A history of the art of war under the Roman Empire has not yet been written, for the simple reason that we do not possess an account by a good military historian of a single campaign between that of Thapsus (46 B.C.) and that of Argentorate (357). Josephus does indeed give a first-hand account of the Jewish war of 66-70, and took some trouble over military details, but his subject limited him to siege operations and street-fighting. The most valuable section in his work is a general sketch of the Roman army and its organization, and a description of the arrangement of troops on the march.311 Tacitus, on the other hand, who is forced by his subject to describe several campaigns, and remains in consequence our chief authority, cared nothing for the technical side of warfare, and does nothing more than record, as a rule correctly enough, details which he found in his sources.312

With strategy we need not concern ourselves, since the subject lies beyond the scope of this essay; but tactics require more consideration on account of the special position assigned to the auxilia in battle formation. From the scanty information given by our authorities it appears that in any regular engagement fought during the first two centuries the legionary infantry were still considered to be the chief arm and employed to deal the decisive blow.313 They occupied the centre of the line, and the light troops and cavalry—that is to say, the auxilia—were expected to do little more than protect them from a flank attack. This formation was employed at Idistaviso in 16,314 against Tacfarinas in 18,315 against Tiridates in 58,316 against Boudicca in 61,317 and at the second battle of Bedriacum in 69.318 It is also prescribed by Arrian in his ‘Order of Battle against the Alani’.319 The only considerable exception is the battle of Issos in 193, in which the legions on both sides formed the first line and were supported by the archers, who shot over their heads. Dio, however, expressly states that this formation was adopted because these armies were fighting in a narrow space with the sea on one side and mountains on the other, so that there was no need to detach a force to protect their flanks.320

There were, however, cases, particularly in warfare against barbarians, where the enemy would not meet the imperial forces in the open field, but took up a defensive position on ground where legionaries could not be employed with success. In these circumstances the auxilia formed the first line and began the attack, and only if they were driven back and pursued by the enemy did the legions come into action. The battle of Mons Graupius, in 84, was conducted on these lines,321 and similar tactics seem often to have been employed by Trajan in Dacia.322 In general, however, the auxilia play a very secondary rôle; we do not hear either of the cavalry being used to strike the decisive blow after the manner of Alexander,323 or of any such combination of archers and heavy infantry as we find in mediaeval warfare.

The subject, however, is still obscure, and it is more satisfactory to turn to the part played by the auxilia in frontier defence, concerning which the archaeological research of the past twenty years has established more certain conclusions. In the frontier policy of the first two centuries we can trace two opposing tendencies at work, each of which is reflected in the disposition of the troops and the duties required of them. At the death of Augustus the Empire had as yet reached hardly any of its natural boundaries, although by means of the system of client kingdoms and ‘protected’ tribes it was asserting its claims and intentions in much the same fashion as the powers of modern Europe are doing in Africa to-day. The first century therefore witnessed on almost every frontier a period of expansion of greater or less duration, in which the sphere of direct administration was gradually pushed forward until some physical or political obstacle was reached which necessitated either a halt or a forward policy on a much larger scale. Throughout this period military operations were always imminent; in Britain, for example, between 50 and 85, the garrison marched out almost every spring, either for a campaign or a military demonstration. In winter, therefore, or in times of peace, the frontier armies were so disposed as to be able to take the offensive at a few days’ notice. The legions often lay in pairs, while many of the auxiliary regiments, instead of being scattered over a wide area, as was the case later, were concentrated at a few strategic points. The extent to which this system was adopted varied, of course, with local conditions, and a few regiments always occupied more isolated positions, but as a whole the auxilia of a province were far more easily mobilized than later when each regiment had its own castellum. On the Rhine frontier Haltern and Hofheim furnish examples of these large hiberna, dating from the beginning and the middle of the first century respectively,324 and we find the same system continued for the defence of the Taunus district annexed by Domitian.325 There is, indeed, here a chain of forts on the frontier, but they are of small size, with an average area of only 1½ acres. The bulk of the auxilia lay some way behind the frontier, in forts which held some two or three regiments apiece.326 In Britain we have traces of a similar system at the same date. The ‘Agricolan’ fort at Barr Hill is a frontier post which would require some two centuries at most for its defence, while the early fort at Newstead, which was probably occupied from about 80 to 100 or later, could accommodate at least 1500 men.327 The essentially temporary nature of such hiberna is emphasized by the character of their defences, which usually consist simply of an earth wall or palisade, little more elaborate in construction than the vallum which an army in the field was expected to throw up round its camp after a day’s march.

