In making a levy for the auxiliary regiments, the imperial government was under no obligation to be at pains to legalize its position. In an ancient state it was assumed, as a matter of course, that the government had the power to call upon every citizen, if need arose, to take his place in the fighting line. Even the privileged cives Romani were never freed under the Empire from the legal obligation to military service, however much they may have been spared in practice, so that there can have been little doubt about the position of peregrini. Only in the case of the civitates foederatae was the government theoretically required to limit its demands to the number of men stipulated in the original foedus.

So much for the position in theory; in practice, of course it was not to the interests of the government to raise troops without considering the susceptibilities of its subjects, more particularly since the inhabitants of those districts which would furnish the best soldiers would also prove the most dangerous rebels if the demands made upon them exceeded their endurance. One instance of the conciliatory policy followed by the early Empire has already been noted; the exemption of the Batavians from all burdens but military service flattered their pride and enlisted their clan-spirit effectually on the side of the Romans. Evidence of a similar policy is apparent in the selection of the ethnical titles borne by the majority of the auxiliary regiments. In spite of the obvious convenience of such a step it was unusual for all the auxilia raised in one province to form a single series with a uniform designation. Wherever the clan-spirit existed, the name of the clan was accepted as the official title of the contingent which it furnished to the imperial forces.158 In Tarraconensis, for example, while the more civilized part of the province was represented by the alae and cohortes Hispanorum, several of the wild tribes of the north and west, such as the Aravaci, Vardulli, and Vascones gave their name to the regiments which they supplied.159 The Gallic levies reveal a similar policy; while the contingents of the comparatively peaceful Lugdunensis seem to be covered by the general title of Galli, a list of the levies of Belgica contains the name of almost every tribe in that warlike province.160 Indeed it is probable that during the first years of the Empire many of these tribal contingents fought, like the Batavians, as allies rather than as subjects of Rome, and knew little of Roman training or discipline.

In the East the historic position of the great city-states of Syria received similar recognition. Among the numerous regiments of archers contributed by this province we can distinguish the contingents of Ascalon, Tyre, Antioch, and Apamea, as well as corps from Chalcis, Damascus, Hemesa, and Samaria, who represented the incorporated armies of the old client states.

The incidence of the levy upon different provinces can best be judged by a statistical table giving the number of regiments raised in each. This is not easy to construct owing to the confusion caused by the duplication of numbering, and the consequent danger of counting the same corps twice over, or of reckoning two corps as one. There were, for example, in Pannonia two cohorts, each bearing the title ‘I Alpinorum’, which can fortunately be distinguished from one another because they are both mentioned in the same diploma, but there are scores of similar cases which can only be decided as yet on a balance of probabilities. This extremely inconvenient system seems to be due to two causes. In the first place, when new regiments were raised some time after the original levy they seem to have begun a fresh series instead of being included in the old ones. This process can be followed most clearly in the case of regiments raised after 70, which were distinguished by a title derived from the name of the reigning emperor. Thus we have cohorts I and II Flavia Brittonum, I Ulpia Brittonum, I Aelia Brittonum, and I Aurelia Brittonum.161 Secondly, it seems probable that when newly-raised regiments were drafted into different provinces they were numbered in a different series in each province. This suggestion is supported by the fact that where a regiment bearing a high number is found, it generally appears that the rest of the series was originally stationed in the same province, whereas isolated cohorts generally have a low number. For example, the greater part of the Gallic levies were originally stationed on the Rhine. Consequently, we find few duplicate numbers and several series which run up to four or even higher. The Thracian regiments, on the other hand, on account of their special utility as archers, were distributed very widely throughout the Empire during the first century, and of the twenty-seven corps known to us, seventeen are numbered I or II, and are distributed over eight provinces.

