 5000 strong and the ones named above were divided into up to five detachments, we may tentatively speculate two things: that these 1000 strong detachments are what the Notitia identifies as ‘legions’ (probably in the west, since the detachments above were all identified as cohorts), and that in any case by the fifth century even the old legions were well under 5000 strong, having been posted into several detachments.

     There is no concrete evidence for the size of the new legions, but from the sources one can put their size by the fifth century AD at much more than 500 men, but much less than 6000 men. There are three pieces of evidence for this: in Ammianus’ account of the AD359 siege of Amida,¹⁹ he notes that some 20,000 Romans – including seven legions – were besieged by the Persians. All that we can deduce from this anecdote is that the ‘legions’ mentioned cannot have been 6000 strong. Though this is unsatisfactory in itself, it lends more weight to our other suggested figure of roughly 1000 per legion – indeed a force of around 7000 soldiers defending 12,000 besieged civilians seems at least plausible.²⁰ If on the other hand one wishes to argue that legions by the late fourth century were mere skeletons, one must note three anecdotes, also from Ammianus’ account: in AD360 Constantius requisitioned 300 men from each unit of Julian’s army, in AD377 Gratian picked 500 men from each unit for special operations, and in the same year Valens picked 300 men from each ‘legion’ of his army to deal with the Goths in Thrace. All of these anecdotes strongly imply that the ‘legion’ of the field army was much larger than 500 men,²¹ if a mere detachment was already that size. Finally Zosimus²² writes that Honorius withdrew from Dalmatia ‘five regiments of soldiers…[these] consisted of six thousand men’. If one hazards to assume that these ‘regiments’ (ταγματα) are the same as the ‘legions’ of the Notitia,²³ then the legion, the main tactical infantry unit of the fifth century, was indeed around 1000 strong. Though none of these three sources decisively gives the number of the fifth-century legion (and indeed none of them actually refers directly to it), at least we now have parametres for its size: much less than 6000, much more than 500, and perhaps around 1000.

     Next let us examine the size of the cavalry vexillatio. There are two sources: the first is Zosimus, who notes the retreat of a cavalry unit at the Battle of Strasbourg numbering 600 men²⁴ – this was almost certainly a vexillatio, as Libanius, in his account of the same event and describing the same men, had associated them with standard bearers (τοις τα σημεια φερουσιν), possibly a semantic mistake on his part.²⁵ Secondly, the size of the Justinianic cavalry vexillatio is given by the sixth-century writer John Lydus as 500 strong; this was also the same size as a contemporary schola, a unit of the imperial mounted bodyguard.²⁶ Since 500 had been the traditional size of a cavalry unit (auxiliary alae had been 500 strong under the Principate), and clearly continued to be so down to the reign of Justinian, it is reasonable to place the size of a fifth-century vexillatio at around 500 men.

     The auxilia are more problematic. Again there are two sources: firstly an anecdote by Zosimus²⁷ in which the emperor of the east sent ‘six regiments of auxiliary soldiers…amounting in number to four-thousand’ to Honorius’ aid.²⁸ Assuming that these ‘regiments of auxiliaries’ were indeed auxilia,²⁹ this would place the strength of an auxilium at around 650 men.³⁰ Secondly, Aurelius Victor notes that when Maximian gained the epithet Iovius for his worship of the god, some ‘auxiliary units’ (Caes. 39.15) which stood out in the army also gained this title. And indeed, the Iovii are a unit of auxilia palatina listed in the Notitia.³¹ Though the connection is not very strong, one could take the semantic similarities between the traditional ‘auxiliaries’ mentioned by Victor and the new auxilia as proof of a common root, and therefore perhaps a common strength of 500 men. Therefore a unit of auxilia in the fifth century perhaps numbered 500 to 700 men. Though the conventional view is that the auxilia were an entirely new formation raised by Constantine and composed of barbarians,³² it is more likely that the auxilia were in fact not totally different from the traditional auxiliary cohorts of the Principate.

     As for the traditional auxiliary cohorts and alae, Jones assumes that they continued to be maintained at 500 men as they had always been;³³ this is based on the presence of several cohorts in the Notitia (albeit in the eastern list) styled ‘milliariae’,³⁴ probably 1000 strong. Given that in the fifth century some provincial border troop formations seem to have remain largely untouched by the Diocletianic and Constantinian military reforms (note the large numbers of cohorts and alae in Britain and Raetia – see Fig. 3), perhaps their theoretical strength stayed approximately the same as well.³⁵

     The strengths of the rest of the units mentioned in the Notitiagentilia, cunei, numeri, milites, limites – are almost impossible to deduce. Cavalry cunei and equites were probably 500 strong like the alae and vexillationes, while numeri and milites, newly raised formations found in embattled provinces such as Gaul and southern Britain, perhaps numbered not much more than 500 men, as the hard-pressed local manpower could probably provide no more. Perhaps they were a stop-gap measure to replace destroyed legions. In the end it is entirely unknown if these were infantry, cavalry, or mixed units.³⁶

     Having examined theoretical sizes of individual units in the fifth-century army, we can now turn to the army itself. There are three ancient estimates for its size: the first is John Lydus, who gives a precise figure of 435,266 men during the time of Diocletian (389,704 land troops, 45,562 fleet troops). The problem with this vague timeframe is that, as we have examined earlier, the army at the start of Diocletian’s reign was, owing to his military reforms, quite different from the one at the end. It is more likely however that Lydus’ figure refers to the latter. Zosimus gives a second figure, claiming that on the eve of the Battle of the Milvian Bridge, Constantine had 90,000 infantry and 8000 cavalry, and Maxentius 170,000 infantry and 18,000 cavalry. If, as Jones has it, these are taken to be total army sizes and not field army sizes, the total strength of the early fourth century western Roman army would have been 286,000.³⁷ The sixth-century historian Agathias gives a third figure, stating that ‘previously’ the Roman army had 645,000 men (this large figure probably refers to the manpower of the entire Roman Empire). 