In provinces whence archaeological evidence is not forthcoming, inscriptions indicate the same system. From Spain, for instance, we have an early inscription referring to an officer who held command over four cohorts,328 and a similar brigade of three cohorts appears at Syene in Egypt in the reign of Trajan.329 On the Danube frontier von Domaszewski has concluded from the epigraphical evidence that Aquincum and Arrabona each held two alae in the first century.330

This period of expansion may be considered to end with Trajan’s annexation of Dacia and his failure a few years later to execute a similar forward move on the Eastern frontier. With the accession of Hadrian a new policy begins, which advertised by the elaborate character of its defensive measures that the imperial government was firmly determined to renounce all further schemes of aggression, a determination which was adhered to until the power of decision lay no longer in Roman hands. The outward signs of this new spirit were the abandonment of the old hiberna, and the removal of their garrisons to stone forts of a new type, each arranged to hold no more than a single unit, which were placed at more or less regular intervals along the frontier instead of behind it.331 The auxilia, that is, were transformed from a potential field army into a frontier police.

This policy of passive defence depended, of course, for its success upon the extent to which the frontier could be made defensible. Fortunately by this date it lay for the greater part of its length along positions of great natural strength. The Rhine, Danube, and Euphrates, when guarded by a continuous line of forts and watch-towers,332 and patrolled by flotillas of guard-boats, formed a serious military obstacle to a raiding force, an obstacle even more dangerous to its retreat than its advance. The desert frontiers of Africa and Arabia were more easily defended, since the routes by which a hostile force could advance were limited in number and the defence could concentrate upon them, and the same of course holds true of the southern frontier of Egypt.

There were, however, districts where such natural obstacles did not exist, as in the case of the trans-Rhenane territory, which was divided between Germania Superior and Raetia, and the northern frontier of Britain, and here Hadrian had recourse to the expedient of erecting artificial barriers, which he hoped would serve the same purpose.333 In the former case the frontier was defended by a palisade and ditch, which were later supplemented by an earth mound in the German section and replaced by a stone wall in Raetia.334 On the British frontier, between the Tyne and Solway, the existing remains are those of a stone wall, although there are also traces of a wall of turf, which may have been an earlier work.335 A turf wall also defended the more advanced line between the Firths of Clyde and Forth, which was occupied between 140 and 180.336 The southern line in Britain in its most perfect form was guarded by a stone wall seventy-three miles long. This wall was between six and nine feet thick, and probably stood originally about twelve feet high. In front of it, except where the precipitous nature of the ground rendered such an additional defence unnecessary, ran a wide ∨-shaped ditch. At intervals of about every Roman mile stood a stone block-house, and between every block-house two towers. The mile-castles contained barrack accommodation for about fifty men, and reveal abundant traces of continuous occupation.337 The garrison of about eleven thousand men lay in stone forts of the ‘cohort’ size, the majority of which are actually on the line of the wall, although a few, which probably belonged originally to an earlier system of defence, are a short distance behind it. The average interval between the forts is some six miles, so that it was easily possible for each regiment to man the adjacent towers and mile-castles and retain a considerable force at head-quarters. In addition to the troops actually stationed on the line of the wall, there were other regiments in outpost forts to the north and in the forts which guarded the three roads leading south to the legionary fortresses of Chester and York. The ends of the line were also guarded against flank attacks from the sea by forts at South Shields and on the Cumberland coast. If, then, we include all troops within three days’ march of the wall, the total force available for its defence probably exceeded twenty thousand men. Taking also into our calculations the natural strength of the position, we may safely say that this was the strongest and best guarded of all the frontier barriers.

The trans-Rhenane frontier, which extends for over three hundred miles from Rheinbrohl on the Rhine to Eining on the Danube, was defended by the same methods, although in certain sections the forts were more widely separated and the garrison was proportionately weaker. There were also fewer troops within call immediately behind the frontier line. Here also, between the cohort forts, stone ‘Zwischenkastelle’ and ‘Wachttürme’ furnished additional safeguards.338

This whole system of frontier defence has been much criticized, and the limitations and possibilities of these artificial barriers must be carefully determined.339 To take the negative side first, they could not, of course, be defended against unexpected attack, like the walls of a town, unless the assailants were only a small raiding party numbering some twenty or thirty men. On the other hand, in spite of the parallel of the ‘Customs Hedge’ in India,340 it seems unlikely that fiscal considerations played any large part in determining the government on their construction. They doubtless acted, when built, as a check on smuggling, but the expense of their maintenance would have been quite out of proportion to the value of the trade done with the German or British tribes.