Apart from this difficulty the following list contains in any case more regiments than ever existed at any one time. Fresh regiments must have been raised to fill the gaps caused by such disasters as the defeat of Varus and the rebellion of Boudicca, but in only a few cases can we distinguish the earlier from the later levies. It is only possible to put in a separate class those regiments which bear a title derived from the Flavians or later emperors, and were probably raised after 70. Still, if these limitations are borne in mind, the following table may serve to show approximately the quota which each province contributed:

The first point to notice in this list is the smallness of the contingent from the senatorial provinces. So small is it that Mommsen desired to see here evidence of a constitutional principle.179 The auxilia were ‘gewissermassen eine Hausmacht des Kaisers’ and as such raised only in the provinces governed by his legati. Such instances as were then known of regiments raised in senatorial provinces were, he thought, susceptible of explanation. The alae Vocontiorum, for instance, represented a civitas foederata which was not, strictly speaking, a part of the senatorial province of Narbonensis. It does not, however, seem possible to maintain this theory. The Cohors I Cretum180 is a certain case of a regiment from a senatorial province, nor can it be really doubted that contingents were also drawn from Cyprus181 and Cyrene.182 Indeed, it is difficult to see what legal or political obstacle should prevent Augustus and his successors from utilizing the military material available in the senatorial provinces. Even if Mommsen is right in believing that conscription, as opposed to the enrolment of volunteers, could only take place in a senatorial province with the authority of the Senate (and this theory is questioned by both Gardthausen and Liebenam),183 there is no reason why levies should not have been made for the auxilia under these conditions when they certainly were made for the legions.184 In no case did any military power remain in the hands of the Senate, since the recruits would immediately be marched away to garrison imperial provinces. As a matter of fact, the reason for the smallness of the senatorial contingent seems to have been a practical one. Few auxilia were raised from Narbonensis and Baetica, because the greater part of the inhabitants of these provinces had received the franchise and were consequently eligible for service in the legions. Achaia, Asia, and to a certain extent Macedonia, were treated as being on the same footing, partly because Greeks did actually serve in the Eastern legions, partly because of the Philhellenic policy of the imperial government, which would not deny to the Greek states, although they were technically unenfranchised, the privileges enjoyed by the enfranchised urban communities of the West. Also no doubt the Greek of the period was not rated highly as a fighting man. On the other hand, from Cyrenaica, Crete, Cyprus, parts of Macedonia, and Africa185 useful troops could be and were obtained. The way in which the system worked is shown by the case of Noricum, which, although an imperial province, included many enfranchised communities and contributed recruits to the Rhine legions in the middle of the first century.186 Its contribution of auxilia in consequence is limited to one ala and one cohort, as against the eighteen regiments furnished by the neighbouring province of Raetia.

In contrast to Narbonensis, it was upon the remaining three Gallic provinces that the levy fell most heavily. From this district came more than a quarter of the auxiliary infantry187 in the pre-Flavian period and nearly half the cavalry. The Gallic troopers indeed maintained for a century the reputation which they had won under Caesar’s command, and Strabo,188 writing in the reign of Augustus, places them above all other cavalry in the imperial army. Arrian,189 too, notes their reputation and the number of Celtic words in the cavalry drill-book, although in his day their position had been taken by the Pannonians, already prominent in the campaign of 69.190 Spain sent the largest contingent after the Gallic provinces, and also contributed a few words to the drill-book,191 but we hear nothing of the quality of the Spanish troops and they soon lost their early importance. The predominance of the auxilia of Spain and Gaul in the pre-Flavian period is, however, a clear indication of the determination of Augustus to base the Empire on its Western provinces. Archers alone, and these in comparatively small numbers, were drawn from the East,192 which was still regarded as the home of dangerous and un-Roman ideals.

Lastly, a word must be said about a group of regiments which do not appear in the above lists and are too numerous to be passed over. These are the cohorts which bear the titles voluntariorum civium Romanorum, ingenuorum c. R., Italica c. R., and campestris.193 Collectively these regiments constitute the cohortes civium Romanorum to the soldiers of which Augustus left by his will a donative equal to that of the legionaries.194 From various passages in the literary authorities it appears that they represent the result of two levies made by Augustus in Italy, the first during the Pannonian rising, and the second after the defeat of Varus.195 When free-born citizens could not be found in sufficient numbers the levy was extended to freedmen.196 This is corroborated by the evidence of the inscriptions, since the title ingenuorum clearly implies the existence of regiments whose members could not make this boast.197 Originally, as the provisions of the will of Augustus show, these cohorts occupied a peculiar position, and were practically on a level with the legionaries, in consequence of which their commanders bear the title of tribunus.198 The presence, however, of the Cohors VIII Voluntariorum on the Dalmatian diploma of 93 shows that unenfranchised recruits had been accepted even during the pre-Flavian period, and in the following century only their title distinguishes these regiments from the ordinary auxilia.