     As with all ancient figures one must treat these with caution. But Lydus’ figure seems reliable; Lydus himself worked in the office of the praetorian prefect and therefore could have had access to official figures. And as Lee states, the very precision of this figure makes it more believable.³⁸ For our purpose however I have conjectured that Agathias’ figure refers to the fifth-century army, which allows us to theorise how the size of the army evolved from the late third to fifth centuries. We will examine these three figures later, but first we will look at the numbers given in the Notitia. Below are three tables based on the units and formations listed in the western half of the Notitia

Troops of the Western Comitatus (Magistri, Occ. 5-6) – Fig. 1

Troops Types under the Magistri Peditum and EquitumNumbers
Palatine legions12 units x 1000 men 🡪 12,000 men
Auxilia palatina³⁹65 x 500 🡪 32,500
Comitatensian legions32 x 1000 🡪 32,000
Pseudocomitatenses18 x 1000⁴⁰ 🡪 18,000
Palatine vexillations10 x 500 🡪 5000
Comitatensian vexillations32 x 500 🡪 18,000
Total: 117,500 men

Distribution of Field Army Troops in the Western Provinces (Distributio, Occ. 7) – Fig. 2

Western ProvincesLegionsAuxilia PalatinaVexillationesPseudo-comitatensesTotal Numbers of Men
the Gauls1015121033,500
the Spains51110,500
the Britains366000

Troops of the Western Limitanei (Occ. 24-52) Fig. 3

CommanderTroop TypesNumbers of Men⁴¹
Count of Africa16 limites8000
Count of Tingitania7 cohorts, 1 ala4000
Count of the Saxon Shore4 numeri, 1 milites, 2 equites, 1 cohort, 1 legion5000
Duke of Mauritania8 limites4000
Duke of Tripolitana11 limites5500
Duke of Pannonia6 cunei, 11 equites, 1 ala, 5 auxilia, 5 legions, 1 milites, 5 classes, 4 cohorts18,500
Duke of Valeria5 cunei, 17 equites, 5 auxilia, 8 legions, 1 classes, 6 cohorts24,500
Duke of Pannonia prima2 cunei, 14 equites, 1 gentis, 8 legions, 3 classes, 5 cohorts19,000
Duke of Raetia3 equites, 5 legions, 4 alae, 1 milites, 7 cohorts, 1 gentis, 1 numeri13,500
Duke of the Province of the Sequanici1 milites500
Duke of the Armorican Tract1 cohort, 9 milites5000
Duke of Belgica Secunda1 equites, 1 classes, 1 milites1000
Duke of Britain1 legion, 3 equites, 11 numeri, 16 cohorts, 5 alae, 1 unknown, 1 cuneus19,500
Duke of Mogontiacensis11 milites5500
Magister militum praesentalisItaly: 4 classes, 1 milites, 17 gentilia;Spain: 1 legion, 5 cohorts, 11 laeti, 2 unknown;Gaul: 4 classes, 1 milites, 2 cohorts, 6 gentiliaItaly: 9000;Spain: 10,000;Gaul: 4500
Total: 157,000 men

When the figures of the field army and the limitanei are added together, this gives a grand total of 274,500 men.⁴² Now if we examine the three estimates mentioned above (and once again I stress my assumption that Agathias is referring to the fifth century), it is apparent that the Roman army shrank over time; from nearly 400,000 around the time of Diocletian, to 286,000 at most by Constantine’s day, and by the early fifth century, 274,500 men, smaller perhaps than the eastern army, if we take 274,500 as the western portion of Agathias’ proposed 645,000. This makes sense given the heavy fighting in which the western army was engaged throughout most of Honorius’ reign.

     Yet despite the shrinkage over time – if our above figure of 274,500 were true – if the western emperor had more than 270,000 Roman soldiers to defend only the western empire (the Principate had in theory only 300,000 men to defend the entire empire), the fifth-century western Roman army should have been in a very strong position. Conventionally it is asked, why then did the late Roman army perform poorly? Why were Roman armies of the fourth and fifth centuries in fact so small compared to Principate armies? Caesar, during his expedition to chastise the Getae and the Parthians, had with him some 16 legions, (90,000 infantry) and 10,000 cavalry.⁴³ When Augustus mustered his troops to deal with the Pannonian revolt in AD6, he could field ‘ten legions, over seventy cohorts, ten alae and more than ten-thousand veterans’,⁴⁴ an extraordinary force numbering upwards of 100,000 men. Julian’s grand expedition to Persia is the usual ‘other’ in the comparison of early versus late Roman army: an extraordinary force as well, the only one to draw from the resources of both eastern and western empires,⁴⁵ yet which came to only 60,000 men. It is important however to remember that logistical constraints necessarily kept Julian’s army small; and even then Julian’s men found it difficult to keep themselves fed the further they advanced from the Tigris.⁴⁶ It might have been quite different if Julian had been campaigning in Pannonia or Gaul.

     On the other hand Tacitus describes Corbulo’s Armenian expedition as consisting of six legions⁴⁷ – perhaps some 25,000 men, excluding auxiliaries and allied troops – certainly a less grandiose force than that of Augustus and Caesar. In comparison, Valens had perhaps 15,000 men with him at Adrianople.⁴⁸ Some 20 years later, Stilicho could muster only ‘thirty cohorts’ (αριθμους) and an unknown number of ‘auxiliaries’ to fend off Radagaisus’ invasion⁴⁹ – 15,000 Roman troops, and perhaps 5000 ‘auxiliaries’.⁵⁰ So why were late Roman armies so small? One should instead ask, what size should an army have been? The scarce evidence regarding army strengths does not allow us to conclude that late Roman armies were necessarily smaller than their Principate counterparts. Sources only tend to give army strengths in the context of unusually large forces (Caesar, Augustus and Julian), heroic victories (Corbulo,⁵¹ Stilicho) or crushing defeats (Valens). We simply do not know the ‘standard’ strength of an army in a ‘normal’ campaign (if such concepts indeed ever existed) – was 15,000 men, the number under Valens and Stilicho, the norm during the late empire? Was it a norm during the Principate?⁵² The Notitia certainly does not help at all in this regard, and we must therefore re-examine its paper strengths, to see if the army it portrays was ‘standard’, understrength or ‘overstrength’.