The first purpose which they served was to furnish a screen behind which patrols could march in comparative safety, both by day and night, and keep the whole line under constant surveillance. Thus the passing of a hostile force could be instantly reported by messenger or signal341 to the nearest castella, whence detachments could at once start in pursuit. Secondly, whereas the defenders nearly always had a mounted force close at hand drawn either from the alae or the numerous cohortes equitatae, the raiders would probably be unmounted, since their start would be lost if they delayed to fill up the ditch and make a gap in the barrier large enough for their horses.342 This barrier, too, had to be crossed again in retreat, and presented a very serious obstacle to a force encumbered with booty. Indeed, the defenders might reserve the great part of their forces for this moment, as is recommended by Byzantine military writers describing similar conditions.343

This sketch of the methods of defence employed applies more particularly to the German and Raetian frontiers. In Britain, more particularly on the southern line, it is probable that a more serious defence was intended. In the first place, the massive stone wall, on which the defenders could stand, was obviously stronger than anything on the trans-Rhenane section.344 Secondly, we have noted that the garrison was stronger than in Germany, and could be more easily reinforced. Moreover, even after the final abandonment of Scotland, forts were still held in front of the southern line. Netherby on the Esk, and Habitancium (Risingham) and Bremenium (High Rochester) on Dere Street, were occupied by cohortes miliariae equitatae well into the third century, and at the last two forts we find a numerus exploratorum attached to the regular auxilia.345 These strong outposts would have been able to check or harass the enemy’s advance and give warning to the garrison of the wall of any impending attack.

All these suppositions, however, both as regards Germany and Britain, are based upon the assumption that a raid would be the sole subject of the attacking force, and that it would not be too numerous to be dealt with by the garrisons of three or four castella. To a more serious invasion the resistance offered was much less effective. The legions, it is true, still remained in reserve, but they formed the only concentrated force at the disposal of the defending general, for the majority of the auxilia, scattered as they were along the entire length of the frontier, could not be quickly concentrated, and a provincial garrison can rarely have taken the field at anything like its full strength. The system also created serious difficulties when it became necessary to send troops from one province to the aid of another. Three regiments, for example, could not easily be sent from Germany to Pannonia, because each of them constituted an essential link in the chain of frontier defence. It became the practice, therefore, to form out of detachments drawn from several regiments a composite vexillatio in which efficiency must have been greatly diminished by lack of esprit de corps. A cavalry vexillatio of this type commanded by a certain Lollianus, probably during the Parthian war of Trajan, was drawn from no less than five alae and fourteen cohortes equitatae.346

In defence of the system it would probably have been urged that on every frontier the hostile forces were equally dispersed and far less easily concentrated, and that a combination of the Celtic or Teutonic tribes would be heard of long before it was ready for action. The existence of the league which attacked and for a time broke through the Danube frontier in the reign of Marcus was certainly known to the imperial government, and the local governors succeeded in delaying the crisis until the return of the vexillationes which had been sent to the eastern frontier, with whose aid they hoped to be able to cope with the situation.347 Their calculations were upset by the havoc wrought in the army by the plague which these troops brought with them. Even so the danger was eventually surmounted, and the frontiers were on the whole successfully maintained for nearly a century more.

But the full consequences of this system cannot be perceived without some consideration of the changes which it brought about in the conditions of military life and their effect upon the general morale and condition of the troops. A very important point to notice is their immobility. Already in the first century there was, except for the officers, no regular system of transfers, and only an important change in the military situation caused troops to be sent from one province to another. In fact such changes were frequent, and considerable transfers took place, particularly during the Flavian period and the wars of Trajan. From the accession of Hadrian onwards, however, such movements cease almost entirely. During the following hundred and twenty years hardly a legion changed its position and the auxiliary regiments remained almost equally stationary.348 We can trace regiments which remained literally for centuries in the same province and for the greater part of the time were in the same castellum. Of the twenty-one cohorts and alae which are mentioned by the Notitia Dignitatum as forming part of the garrison of Britain, fifteen are shown by the evidence of diplomata to have been in the province long before the end of the reign of Hadrian; and two more, which occur in a diploma of 146, are probably only not mentioned earlier because they were creations of that emperor and had consequently no veterans ready for discharge until after his death.349 Similarly the Cohors V Lucensium et Callaecorum was in Pannonia at least from 60 to 198, the Cohors I Hemesenorum from 138/46 to 240, and the Ala III Augusta Thracum from 148 to 268/71.350 The best instance, however, is that of the Cohors II Ituraeorum Felix. This regiment is placed by the Notitia in Egypt, and other evidence shows it to have been in the province in 147, 136, 98, 83, and probably 39.351 As this section of the Notitia seems to date, at the earliest, from the beginning of the fifth century the regiment was probably quartered in the same province for at least three hundred and twenty years.352