The evidence hitherto considered has mainly served to illustrate the original distribution of the burden of military service and the respective quotas furnished by the different provinces to the auxilia at the time of their organization. To trace the further workings of this system it is necessary to examine the principles on which the auxiliary regiments were distributed among the military areas and to trace the relations between this distribution and the method of recruiting.

A casual glance at the military diplomata, which give a fair idea of the composition of the more important provincial garrisons between the reign of Vespasian and that of Commodus, suggests that it was the settled policy of the imperial government to destroy the possibility of national cohesion and local sympathies among the regiments raised from their subjects by distributing the contingents of each recruiting district over as wide an area as possible, and making every frontier army corps a mosaic of different nationalities. It will be shown later that this theory, which has been frequently adopted by modern writers, will not stand before a closer scrutiny of the evidence as an explanation of the state of things existing in the second century; it can also be shown that such a principle of distribution was not the original policy inaugurated by Augustus.

Our earliest evidence relates to the composition of the garrison of the Danubian provinces, and the account of the great rising which took place here in the year A.D. 6 by the contemporary observer Velleius makes it clear that the strength of the rebels lay in the training which many had received in the Roman army. His reference to the military knowledge of the leaders and the discipline of the rank and file indicates that regular auxiliary regiments, raised locally and stationed near their homes, had mutinied in sympathy with their fellow tribesmen.199 Concerning the state of things on the Rhine frontier we have more detailed information which points to the same conclusion. The account in the Annals of the campaigns of Germanicus mentions cohorts of Raeti, Vindelici, and Gauls in addition to the tumultuariae catervae of the local militia.200 Later in the century we find an Ala Treverorum engaged in putting down a revolt of their own countrymen in 21,201 an Ala Canninefatium engaged in the disastrous expedition of L. Apronius against the Frisii in 28,202 and Vangiones and Nemetes helping to repulse a raid of the Chatti in 50.203 Finally, when we turn to the narrative contained in the Histories of the events of the disastrous year 69, we find abundant evidence that at this date three-fourths of the Rhenish auxilia were drawn from Gaul proper or the Teutonic tribes of Belgica. The only regiments mentioned by Tacitus which are not of local origin are (1) Thracians, who appear on every frontier owing to their special qualifications as archers;204 (2) Spaniards, who may have entered the province in 43 with Legio IV Macedonica, which was transferred from Spain to the Rhine to replace the troops sent to Britain, and (3) Britons, who probably began to arrive from the newly conquered areas a few years later.205 Epigraphical evidence adds to the list a few regiments from the Danubian provinces and some corps of oriental archers.206

In other provinces the same policy can be traced, although the evidence is less abundant. In Africa, for example, the deserter Tacfarinas seems to have served in his own province,207 and in Palestine we find Samaritan regiments garrisoning Caesarea.208 On the whole there is sufficient evidence to show that although each of the great frontier armies contained imported elements, in particular the ubiquitous Thracian and oriental archers, the original policy of the imperial government was to draw the auxilia in each case from the nearest recruiting-areas.

Both the advantages and the defects of this system are sufficiently obvious. It saved trouble, a reason which had already commended it to the administrators of the Republic, and it avoided the dangerous and widespread discontent which, as the case of the Thracians shows, would have followed any wholesale attempt to remove the newly organized regiments to distant provinces.209 Lastly, the men would be fighting on ground which they knew against an enemy with whose methods of fighting they would already be acquainted. On the other hand, there was of course the obvious danger that in a border war which assumed the character of a national struggle the local auxilia might desert to their own countrymen and use the training which they had acquired in the Roman service to increase the strength of the hostile resistance. As a set-off to this danger the Romans reckoned with some justice that tribal enmity was usually stronger than national feeling, and in fact there were many tribal chiefs like Flavus, the brother of Arminius, who were well content with the rewards and distinctions which recompensed their fidelity.210 Events, however, made it clear that this confidence was misplaced. A time was to come when the border tribes would identify themselves readily with the cause of imperial defence, but the influences which were to bring about this result were often slow in their operation, and the first century saw on almost every frontier a more or less serious outbreak of national feeling, in which the auxilia often participated. Yet even the most serious of these revolts, that of Civilis in 69, showed how the new leaven was working. The political conceptions of the mutineers were borrowed from their conquerors, not from their ancestors, and in the darkest hour of the revolt a Gallic cavalry regiment, the Ala Picentiana, was the first to return to its fidelity.211