     We have already explained the different functions of the magistri lists and the distributio, and how the two do not complement each other. Inconsistencies have naturally arisen out of their two very different functions. For example, two units of pseudocomitatenses, the Taurunenses and the Antianenses, appear in the magistri, but nowhere in the distributio. Had they been destroyed yet recorded in the magistri for the sake of ranking other, surviving units? They certainly were not the only ‘ghost’ units. Tomlin notes that two African units, the equites IV sagittarii and the Constantiniaci, had been destroyed in AD373⁵³ – yet they were kept on the rolls, their rations and pay lining the pockets of their tribunes and duces.⁵⁴ Units were also sometimes duplicated upon promotion or demotion,⁵⁵ or upon transfer to other provinces.⁵⁶ The Notitia also gives no indication of unit size, particularly for the myriad different types of limitanei,⁵⁷ nor any indication of under-strength (which would have been acute after the chaos following the AD406 invasions). These minutiae aside, it is now clear that the ‘army’ portrayed in the Notitia is in fact a chimera stretching from the early fourth to fifth centuries. Thus it credits to the fifth-century army far more men than it probably had, and is far from helpful for trying to calculate the army’s gross number.

     Thus there is still no final verdict on the total size of the fifth-century army. All that we can deduce is that it was smaller than the army at the end of Diocletian’s reign, and, if we are to trust Zosimus’ figures, slightly smaller than that of Constantine. More importantly, it was not radically different from armies preceding it: many traditional infantry units remained in place, and while legions were now smaller in size, small infantry detachments had been in use since the Principate. Many units, such as the auxilia and vexillationes, also appear to have been old units under different names. Despite this continuity however and the army’s apparent position of strength, much of the elite field army had already been dispersed into urban centres, where they presumably did not see much active service. Though it is true that Roman army throughout the Principate, whose fighting ability has rarely been called to question, habitually billeted its troops in urban centres, it is probably equally true that prolonged postings in towns would have made soldiers at least slightly rusty. Their eventual decline however into ‘mere gendarmerie’, as Jones puts it, is probably an exaggeration.⁵⁸ The heaviest fighting in Britain, Gaul and along the Danube took its toll on the local limitanei, as is evident from the presence of so many newly raised milites and numeri, as well as recently promoted pseudocomitatenses. Meanwhile the numbers of auxilia, the single largest element in the Roman army of the Notitia, points us back to our main question. Since many of the auxilia were in fact promoted foederati, does this mean the Roman army was heavily barbarised?

Barbarians in the army

What was barbarisation of the Roman army? It cannot be quantified in any meaningful way, though common examples of barbarisation include the adoption of barbarian trousers, the Germanic long-sword (which replaced the old gladius) and the Germanic war cry known as the barritus. There were also large numbers of non-Roman officers,⁵⁹ many of them rising to high rank. A famous example is the magister militum Stilicho, a Vandal by birth (though thoroughly romanised).

     It is hard to prove that the adoption of ‘barbarian’ equipment and dress ruined the fighting ability of the Roman army. One should remember that the Roman army had always adopted enemy equipment when it proved useful. The fighting style itself of the legion evolved over centuries of contact with barbarian warriors, and the scutum and the gladius, the most iconic gear in the Roman arsenal, were adopted from the Celts and Spanish. Archaeological and pictorial evidence⁶⁰ suggest that the Roman army continued adopting enemy gear down to the fifth century and beyond, a continuation of time-honoured tradition. And despite the low esteem in which he held the Roman army of his day, even Vegetius admits that learning from the Goths, Alans and Huns had benefited the arms of the Roman cavalry.⁶¹

     We may however examine the barbarians themselves in the Roman army. Barbarian soldiers in the fifth-century army would have come from many different tribes, Huns, Alans, Germans, Goths and Sarmatians among others, but in the west the majority would have been of Germanic extraction.⁶² Was the practice of recruiting these barbarian soldiers into the army detrimental to discipline? There is little evidence to support this view. Nevertheless for the sake of the argument let us assume that barbarian soldiers were a bad influence. There certainly were units bearing Germanic names (and therefore originally raised from Germanic tribesmen) which feature throughout the Notitia, but cross-postings would have diluted their ranks, and even then there is no indication that efforts were ever made to maintain their tribal character.⁶³ Thus any barbarian ill-discipline would have been flushed out by service under Roman officers,⁶⁴ obedience to Latin orders and drill, and postings in (sometimes) far off provinces.

     And yet it is difficult to prove that barbarian soldiers were truly bad for the army. How exactly did barbarian recruits undermine Roman discipline if they served under Roman officers? This seems to be a wholly modern idea; Vegetius in his long list of complaints never suggests that the decline of the army is due its recruitment of barbarians; even Synesius has nothing but praise for the barbarian troops in Cyrenaica.⁶⁵ Ammianus, as an ex-officer who had served with Germans, never once suggests that they were poor soldiers.⁶⁶ In fact, as his account implies,⁶⁷ Germanic recruits were most desirable for the crack imperial guards of the east.

     If there is no proof that German recruits were unreliable soldiers, what of their loyalty to Rome? As non-Romans by birth, could barbarian soldiers have harboured dual loyalties? Indeed, a supposed leak of intelligence to the Alamanni before the Battle of Strasbourg, as Ammianus records (he is our main source regarding barbarian regulars), brought suspicion upon the high-ranking German officers Latinus, Agilo and Scudilo.⁶⁸ There are in addition several cases of ‘treachery’ in Ammianus’ account: again before the Battle of Strasbourg, another deserter from the Scutarii went over to the Alamanni and told them of the size of the Roman force, while during the reign of Gratian a Germanic guardsman who had returned to his home on private business accidentally gave away the weakness of the Rhine frontier.⁶⁹ 

     These are the only two confirmed cases of ‘treachery’ from barbarian soldiers in Ammianus’ account; there is no mention of betrayal on the battlefield or mass desertions. And in any case, the Strasbourg deserter was motivated by fear for an unpunished offence not sympathy for his kinsmen, while the guardsman could not be convicted of much more than carelessness with words. In truth, Germanic tribes were as apt to fight among themselves as they were to fight the Romans. There was no concept of a German ‘people’ during the fifth century (and indeed for very long afterward) which could have led to a battle of loyalties in the minds of Germanic recruits. One tombstone from the later empire dedicated to a Franco-Roman soldier described the man as ‘a Frankish citizen and a Roman soldier’.⁷⁰ Apparently while Germanic identity became more defined in the later empire, it could co-exist quite smoothly with a Roman identity. This case aside, most Germanic soldiers in the sources come across as thoroughly romanised.⁷¹ One example is the Frankish magister peditum Silvanus who served under Constantius II. Having been accused of treason, the first place in which he thought to seek refuge was among his own people; but one of his tribunes reminded him that they would probably kill him or betray him to the emperor. As a result Silvanus was forced to appeal to his own Roman troops.⁷² If Germanic soldiers could expect such a reception back home it would have given them much incentive to stay loyal to Rome. 