Evidence of continued stay in one castellum is naturally more difficult to find, but the way in which the names Ulpius, Aelius, and Aurelius follow one another on a series of inscriptions of the Ala I Ulpia Contariorum from Arrabona in Pannonia Superior suggests that the regiment remained there throughout the second century, and the title Antoniniana shows that it had not moved before the reign of Severus Antoninus.353 At the fort of Veczel in Dacia the Cohors II Flavia Commagenorum has left inscriptions dating from the reigns of Hadrian, Marcus, Septimius Severus, Severus Alexander, and Philip, which cover practically the whole period during which the province was in existence.354 In Britain a remarkable series of dedications from Amboglanna (Birdoswald), which has already been referred to, shows that the Cohors I Aelia Dacorum was stationed there from about 211 to 271.355

Had the practice of employing a secondary title derived from the name of the reigning emperor commenced before the third century it would probably be easy to prove stays of much longer duration. The figures given above must certainly be taken as a minimum. A second-century auxiliary could thus make himself at home in his quarters in the practical certainty that, with the exception of a few temporary absences as member of a vexillatio, he would spend the whole of his twenty-five years of service patrolling the frontier on each side of his castellum.

In considering the life which the frontier guards would lead under these conditions we must remember that the character of the auxiliary soldier in the second century had changed considerably since the force was first organized by Augustus. In the early first century enrolment in the Roman service had little effect on the levies of wild tribesmen who composed the greater part of the auxilia at this period. They might be organized in Roman fashion, but the military qualities which they displayed and their whole manner of fighting were inherited from their ancestors. Promptam ad pericula nec minus cantuum et armorum tumultu trucem is Tacitus’s description of a cohort of Sugambri employed in Thrace in the reign of Tiberius, and in like fashion the German cohorts of Caecina’s army shouted their war-songs and rattled their shields beneath the walls of Cremona.356 In the second century all this was changed: the progress of Romanization had raised the majority of the provincials, even in the frontier districts, to a level of culture which placed them far above their ancestors of three generations back, although they might still seem barbarous to a cultured Greek or Italian.357 In the conditions of the service there was nothing to prevent the auxilia from participating in this general advance, and the soldiers who spent the best years of their lives in these little frontier stations gathered around them all the amenities of provincial life which would have been found in any country town in the neighbourhood. On the sheltered side of the fort a civil settlement, technically known as the canabae, quickly sprang up, and soon contained as many inhabitants as the fort itself, if not more. It was here that the soldiers placed their wives and children, that retired veterans settled near their old comrades, and traders erected their shops. A bath-house or two and a few small shrines, particularly those dedicated to the popular military cults of Mithras, ‘the Unconquered Comrade,’ and Juppiter Dolichenus satisfied the highest material and spiritual needs.358

At the fort of the Saalburg, where such a settlement has been carefully explored, an area of something like seventy-five acres was covered with buildings and gardens. Shrines dedicated to the Mater Deum and to Silvanus and Diana have been found, as well as those of Mithras and Juppiter Dolichenus, and two others remain as yet unidentified. On the outskirts, here as elsewhere, lay the cemetery with its inscribed sepulchral monuments, the chief source of our information on so many points of military life.359 On the British wall no canabae have been so carefully explored as those on the German limes, which is the more to be regretted since the buildings are usually in a better state of preservation; but it is still possible to see near the fort at Borcovicium (Housesteads) the terraces on which a scanty crop was raised, while the remains of buildings extend down the hill from the fort at the top to a small Mithraeum in the valley.360 At Cilurnum an elaborate bath-house was erected for the use of the soldiers of the Ala II Asturum on the banks of the Tyne, and further excavation would doubtless show that it did not stand alone. Where excavations have not taken place the existence of these and other buildings is testified to by inscriptions. At a fort on the Lower Rhine we even find the praefectus repairing at his own cost the regimental clock.361