The first district in which the Augustan policy broke down was the Danubian provinces, and a glance at the names of the regiments stationed here in the pre-Flavian period shows that the lesson of the great rebellion was not thrown away upon the imperial authorities. In Pannonia a diploma of the year 60212 shows us the following seven cohorts, I and II Alpinorum, I Asturum et Callaecorum, I and II Hispanorum, I Lusitanorum, and V Lucensium et Callaecorum, forming part of the garrison of the province, and we may add the Ala Aravacorum on the strength of an early inscription.213 In Dalmatia early inscriptions give the following cohorts:

I Campanorum Voluntariorum civium Romanorum. iii. 8438.

VIII Voluntariorum civium Romanorum. iii. 1742.214

III Alpinorum. iii. 8491, 8495, 14632.

I Lucensium. iii. 8486, 8492, 8494, 9834. All these must date before 80, when the regiment appears in Pannonia.

This list might perhaps be lengthened, but it is sufficient for our purpose. It is clear that after the rebellion Augustus imported into the disturbed area a number of regiments from other provinces, particularly from Spain, where the large garrison maintained during the earlier part of his reign could now safely be reduced. The Pannonian and Dalmatian regiments, on the other hand, were transferred elsewhere—several of them, as we have seen, to the Rhine, where they served to replace the troops who shared the fate of the legions of Varus.215

The same sequence of events took place on the Rhine in the years 69 and 70. The temporary success of Civilis was largely due to the wholesale defection to his standard of the Gallic and Teutonic regiments then stationed on the Rhine frontier. After the suppression of the rebellion in the summer of 70 a number of these regiments were disbanded or sent elsewhere,216 and their place was taken by the auxilia who had accompanied the new legions sent into the province by Vespasian. Of the 29 regiments which appear in the Rhine in the second century only 11 bear titles indicating a local origin, and some of these had probably not belonged to the pre-Flavian garrison but had only returned to their native country in 70 after a long stay in other provinces. It has been noticed, for example, that of the two veterans of the Cohors I Aquitanorum, to whom the diplomata of 82 (D. xiv) and 90 (D. xxi) were granted, one is a Thracian, the other a Galatian; further, that one of these diplomata was found near the site of the later town of Nicopolis ad Istrum, where the owner had presumably settled after his discharge. This suggests that the regiment had been stationed in Moesia and only returned to its native province in 70 with the Moesian legion VIII Augusta.

It is on these two frontiers, the Rhine and the Danube, that the transfer of troops can most easily be traced, because of the importance of the military events which caused it to take place. In other parts of the Empire other tendencies were at work during the first century which produced the same result in less noticeable fashion. One need only mention the steady drift of troops from the Danube to the East in the reign of Nero,217 and from the Rhine to the Danube a little later,218 and it is easily intelligible that the second-century army list shows few traces of the original policy of Augustus.

If, then, it were correct to assume that the title of an auxiliary regiment is always a correct index of its composition, it would certainly be justifiable to comment on the extraordinary mixture of nationalities in the frontier garrisons of the second century. Fortunately, however, the frequent mention of the origin of individual soldiers on diplomata and sepulchral inscriptions219 gives us the means of checking this assumption and of working upon a surer basis of fact. The following lists give the inscriptions of this type from Pannonia arranged in two groups according to their date, the year 70 being taken as the dividing line; that is to say, the soldiers mentioned in the first group were enrolled before that date. Some inscriptions which could not be dated with any certainty have necessarily been omitted, also others where there was reason to believe that the soldier mentioned was enrolled when his regiment was in a different province.220 To the second group, which illustrates the recruiting system from the Flavian period onwards, a list of similar inscriptions from Dacia has been added. In each case the title of the regiment is followed by the nationality or place of origin of the soldier, stated in the form given on the inscription, and by the name of the province from which he was drawn. For reasons which will appear later the evidence concerning the oriental regiments is omitted.

We may add here a recently discovered inscription from Samaria:

I(ovi) O(ptimo) M(aximo) mil(ites) v[e]xil(larii) coh(ortium) P(annoniae) sup(erioris) cives Sisc(iani) Varcian(i) et Latobici sacrum fecerunt. A. E. 1909. 235. 1910. p. 6.