     It is worth mentioning also the foederati. These were barbarian migrants who had entered the empire and were allowed to live as settled, autonomous communities. In return they paid Roman taxes or supplied military manpower for particular campaigns. The conventional view is that as the empire weakened, emperors found it more convenient to hire foederati, trained from birth in warlike societies, than to train new soldiers. But in many cases ‘it was not entirely a matter of choice’.⁷³ The authorities could either absorb these groups or destroy them, and the latter became increasingly difficult as the fifth century progressed.⁷⁴ Thus, the common view goes, by the mid fifth century the foederati had effectively replaced the regular Roman units where defence and law and order were concerned; the foederati became a bad influence to what remained of the army as far as discipline was concerned, since they themselves did not serve under Roman officers, but under their own chieftains.

     This common view seems to agree with the sources. Ammianus notes that Goths of his day (early fifth century) did not serve under Roman officers as they had once before.⁷⁵ Jones takes this as an example of early fifth century foederati.⁷⁶ Orosius also notes that the emperor Honorius promoted many foederati to the rank of auxilia,⁷⁷ while Synesius mentions units of barbarian Unnigardae and Marcomanni stationed at his city.⁷⁸ While there is no reason to dismiss these sources outright, one should look back to the Notitia. In the embattled prefecture of Gaul (see Fig. 2), of 33,500 field army troops only 7500 are auxilia, while in Italy, 11,000 of 31,500 field army troops are auxilia; even if we are to believe that by the mid fifth century most auxilia were in fact promoted foederati, the gist of the Notitia’s figures is that well over half the provincial field armies were still made up of regular troops. Also, the barbarian troops mentioned by Synesius are described as auxiliaries sent by the emperor; if anything they were regular auxilia (under Roman officers) not foederati. As for the foederati being a bad influence on the rest of the army, one must remember that like many of the components of the fifth-century army, the foederati were not novel. The Batavi had been serving in a similar capacity since the first century AD, and while they did revolt in AD70 they were never accused of being a bad influence on Roman troops.

     Overall, barbarians in the army cannot be counted as a weakness in the army, so long as they were properly incorporated, and trained in the Roman way. The foederati were probably not as prevalent as the common view has it, or as destructive to discipline. While it is true that the regular army did eventually become sidelined by the foederati,⁷⁹ this did not happen until the final few years before AD476, when the regular army truly withered away, and with it a major component of imperial power. It would have been different if the Romans had properly incorporated their foederati into the regular army; but this would have required time and stability, which for the fifth-century western empire were in short supply.⁸⁰

The Limitanei

Contempt for the limitanei is, together with barbarisation of the army, a cherished and popular component in the fifth-century military decline theory. The limitanei, separated from their founding from the elite comitatenses, and subordinated to them in status by Constantine, are hardly mentioned specifically in any of the literary sources. There is no record of their ability as combat troops, which tempts one to believe that they were beneath mention. Even more damningly, according to Procopius, by the sixth century Justinian considered them soldiers no longer.⁸¹ The sources of the fifth century however are scattered and varied, and few are too concerned about objectivity. Many of the letters or poems from this period mention only heroic deeds or victories, and the relatively small-scale operations of the limitanei against endemic barbarian raids would have attracted little attention. An oft-quoted source regarding the weakness of the late Roman army is Synesius, who at length berates the unreliability, cowardice and rapaciousness of the soldiery in his city of Cyrenaica.⁸² But he cannot be counted on for rational military analysis: firstly Synesius was not a military man, and secondly, as mentioned previously, he obviously had an axe to grind, and his criticisms extend to nearly all people from all walks of life. Some of his criticisms are also somewhat irrational, such as his continuous praise for barbarian auxiliaries, which he contrasts with the useless local soldiers.⁸³ This bias possibly stems from the fact that the local Roman commander had had to abandon stretches of land, including Synesius’ estates, to the barbarians.

     Beside their invisibility in the sources, Procopius’ anecdote (which goes beyond the period we are interested in) and the irrational bias in Synesius’ letters, there is little evidence to suggest that the fifth-century limitanei were the run-down, hereditary peasant militia they are believed to have been. There are however two pieces of evidence which seem to support the common view: the first is an anecdote from the Historia Augusta regarding Alexander Severus. The emperor rewarded his soldiers with land, which they could cultivate on the condition that their heirs entered the service as well.⁸⁴ The second piece of evidence is a law enacted by Valens dating to AD365, declaring that the limitanei must receive rations in kind for nine months of the year, and money for the other three.⁸⁵ Even by the mid-fourth century the limitanei seem to have been 9/12 of their way to becoming unpaid militia – perhaps they made up for the loss in wages by farming.

     Firstly one must remember that the Historia Augusta is often fiction, and there is no indication at all that land with attached military obligations was given to soldiers during the reign of Severus Alexander. Service was also hereditary (theoretically) for all soldiers since the reign of Diocletian; indeed most professions were theoretically hereditary, and this applied as much to the comitatenses as to the limitanei.⁸⁶ Overall the Historia has clear didactic and moralising tones, probably meant as a piece of advice for the emperor Constantine.⁸⁷ If so, then the reference to land with military obligations is, if anything, merely a suggestion. And all things considered it is not necessarily the case that soldier-farmers were ineffective – the Byzantine army of the eleventh-century re-conquests was made of soldier-farmers. Even if fifth-century soldiers did farm their land, firstly they probably would have had tenants or slaves do the actual work while they themselves served in the army, and secondly, this was not such a radical departure from what happened under the Principate, when land grants on discharge were part and parcel of a successful army career. By the fifth century many sons of veterans (both in the comitatenses and the limitanei) would have inherited, and profited from, land from their fathers. In any case the Notitia strongly suggests that the western limitanei were not tied to the land during the fifth century. One should recall the magistri list in the Occidentis, and its 18 units of pseudocomitatenses. If the limitanei were indeed settled and unprofessional troops, it would have been extremely difficult to incorporate them into the comitatenses.

