Fectio, Late Roman re-enactment society

What's New I Sitemap I Bibliography I Vortigern Studies l POLLS I LINKS l Games I Sitemaster
Vortigern I Wansdyke I Arthurian Collection I View Guestbook I Sign Guestbook l Webrings

Vortigern Studies > Fectio > Articles > Late Roman army and unit strength

Fectio Index



click here

It's all in the numbers -
The strength of Late Roman armies and units

By Robert Vermaat

A question often heard is how the armies and units of the late Roman army compare to those of the Roman armies of the Republic and the Principate. This article provides a short discussion of the numbers provided by the sources.

The Late Roman army

First of all, how large was the Late Roman army anyway? The Notitia Dignitatum , written in c. 394 AD and updated over the next two decades, actually lists the following units for both parts of the Empire:
12 scholae palatinae,
146 field army legions or legiones comitatenses,
42 border legions or legiones limitanei,
97 auxilia palatinae,
85 field army vexillations,
196 cohortes, auxilia and milites,
253 border units of cavalry (ale, cunei equitum and equites)

I will discuss the actual unit strength below, but if we assume that, on average, the scholae numbered 500 per unit, field army legions 1.000, border legions 3.000, auxilia palatina 800, border infantry units 300 and cavalry units 350, the total of units mentioned in the Notitia Dignitatum yields an army of about 450.000 men.
However, scholars such as Nicasie think this might be seen as a minimum rather than a maximum, to be taken as a more realistic number, as opposed to a 'paper strength'. The establishment strength could even be 650.000 if each cohort indeed was to be 500, and this number is what Agathius gives us for the 4th-c. army: 645.000 (Agath. V.13). Of course, it could well be argued that since each unit may have been understrength at many times during their existance, the lower number would probably be a good guess too.

Armies on campaign

Was there a 'typical' Roman army? Did Roman generals have an 'ideal' number in mind, depending on circumstances such as the season or logistics? Vegetius, talking about what he believe was practise during the Principate, thought 24.000 was the maximum for a field army in a large campaign (Veg Epit III.1.15).
Maurice/Maurikios thought a field army of 36.000 (24.000 infantry and 12.000 cavalry) was the upper limit for campaigns, and 5.000 to 15.000 average forces (Maur Strat II.4.28-33, III.8.31-40 and III.10-16-28. he clearly regarded 12.000 cavalry as many (XII.B.13.5-7, XII.B.8.33-36, XII.9.3-6, XII.13.5-7).

Of course, such theories could differ from reality. A few examples:

  • A field force of 38.000 is reported in the Balkans in 478 AD (Malchus fr. 18.2.12-23), being two forces combined: one of the regional armies under the magister militum per Thracias comprised of 10.000 infantry and 2.000 cavalry, while the other praesantal army comprised of 20.000 infantry and 6.000 cavalry.
  • In early 503 AD, a field force of 52.000 strong was created combining two armies of roughly 20.000 each under the two magistri militum preasentalis with 12.000 troops of the magister militum per Orientem. This number did not include either local limitanei units or Arab federates. Procopius assumed this was the largest Roman field army ever (Proc. Bella II.24.12-17). That he was quite mistaken can be seen in the list below.
  • Belisarius fielded 30.000 men at Dara, 530 AD (Proc Bella I.13.23), 20.000 men at Callinicum, 531 AD (Proc Bella I.18.5-7) and over 17.000 men in Africa, 533 AD, of which 5.000 cavalry and 10.000 infantry (Proc Bella III.11.11).

Numbers: Late Roman Armies on campaign :

188.000 - army of Maxentius, 312 AD (Zosimus II.15.2)
165.000 - army of Licinius, 324 AD (Zosimus II.22.1)
130.000 - army of Constantine, 324 AD (Zosimus II.15.2)
130.000 - army of Licinius at Chrysopolis, 324 (Zosimus II.26)
100.000 - army sent to Africa 457 AD (Procopius Bella III.6.1)
98.000 - army of Constantine, 312 AD (Zosimus II.15.1)
70.000 - army of the rebel Gildo, 398 AD (Orosius VII.36.12)
65.000 - main Roman army in Persia, 363 AD (Zosimus III.13.1)
50.000 - garrison of Egypt, 269 AD (Zosimus I.44.1)
35.000 - army of Licinius at Cibalae, 317/8 AD (An. Val. I.5.16)
30.000 - army of Macrianus, 261 AD (SHA, Gall. Duo II.4)
30.000 - army of the magister peditum Barbatio, 357 AD (Lib. Or. 18.49), Ammianus has it at 25.000 (AM XVI.11.2)
30.000 - army of Procopius, Persia 363 AD (AM XXIII.3.5)
30.000 - Roman army in Armenia, 543 AD (Proc. Bella II.24.16)
25.000 - army of Galerius in Persia, 298 AD (Festus Brev. XXV)
23.000 - army of Julian, 360 AD (Zosimus III.10.2)
20.000 - army of Constantine at Cibalae, 317/8 AD (An. Val. I.5.16)
20.000 - army of Vetriano, 350 AD (Jul. Or. 2.77B)
20.000 - army of Procopius, Persia 363 AD (Lib. Or. 18.214) though Zosimus has it at 18.000 (Zosimus III.12.5) and Malalas at 16.000 (Chron. XIII.21)
20.000 - Roman army in Mesopotamia, 531 AD (Proc. Bella I.18.5)
15.000 - Roman army in Africa, 533 AD (Proc. Bella III.11.2)
15.000 - Roman army in Illyria, 548 AD (Proc. Bella VII.29.3)