The married quarters mentioned above require a few words of explanation. Numerous critics of the Roman army have assumed not only that celibacy is a valuable military ideal, but that it was actually attained until Severus issued his famous edict permitting soldiers to marry while still on active service.362 Previous to this it is assumed that they had no relations with women but those of the least binding description. Seeck, indeed, has carefully explained that the ‘children of the camp’ could not have been reckoned upon as a valuable source for recruits, because the rate of mortality is notoriously higher among illegitimate than legitimate children.363 This theory is sufficiently refuted by the fact that, as we have seen, nearly fifty per cent. of the recruits for the Legio III Augusta in Africa were giving castris as their birthplace long before the reign of Severus.364 A recently discovered edict of Domitian has shown further that such unions were sufficiently permanent to be officially recognized by the government during a soldier’s period of service, although only legalized at his discharge.365 The effect of Severus’s edict was merely to anticipate this act and give legal sanction to existing and perfectly well understood social conditions. Practically the change was probably of small importance, since it seems fairly clear that married quarters were not allowed inside the fort walls after this edict any more than before it, nor were married men allowed to remain permanently outside. Cagnat has shown that the arrangement of the internal buildings of the legionary fortress at Lambaesis, which are proved by epigraphical evidence to have been still existing in the third century, is entirely opposed to such a supposition, and to the general theory, which has often been advanced, that from the time of Severus onwards such a fortress became merely a club-house and exercise ground for the greater part of the troops.366 These arguments are concerned only with the legionaries, but they are worth introducing because the erroneous views here discussed have often been made to apply to the army as a whole. In the case of the auxilia, indeed, there was never any justification for their acceptance. The evidence of the diplomata was always sufficient to show that even in the first century the auxiliary soldiers, like the legionaries, formed family ties during their period of service which were officially recognized on their discharge.367 The same picture is given by early sepulchral inscriptions, of which the following, from the Pannonian fort of Teutoburgium, may serve as an example:

‘Ti(berio) Cl(audio) Britti f(ilio) Valerio, dec(urioni) alae II Aravacorum, domo Hispano, annor(um) L, stip(endiorum) XXX, et Cl(audiae) Ianuariae coniugi eius et Cl(audiae) Hispanillae filiae vivis ex testamento Flaccus dec(urio) frater et Hispanilla filia heredes faciundum curaverunt.’368

This tendency towards matrimony was naturally intensified by the more settled life of the second-century auxiliary. The systematic investigation of the cemetery attached to one of these permanent garrisons reveals as orderly a family life as could be found in any country town of the peaceful inland provinces. The following inscriptions, which are drawn from different parts of the Empire, are but few among many which might be advanced to support this contention.

xiii. 6270. From Borbetomagus in Germania Superior:

iii. 10257. Teutoburgium in Pannonia Inferior:

iii. 10609. From Pannonia Inferior: exact provenance unknown:

These examples alone show how far from reality are Seeck’s licentious mercenaries and their neglected bastards. In fact the suggestion of many critics that celibacy is a valuable military ideal, which was attained, at any rate partially, until the relaxation of discipline by Septimius Severus, proceeds upon false lines. In a short service army, like those of modern European states, in which the whole time of the men is necessarily occupied in learning their military duties, such an ideal is practical enough. In the Roman Empire the adoption of a professional army with a service of twenty-five years put it beyond the power of any government to enforce such monastic conditions, and the facts of the situation were, as we have seen, never misunderstood by the imperial authorities.

This is, of course, far from saying that the resulting state of things was all that could be desired. The long service system is, on this account, open to serious objections in principle, and these objections are intensified when we consider the lines on which this system developed. The second-century auxiliary, encouraged by the settled conditions of his service to form matrimonial ties, with his wife and children comfortably settled just outside the fort walls, is perhaps a more satisfactory spectacle from the moral than the military point of view. Military service in the same regiment had not yet become actually hereditary, because the enfranchised son of the auxiliary was advanced a step in the social scale and enabled to take service in the legions. When, however, in 212 the Constitutio Antoniniana swept away a distinction which had long ceased to have any real basis in a difference of race or culture, this obstacle was removed.370 Two sepulchral inscriptions of the Cohors I Hemesenorum, so often referred to, illustrate this change. The first is erected to a veteran of this cohort and to two sons and a grandson who had taken service in the neighbouring legions I and II Adiutrix, while in the second we find the son of a veteran from the latter legion who has taken service in the auxiliary cohort.371

It has already been noticed that the system of frontier defence organized by Hadrian made it difficult either to concentrate rapidly the garrison of a province at one point, or to send reinforcements from one province to another. The more settled the auxiliary regiments became, and the more local ties they formed, the more difficult did it become to order any dislocation of troops on a large scale.372 In fact when Severus Alexander granted to the frontier garrisons any adjoining territory which had been captured from the enemy, insisting at the same time that their heirs could only inherit it on condition of military service, this act was the natural culmination of a long process of development which had transformed what had once been the finest field army in the world into a rural militia.373 Unfortunately just as this development was completed and the result stamped with the seal of official approval, the emperors of the third century found themselves faced by new military dangers of a type with which the old system was least fitted to cope.