     As for the 365 law, firstly the very reliance on payment-in-kind suggests that the limitanei did not in fact possess the means to grow their own food. In fact, this law was predated by one which allowed for an allotment of land or a cash bonus to be granted to all veterans,⁸⁸ which would make sense only if all the soldiers were professional troops. This brings us to our second point regarding laws, which is that they tended to swing back and forth: another law dating to AD409 and attributed to Arcadius, declared that military supplies throughout the three Palestines (including those of the limitanei) were to be commuted at a fixed tariff – and so the limitanei once again were paid in money all year round.⁸⁹ Therefore no one law necessarily dictated how the limitanei were paid. True to this cycle of change, by AD443 the eastern limitanei appear to have become soldier farmers (though again not necessarily ploughing their own land).⁹⁰ But there is simply no conclusive proof that the same could be said of the fifth-century western limitanei. Nevertheless what we can conclude with some certainty is that as late as AD409, the eastern limitanei were still professional troops, and that perhaps the same went for the western limitanei. Hereditary service was not exclusive to the limitanei, and even if they did own land, it is unlikely that they would have worked it personally. Moreover many of the western limitanei were organised and effective enough as soldiers to be promoted to pseudocomitatenses and entrusted with such vital provinces as Gaul.

     So far we have established that the limitanei were not peasant militia. It would nevertheless be useful to further distinguish them from two military ‘units’ which were not regular border troops.⁹¹ First are the burgarii mentioned specifically in Spain,⁹² listed in the section of the Theodosian Code dealing with military affairs (so they were probably were soldiers). Like all soldiers they were hereditarily tied (in theory) to service. Unlike soldiers, however, they were tied to their actual areas as well. This is all we know of them, as they are even more poorly attested to than the limitanei (they are not mentioned at all in the Notitia). In any case there is no reason to equate them with the limitanei.⁹³

     The second military unit are what I have termed military settlers, what the Notitia styles as laetora and gentilia, and perhaps also the limites (found mostly in Africa). The laetora and gentilia perhaps refer to units drawn from settled military immigrants (laeti and gentiles). In the Notitia several laetora named after Germanic tribes are listed under the magister militum in Italy – the Teutones, Batavi, Sueves and Franks – while several gentilia of Sarmatians are also listed.⁹⁴ It is unclear whether these laetora and gentilia were actual military units, or communities which served as recruiting grounds for regular soldiers – similar to the foederati, except that they were overseen by Roman officers (praepositi). The practice of hiring laeti was not new however; some laeti had lived in the Empire since the days of Augustus. Tacitus describes the Ubii being settled on the Roman side of the Rhine⁹⁵ as well as Moors settled in parts of Spain.⁹⁶ While we know little of them, it is clear that the laeti and gentilia were not the same as the regular limitanei.

     The limites are more peculiar. Like the numeri and milites they were particular to specific regions, in this case Africa. From Fig. 3 it is apparent that while the Count of Africa and Duces of Mauritania and Tripolitana commanded between them the third largest concentration of field army troops in the western empire, they had under them no conventional units of limitanei at all. The only non-field army troops at their disposal were the limites. How then can we prove that the limites were distinct from the limitanei? Firstly we know that the limites, like the laetora and gentilia, and unlike the numeri and milites, were not a new development. Jones notes that a third-century inscription from Tripolitana mentions a praepositus limitum.⁹⁷ Secondly, the limites seem to have been military settlers under Roman authority; Ammianus writes of a prefect⁹⁸ of the Mazices (a north African tribe) supporting their chieftain Firmus in a revolt, and how Theodosius the Elder crushed the revolt and re-imposed Roman authority on the desert tribes by appointing reliable prefects over them.⁹⁹ Augustine mentions in his letters the Arzuges, who were clearly not Roman troops, but tribesmen who, under oath and ‘in the presence of the decurion’, carried out guard and escort duties.¹⁰⁰ These were probably what the Notitia styles as limites. Finally, Jones points out that the African desert boundary was marked by a series of discontinuous and overlapping earthworks, and that there is evidence of substantial areas of irrigated land on either side of it.¹⁰¹ If these lands belonged to the limites, and they probably did, then it is now very clear that the limites were mainly North African allied tribesmen who farmed border land, carried out patrol duties and functioned as local militia at best. They were clearly not regular Roman border troops.

     Yet there remains the question of how militarily effective the limitanei in fact were. We know that the best units of the limitanei were transferred to the comitatenses. The most physically fit recruits also went to the comitatenses, the less fit ones to the limitanei.¹⁰² From what the sources imply, the limitanei were also regularly cheated out of pay or rations by their officers.¹⁰³ By AD438 even the eastern government admitted that the limitanei ‘with difficulty repel the pangs of hunger on their meagre pay’.¹⁰⁴ Under such demoralising and unhealthy conditions, the limitanei probably did in fact decline, perhaps even severely, in combat effectiveness.¹⁰⁵ Yet, as shown above, they were still regular, organised troops.

     Their single greatest military failure which we can pin down was perhaps their inability to contain the Rhine crossings of the winter of AD405. At this point it would be difficult to try to exonerate the limitanei from giving a poor account of themselves because we simply do not know how many of them there were on the eve of AD406. The Notitia (see also Fig. 3) attributes to the Duces of the Sequanici, Armorica, Mogontiacensis and the Belgican Tract 26,000 border troops, and ten more units of pseudocomitatenses – but these are only paper strengths. We do not know how many of these units were duplicated from the existing border troops, or the size of a unit of pseudocomitatenses.