Unit strength

As referred to above, a lot can depend on what we determine was the actual unit strength for each Late Roman unit. Nicasie estimates the following Late Roman unit strengths:

  • scholae - 500
  • legiones (comitatenses) - 1000
  • legiones (limitanei) - 3000
  • auxilia palatinae - 800
  • infantry units (limitanei) - 300
  • cavalry units (limitanei) - 350

The traditional legion had been 6000 strong. Johannes Lydus also mentions this high number (De Mag. I.46). Vegetius mentions two very strong legions of 6000, but he stresses that legions 'in his day, although the name legion still existed, were much smaller in size.' (I.17, II.3). Vegetius' 'old legion' numbered 10 cohorts, the first numbering 1105 infantry and 132 cavalry, the others 555 and 66, totalling 6100 infantry and 726 cavalry (II.6). However, it is generally assumed that if this organisation ever existed, it probably belonged to the early 3rd century.

It was Mommsen who first noticed that old style legions were probably broken up into 6 detachments of a 1000, each of these commanded by one of the 6 tribunes of the old unit. Only the legions of the limitanei were still commanded by praefecti. Only Nischer proposed that 'new' legions existed of two units of 500, drawn from every border legion, but his theory lacks evidence. However, the Beatty papyrus mentions detachments of 500 men. One vexillatio consisted of 1000 men from legio III Gallica and legio I Illyrica. Another, from legio II Traiana, consisted of two 500-men units. Two vexillationes of legio III Diocletiana numbered 1100.

Border legions seems to be much larger than field army legions, because they are usually divided over as much as 7 stations.

When in 359 AD 7 legions (including equites indigenae, two units of superventores, praeventores and comites sagittarii) were trapped at Amida, they numbered 20-25000 men together with the inhabitants and refugees (XIX.2.14). Two of these legions suffered 400 casualties during a sortie (XIX.6.11). If we assume there were 7500 civilians, this hardly allows for more than 1500 per legion, if the garrison was not a legion of 6000 or if there weren't even more civilians than assumed.

All in all, it seems that Late Imperial legions numbered between 1000 and 1200.

Infantry cohorts
During the time of Septimus Severus, a reorganisation apparently resulted in cohorts 550 strong (up from 480), possibly those meant by Vegetius (II.6). But see below.
Ammianus mentions that Julian sent detachments of 300 men from each unit to Constantine (XX.4.2). In 378, Gratian sent detachments of 500 men to storm the position of the Lentienses (XXXI.10.13). Also in 378, Valens sent detachments of 300 from each legion to meet the Goths at Adrianople (XXXI.11.2).
It is thought that some of these detachments never returned to their parent unit but retained the original name (hence the repetition of such names in the Notitia Dignitatum lists). It is also possible that many of the field army units were split up into iuniores and seniores, reducing the original strength of these units. But another view on this is that the iuniores were a cadre of the original (hencefort designated as seniores) unit, which was therefore doubled in the process.
Infantry units mentioned by Ammianus number 300, 500, 800, 1000 and 1500 (XVII.1.4, XVIII.2.11, XXIV.1.6, XIV.1.2, XXIV.6.4, XXV.6.13-15, XXV.7.3).
Johannes Lydus mentioned cohortes of 300 and vexillationes of 500 (De Mag. I.46).

Cavalry units
The old style alae numbered 500 and seem to have remained that way. On paper at least. On the basis on the Beatty papyri, Duncan Jones calculated that around 300 AD in the Thebaid (Egypt), a unit of equites was 121 strong, an ala 116 and a cohors 164. These may not have been complete units, or else very much understrength.
Nothing much else is known about cavalry unit strengths. The ala III Assyriorum was organized in old-style 11 turmae, giving it a possible strength of 350 (ChLA XVIII 660). Ammianus mentions that the cataphracti defeated at Strasbourg were 600 strong, which is echoed by Johannes Lydus who says that alae were that number, and turmae 300 but also 500 (De Mag. I.46). Ammianus also mentions two turmae at Amida numbering 700 together (XVIII.8.2). Procopius has various sizes, between 200 and 800 strong (800: Bella VI.5.1, VI.7.25-6). Some units are larger, between 1000 and 1500, but it is unclear if these are units grouped together, or maybe allied forces (1500: Bella V.27.22-3 and VII.34.42). Maurikios mentions cavalry units should be between 300 and 400, but in any case not less than 200 and not above 400; if understrength, they should be combined.

Finally, the Anglo-Saxon Chronicle mentions that in a battle in 457 AD, '4 units' of Britons were destroyed, while another version of the text mentions a number of 4000. (ASC. Laud Chronicle year 456, Parker Chronicle year 457).


Two main conclusions can be drawn of Late Roman units (Nicasie):
a) in practise, the actual number of troops fielded will have been lower than the paper strengths.
b) it seems possible that Late Roman units did not have fixed establishment strengths at all, but varied between a certain minimum and maximum according to need.



Secondary literature

  • Hoffmann, Dietrich (1969): Das Spätrömische Bewegungsheer und die Notitia Dignitatum, 2 vols., Epigraphische Studien 1, (Rheinisches Landesmuseum Bonn).
  • Nicasie, Martijn (1997): Twilight of Empire, the Roman Army from the Reign of Diocletian until the Battle of Adrianople, (Thesis Publishers Amsterdam).

VortigernStudies is copyright © Robert Vermaat 1999-2008. All rights reserved