The vexillation had presumably taken part in suppressing one of the Jewish rebellions in the first half of the second century.

The facts disclosed by these inscriptions are very significant. In the first list, as is natural, we find traces of the troops transferred into Pannonia from other provinces after the great rebellion. It is more important, however, to notice that before the end of the reign of Tiberius natives of the province were already being accepted for service in these imported regiments.228 In fact there is nothing here to suggest that any attempt was made to preserve the national character of these Spanish and Alpine corps by obtaining fresh drafts from the districts in which they were originally raised. Those recruits who do not come from Pannonia itself are drawn merely from the neighbouring provinces of Dalmatia and Thrace.

But it would perhaps be misleading to infer from this evidence alone that local recruiting was universally adopted in the first century, although it was certainly common. It is possible that in the Flavian period, when the memory of the rebellion of Civilis was still fresh, some attempt was made to check a national cohesion by combining drafts from different provinces in the same regiment. This at least is suggested by the nationalities of twenty-one soldiers of an auxiliary regiment which are recorded on a sepulchral inscription at Tropaeum Traiani in Moesia Inferior.229 This monument was erected in memory of men killed in action during one of the Dacian campaigns either of Domitian or Trajan, so that its evidence applies to the recruiting of the Flavian period. Twelve of these men came from the Lower Rhine, two from Lugdunensis, and three from Spain, while Raetia, Noricum, Britain, and Africa supply one each.230 In Pannonia, too, some Spanish soldiers appear rather mysteriously in an Ala Pannoniorum on two inscriptions which can hardly be later than the beginning of the second century.231 There are even traces of a similar policy having been pursued in the recruiting for the legions during the same period. In a list of seventy-six soldiers who were apparently enrolled in Legio III Augusta towards the end of the first century, we find men from seven different provinces.232 In any case, however, no attempt seems to have been made to preserve any connexion between an auxiliary regiment and the tribe from which its title was derived.

When we come to the second century there is no more room for doubt; for all cohorts and alae on the Pannonian frontier, leaving out of account, as before, the oriental regiments, local recruiting has become practically universal. Seventy per cent. of the recruits come from the two Pannonian provinces, the majority from the Azali and the Eravisci, tribes which never gave their name to an auxiliary regiment. Even the Thracian regiments, which might have maintained their original character without much difficulty, form no exception to the rule. In Dacia the exceptionally large auxiliary garrison233 could not be supported entirely by local levies, but the deficiency was mostly made up in the nearest available recruiting-grounds of Moesia and Thrace.

A few examples may be adduced from other provinces to show that the methods employed on the Danube frontier were not exceptional. In Germania Superior three soldiers of the Alae I and II Flaviae Geminae describe themselves as Baetasius, Elvetius, and Secuanus, and the Raetian diploma of the year 107 was granted to a Boian who had served in the Ala I Hispanorum Auriana.234 In Africa a soldier of the Cohors VII Lusitanorum gives ‘castris’ as his place of origin, as do the majority of the veterans discharged during the second century from the African legion III Augusta.235 Concerning the Eastern provinces we have very little evidence, but it may be noted that of the large number of regiments raised by Trajan in this part of the Empire the majority remained stationed in the East throughout the following century, and there is no reason to suppose that they were not kept up by local levies.236

The recruiting of the legions during the second century seems to have followed the same lines. The high proportion of men of Legio III Augusta in Africa who give ‘castris’ as their birthplace has already been noted. Similarly of 39 soldiers discharged from Legio II Traiana at Alexandria in 194, 22 come from the ‘castra’, 8 from the Greek towns in Egypt, and only 9 were not born in the province.237 Of 133 soldiers discharged in the following year from Legio VII Claudia stationed at Viminacium, 104 come from Upper or Lower Moesia, and of the remainder all but one come from the Danubian provinces.238 Further evidence on the recruiting-area of the auxilia during this period can be obtained from another source, the inscriptions of the Equites Singulares Imperatoris. This corps, which seems to have been raised towards the end of the first century, possibly by Domitian,239 formed thenceforward a part of the imperial guard, and was stationed at Rome. It was recruited mainly from the same area as the auxiliary alae, and a certain number of the men were selected from them.240 On a hundred epitaphs of members of this corps who recorded their place of origin, the provinces are represented in the following proportions:241