     We can guess that the heavy fighting probably decimated the limitanei in Gaul,¹⁰⁶ and that many of the surviving units were promoted to pseudocomitatenses to meet continuing threats, while hastily-raised milites took their place. Is this evidence of the failure of the limitanei as combat troops? Probably not; firstly since the emperor saw fit to promote such troops to the elite field army, the limitanei must not have been too poor as soldiers – otherwise he could have simply hired or promoted more foederati into Roman service. Secondly the Rhine crossings of the winter of AD405 were unprecedented – enormous forces were on the move. It is unknown how many people were on the move, but between the Sueves, Alans and Vandals, Heather has guessed that there could have been hundreds of thousands of migrants – for every five of them, one was a warrior.¹⁰⁷ The limitanei, posted in units 1000 strong at most, were supposed to deal with small-scale, endemic barbarian raids, and their failure to contain this migration was due less to tactical ineffectiveness than an out-of-control situation.¹⁰⁸

     To end this general defence of the limitanei, it is fitting to note that while there is no specific mention of the last comitatenses, two anecdotes from the Life of St. Severinus by Eugippius mention soldiers who are probably limitanei, or at least their descendants. Severinus was a holy man in Noricum, which in the Life had become cut off from the imperial centre, or was well on its way to becoming so (therefore the soldiers described by Eugippius could be attributed to the mid to late fifth century). Despite this apparent decline, the local soldiers continued to do their jobs: the tribune Mamertinus and his soldiers continued to fight off barbarian ‘robbers’, while even when Roman military presence from the imperial centre faded, and border troops slowly withered away for lack of pay, ‘the troop at Batavis held out’. In the end however, even these soldiers disappeared; the last we hear of them is that a detachment was killed en route to Italy, to fetch their unit’s pay.¹⁰⁹


Was barbarisation of the Roman army a major cause of the end of the western empire in AD476? The theory is an attractive way to explain what might otherwise seem inexplicable. From our analysis of the sources, however, it seems the end of the empire cannot be explained by barbarisation of the army. The regular army of the fifth century cannot be proven to have been qualitatively, or even quantitatively, very different from that of the Principate. Germanic soldiers were employed much as they had been for hundreds of years (albeit on a much larger scale by the fifth century), yet they do not come across as unreliable in the sources. Neither do the limitanei seem to have been ineffective peasant militia. The only ‘barbarisation’ which took place came perhaps only in the last decade before 476, when the regular army was effectively replaced by units of foederati; but again it is difficult to prove that this was the case.


¹ Ferrill’s ‘The Fall of the Roman Empire’ argues strongly for this view; in Ferrill’s opinion the barbarisation of the army made it incapable of defeating its barbarian enemies.
² It is true that Vegetius wrote during the fifth century, and that his manual, De re militari, could therefore be seen as a reflection of the Roman army of his day. It is equally true however that Vegetius was not a military man. Criticism of the soldiery was also a common Roman literary topos (Livy had accused the conquerors of Asia of weakening the soldiery with luxuries), and its appearance in a fifth-century civilian manual is not convincing proof of any serious deterioration in Roman military discipline.
³ Also it is interesting to note that while Synesius berates Roman soldiers he generally praises barbarian soldiers, the supposed source of the army’s weakness; see Letters 78 (the Unnigardae) and 110 (the Marcomanni).
⁴ In Occ. 5 of the Notitia, there are 9 auxilia palatina with the epithet Honoriani, probably raised by Honorius or even Constantius III or Valentinian III, and named in honour of their predecessor. Jones (1973, 353) gives several other units as examples: the Equites Constantes Valentinianses Seniores (Occ. 6.52), raised perhaps by Constantius III and named after himself and his son; and the Placidi Valentinianici Felices (Occ. 7), probably raised sometime around AD420 and named after the empress and her son.
⁵ 1973, 351-53.
⁶ Southern and Dixon 2000, 4-38 contains a good discussion of this widely-accepted view.
⁷ 2.34.
⁸ Zos. 2.15; also note that many units of auxilia palatina in the Notitia bear Germanic names.
⁹ Diocletian probably laid much of the groundwork which made Constantine’s reforms possible; in any case the army that emerged after Constantine’s reign was very different from the one at the start of the Tetrarchy.
¹⁰ Notitia, Occ. 5.
¹¹ This ranking is based on the work of Jones (1973, 348). In the magistri list itself, palatine legions come first, followed by auxilia palatina, field legions, pseudocomitatenses, palatine cavalry, field cavalry, and finally the limitanei.
¹² The practical effects of a demotion are not entirely clear, though it probably meant less pay or fewer rations.
¹³ The regionalisation of the comitatenses seems to have already begun under Constantine, who, if we are to trust Zosimus (2.34), billeted them in towns which had no need for protection. Also Julian’s Gallic troopers almost mutinied upon learning that Constantius intended to transfer them eastward (Ammianus 4.9) – they were probably very much settled by the late fourth century.
¹⁴ Jones 1973, 355-56, 612.
¹⁵ Though the use of vexillations had begun as early as the first century AD, when Nero deployed vexillations in his eastern campaign – these probably started off as detached legionary cohorts, which had self-contained command structures, and could operate semi-independently. For the sake of clarity, I have not italicised these early vexillations to differentiate them from the cavalry vexillationes of the fifth century.
¹⁶ Under the Principate, milliariae had been double-strength auxiliary units, which according to Jones numbered up to 1000 men (1973, 680,1280); the milliariae of which we know numbered somewhat less. Examples include the Ala Milliaria, with 760 cavalry, and the Cohors Milliaria Peditata, with 800 infantry (Goldsworthy 1996, 22).
¹⁷ One example is a detachment from the XII Gemina, which Jones (1973, 681) suggests had accompanied Diocletian to Egypt in 295. Nearly a century later it is found in the Notitia under the Comes of Egypt (Or. 28), listed as the Legio Tertiadecima Gemina – a full legion.
¹⁸ Jones, 1973,1280.
¹⁹ 18.9.3-4.
²⁰ Jones however suggests that the small size of the legions could have been due to attrition in battle, for it had been a long war (1973, 682).
²¹ 20.4.2 (Julian); 31.11.2 (Gratian); 31.10.13 (Valens) – note that in the third anecdote Ammianus writes ‘per legiones’.
²² 5.45.
²³ Zosimus ascribes to them ‘strength and discipline’, qualities one could readily associate with the steadfast legionary footsoldier.
²⁴ 3.3.4.
²⁵ Or. 18.56 – the vexillum (from which the vexillatio derived its name) was a cavalry standard; see Woods, ‘On the Standard-Bearers at Strasbourg’, Mnemosyne Vol. 50, Number 4, 1997, 479-80.
²⁶ Mag. 1.46.
²⁷ 6.8.
²⁸ The Green and Chaplin translation (London, 1814) puts the figure at six thousand.
²⁹ This is probable, given that the Heruli, Batavi, Celtae and Petulantae, all units of auxilia palatina (Notitia, Occ.5), appear also in Ammianus account (20.4) and are similarly styled as ‘auxiliaries’.
³⁰ Jones also explains how these men were used to man the walls of Ravenna – therefore they were probably infantry (1973, 682).
³¹ Occ. 5.
³² Libanius, Or. 30.6; Zos. 2.15.1. Both writers had personal reasons to associate the Christian Constantine with uncivilised barbarians, and their words are probably not entirely true.
³³ 1973, 680.
³⁴ Or. 34.32; 37.25, 31; 38.27.
³⁵ Most auxiliary formations were not up to strength: the I Augusta Lusitanorum had 477 men, while the I Apanenorum had 434 (Goldsworthy 1996, 23).
³⁶ Note the existence of a large fort at which the numerus exploratorum Germanicianorum Divietiensium were stationed. The fort could have easily held 1000 men (Southern 2006, 124); but numeri had always been ad hoc units, and in any case, given the massive losses in military manpower incurred within the first decade of the fifth century, it is doubtful that a numerus would have in fact numbered 1000 men.
³⁷ Jones 1973, 679; even then however these figures are very likely exaggerations, particularly the conspicuous ratio of roughly 1:2 for Constantine and Maxentius’ forces. Even so we can say that the Constantinian army was smaller than the one left behind by Diocletian.
³⁸ Lydus: De Mensibus 1.27; Zosimus: History 2.15; Agathias: Histories 5.13.7; Lactantius: De Mortibus 7.2; Lee 1998, 219.
³⁹ In Figs. 1, 2 and 3, a unit of auxilia is taken to be 500 strong for the sake of simplicity.
⁴⁰ Although there is no proof of a ‘standard’ size of a unit of the pseudocomitatenses, here I have assumed that the all-infantry pseudocomitatenses (which certainly contained legions – several units bear the epithet seniores or iuniores), had an average strength of 1000 men.
⁴¹ Excluding the classes, which were fleet troops; also for the sake of simplicity, units of limites, milites, numeri, laeti, gentilia and unknowns are taken to be 500 strong.
⁴² Excluding five units of scholae under the Magister Officiorum (Notitia, Occ. 9), totalling perhaps 2500 men.
⁴³ Appian, BC 2.110.
⁴⁴ Velleius Paterculus 2.113.1.
⁴⁵ Ammianus 22.12.6 – note the presence of the Gallic units Celtae and Petulantes at Antioch.
⁴⁶ Ammianus 25.10.
⁴⁷ Annals 15.26.
⁴⁸ Macdowall, 2001, 26.
⁴⁹ Zos. 5.26.
⁵⁰ Jones (1973, 684); the number of ‘auxiliaries’ seems to be entirely conjectural.
⁵¹ Frontinus, Strategemata 4. 2.3 – Corbulo held back an entire Parthian force with ‘two legions and very few auxiliaries’ – the asymmetry between the strengths of the two forces is probably an exaggeration.
⁵² 15,000 men would have amounted to around 3 legions – the same size as the force Varus took with him to Germania in AD9. Given Augustus’ reported anguish at the loss of Varus’ forces (Suet. Augustus 23.2), 15,000 men must have been a considerable number even under the Principate.
⁵³ ‘Seniores-Iuniores in the Late-Roman Field Army’, The American Journal of Philology 93 (1972), 253-78.
⁵⁴ Jones 1973, 645; Synesius also implies that pocketing troops’ pay had become the norm (Letters 62) – though as always one must treat his words with some suspicion. Procopius also records this practice in his day (Secret History 24).
⁵⁵ One unit of pseudocomitatenses, the Septimani, is probably a duplicate of the Septimani seniores, a field legion. Evidently the Septimani were demoted to pseudocomitatenses, yet remained on the magistri rolls both as a field legion and a unit of pseudocomitatenses.
⁵⁶ In the cavalry lists of the distributio, the equites cornuti seniores have been listed in Gaul and Britain; our old friends the Septimani are listed as a whole unit in the distributio three times, in Gaul, Italy and Tingitania, sometimes as a field legion, sometimes as a unit of pseudocomitatenses.
⁵⁷ Elton (2007, 285) gives an example of the ala I Hiberorum, which Duncan-Jones (1990 [1978], 105-17, 214-21), through analysis of payrolls, claimed was effectively only 100 strong in AD300. Under-strength was not unique however to the late empire; an extract from the Vindolanda tablets dating to May c.AD95 (II 154), records that of the 752 strong First Tungrian cohort, only 265 men were active. In any case Elton claims that in the fifth century 800 men per legion and 400 per vexillatio was realistic (1996, 89) – cf. footnote 16.
⁵⁸ Jones 1973, 686.
⁵⁹ Most of the officers mentioned by name in Ammianus’ account have Germanic names: Agilo and Scudilo (14.10.8), Malarich (15.5.6), Bainobaudes (14.11.14) and Laniogaisus (15.5.16), among others.
⁶⁰ Late Roman pictorial evidence also clearly attests to the existence of armoured Roman infantry, which demolishes Vegetius’ oft-quoted claim that the infantry had abandoned body armour by his day (De re militari 1.19).
⁶¹ cf. above.
⁶² Zosimus (2.15) mentions Germans, Gauls and Britons serving under Constantine.
⁶³ ‘Out of place’ tribally-named units include the Mauri Honoriani Seniores found in Illyricum, the Batavi Iuniores in Gaul and the Celtae Iuniores in Africa (Not. Occ. 7) – it certainly would have been difficult to keep the latter supplied with Gallic recruits!
⁶⁴ This was of course quite different from contemporary foederati, who did serve under their own officers. Hiring barbarian soldiers under their own officers, and having to quash their consequential revolts, was a tradition dating back to the Principate: the Pannonian revolt, the Varian disaster and the Batavian revolt were all the result of treachery by barbarian officers in Roman service.
⁶⁵ Letters 78 (the Unnigardae) and 110 (the Marcomanni).
⁶⁶ Jones 1973, 621.
⁶⁷ 20.8.13 – Julian offers to supply Constantius with Germanic recruits to avoid having to confront him for the purple.
⁶⁸ Amm. 14.10.8; interestingly Ammianus adds, ‘That at any rate was what some believed’.
⁶⁹ Amm. 16.12.2 (the deserter is not stated to have been an Alamann or even a German, but given his position in the imperial guards and his decision to go over specifically to the Alamanni, he probably was); 31.10.3 for the guardsman.
⁷⁰ Southern 2006, 260.
⁷¹ These included not only Stilicho, but also the Pannonian ‘soldier-emperors’ of the third century.
⁷² Amm. 15.5.16.
⁷³ Jones 1973, 199.
⁷⁴ Aetius’ near extermination of the Burgundian federate groups in AD436 was an exceptional case (Chronica Minora, 1.475).
⁷⁵ 31.16.8.
⁷⁶ 1973, 621.
⁷⁷ 7.40.7 – they were then called Honoriaci.
⁷⁸ Letters 78, 110.
⁷⁹ Jones 1973, 612.
⁸⁰ Southern and Dixon, 2000, 55.
⁸¹ Secret History 24.
⁸² ‘these latter, even when they are much superior to the enemy in numbers, never yet gave battle with courage’ (78); ‘We always wait for our soldiers to defend us, and a sorry help they are!’ (125); ‘the soldiers were hiding themselves in the gorges of the mountains to take care of their precious lives’ (122). Admittedly these refer to soldiers of the east, but given the impassioned bitterness in Synesius’ writing, as far as he was concerned these criticisms were probably equally applicable in the west.
⁸³ Letters 78 (the Unnigardae) and 110 (the Marcomanni).
⁸⁴ Alexander Severus 58.4.
⁸⁵ Cod. Theod. 7.4.14, 365.
⁸⁶ The final law demanding hereditary military service dates from AD398 (Cod. Theod. 7.12.12, 398), though military service was likely to have remained hereditary for many, as it had been since the Principate.
⁸⁷ Jones 1973, 650.
⁸⁸ ‘ad universos veteranos’ – Cod. Theod. 7.20.3, 320.
⁸⁹ Cod. Theod. 7.4.30, 409.
⁹⁰ Th. II, Nov. xxiv, 443 – their land was known as the agri limitanei.
⁹¹ Jones 1973, 651.
⁹² Cod. Theod. 7.1, 398.
⁹³ Elton (1996) makes a case of the limitanei being employed as customs officials; though this is not impossible, my guess is that the burgarii, who were definitely not combat troops, carried this out wherever they were stationed instead of the limitanei. Also, like the magdolophylakes of the east the burgarii perhaps also served as police or local militia.
⁹⁴ Occ. 42
⁹⁵ Germania 28.
⁹⁶ Historiae 1.78.
⁹⁷ 1973, 652; Jones quotes the inscription from IRT 880.
⁹⁸ This was one of the possible ranks of the praepositi listed in the Notitia.
⁹⁹ 29.5.21 (Mazices’ prefect); 29.5.35 (Theodosius).
¹⁰⁰ Aug. Ep. 46.
¹⁰¹ Jones 1973, 652.
¹⁰² Cod. Theod. 7.22.8, 372.
¹⁰³ The source du jour is Synesius’ tirade against Cerealis, who not only sold the mounts of the Balagritae horse archers for personal profit (Letters 132) but whose sole purpose was to make money (130). Despite Synesius’ bias, his accusations are not unique: Themistius (Or. 10.135) claims that frontier troops had lacked clothing and weapons before Valens became emperor, while Ammianus (28.6.16) writes of the notary Palladius pocketing the troops’ pay which he was supposed to himself deliver. Despite some exaggeration, the re-occurrence of peculations by officers in the sources suggests it was a common enough problem.
¹⁰⁴ Th. II, Nov. iv, 438.
¹⁰⁵ Th.II, Nov. xxiv, 443 stresses that the training of the limitanei should be improved. While this is quite strong proof that the limitanei had declined as effective combat troops, it is also evidence that they were reasonably effective enough for the government not to write them off completely.
¹⁰⁶ Jones (1973,198) claims that some two-thirds of the field army after AD406 were comprised of promoted limitanei; this gives an idea of how devastating the fighting was to the army’s manpower.
¹⁰⁷ ‘The Huns and the End of the Roman Empire in Western Europe’, The English Historical Review 110 (1995), 4-41.
¹⁰⁸ Jones (1973,198) defends Stilicho as well, claiming that the accusation that he had stripped Gaul of its defences were unfounded; even if the Roman forces in Gaul had stood at full strength, Jones argues, they would have hardly stood a chance against the migrating forces in the winter of AD405.
¹⁰⁹ Life 4 (Mamertinus); 20 (Batavian troops).


H. Elton, Warfare in Roman Europe AD350-425 (New York 1996)

H. Elton, ‘Military Forces’ in: P. Sabin et al (eds.), The Cambridge History of Greek and Roman Warfare vol. II (Cambridge 2007)

A. Ferrill, The Fall of the Roman Empire (New York 1988)

A. Goldsworthy, The Roman Army at War, 100BC-AD200 (Oxford 1998)

P. Heather, ‘The Huns and the End of the Roman Empire in Western Europe’, The English Historical Review 110 (1995), 4-41.

P. Heather, The Fall of the Roman Empire (London 2005)

A.H.M. Jones, The Later Roman Empire, 284-602 (Oxford 1973)

A.D. Lee, ‘The Army’ in: A. Cameron et al (eds.), Cambridge Ancient History vol. XIII (Cambridge 1998)

S. Macdowall, Adrianople AD378: The Goths Crush Rome’s Legions (Oxford 2001)

P. Southern, The Roman Army, a Social and Institutional History (New York 2006)

P. Southern and K.R. Dixon, The Late Roman Army (London 2000)

R. Tomlin, ‘Seniores-Iuniores in the Late-Roman Field Army’, The American Journal of Philology 93 (1972), 253-78.

D. Woods, ‘On the Standard-Bearers at Strasbourg’, Mnemosyne Vol. 50, Number 4 (1997), 479-80

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

Blog at

Up ↑

%d bloggers like this: