INTRODUCTION
Nowadays the Ninth Legion is generally thought – in academic literature, at least – to have been transferred from Britain before its disappearance in unknown circumstances. Any person consulting scholarly general works on Roman Britain over the last couple of generations could easily get the impression that the issue is resolved, at least at the British end.Footnote 1 On the assumption that it had already left the province, the Ninth Legion does not play a role in recent discussions of possible war in Britain in the 120s or of the building date and sequence of Hadrian's Wall.
It will not be argued here that it can be proven that the Ninth Legion was lost in Britain. This paper does not seek to prove that particular things happened at a particular time, merely to point out a possible reconstruction of events that does not involve the departure of the Ninth Legion from the island. It might have been entitled ‘An unnecessary departure’, in the spirit of Michael Jarrett's ‘An unnecessary war’.Footnote 2 Lawrence Keppie has ploughed a lonely furrow of scepticism about the proposed extended life for the legion outside Britain;Footnote 3 the present paper seeks to build on his work to show that it is circumstantially most likely that it met its end in Britain and to suggest that discussions of events in Hadrianic Britain – of which there have recently been many – should no longer take the permanent transfer of the Ninth Legion out of Britain before its disappearance – and before the building of Hadrian's Wall began – as an a priori fact.
THE NINTH LEGION
Legio IX Hispana, whose last dated attestation is in Britain at its fortress in York in 107–08,Footnote 4 had ceased to exist by c. 165, when a list of the legions was set up in Rome.Footnote 5 Once it was supposed that the Ninth was lost in Britain at the outset of Hadrian's reign (117–19), on the basis of a literary statement that at Hadrian's accession ‘the Britons could not be kept under Roman control’Footnote 6 and the apparent replacement of the Ninth within a few years by legio VI Victrix, known to be permanently based in Britain under Hadrian. The loss of the legion in Britain early in Hadrian's reign is no modern novelist's invention, but was deduced as a possibility by a succession of eighteenth- and nineteenth-century epigraphic scholars, from Horsley in Britain to the continental giants Borghesi, Hübner and Mommsen.Footnote 7 A coin issue depicting Britannia was once firmly dated to 119 and associated with the supposed war in Britain, but not all would now date this coin so closely or confidently.Footnote 8
Almost a century ago it was realised that the Ninth cannot have ceased to exist quite so early as that. The problem lay in the career inscriptions of two of its officers, thought most unlikely to have served in the legion as early as 117–19; this was first observed by Emil Ritterling,Footnote 9 the point being developed by Eric BirleyFootnote 10 with the addition of further relevant careers. Ritterling did not see this as necessarily disproving that the legion was lost in Britain, merely that the event in question might have occurred later in Hadrian's reign than 117–19. The data on careers which indicate that the Ninth was in existence later than 117–19 are set out separately in appendix 1. Appendix 1 also reviews the epigraphic evidence taken to show that the legion existed beyond the 120s.
Loss in Britain was still an option, at first, for Birley, who in 1948 considered the possibilities that the Ninth had been transferred out of Britain during Trajan's Parthian War (114–17) or, having stayed in the province during the building of the Wall and the legionary garrison having been strengthened for that building programme by the addition of VI Victrix, transferred to the east c. 132 with Julius Severus – perhaps to be destroyed in the second Jewish War of 132–35 – or, just possibly, lost in a ‘second British rising’, later in Hadrian's reign.
The 1960s saw the publication of two tile stamps of the Ninth Legion found at Nijmegen, a legionary fortress on the Rhine frontier of Lower Germany (actually on the tributary river Waal, 12 km behind the Rhine). There are other traces of the legion in the Lower Rhine area (listed and discussed, along with the Nijmegen evidence, in appendix 2). This information strengthened the feeling that the legion was transferred out of Britain at some point before its loss. J.E. Bogaers argued that the Nijmegen material was best placed in the Hadrianic period and suggested that the legion had been moved out of Britain to Nijmegen to recover from losses suffered in the warfare early in the reign.Footnote 11 Following Bogaers, E. Birley in 1971 argued more firmly that the legion had left Britain, postulating that, with the Wall completed and ‘VI Victrix … comfortably installed at York’, the Ninth was transferred to Nijmegen c. 126 and picked up a few years later by Julius Severus on his way to the Jewish War of 132–35. In 1972 Werner Eck, discussing a diploma which seemed to show that a former military tribune of the Ninth was consul in 161 and could not therefore have served in the legion before about 135–40, added weight to another suggestion by E. Birley, that the legion might have been moved east to be destroyed by the Parthians at Elegeia in 161.Footnote 12
Bogaers's dating of the presence of the Ninth at Nijmegen is no longer universally accepted and was not based on any precise dating evidence (see appendix 2). There is nothing about the Ninth Legion tiles at Nijmegen or their context that dates them to the Hadrianic period.Footnote 13 J.K. Haalebos preferred to see the legion's traces there as connected to the vexillatio Britannica also attested at this fortress and thought this best placed between the withdrawal of X Gemina from Nijmegen in 103/4 and the beginning of Hadrian's reign, when part or the whole of XXX Ulpia may have replaced it.Footnote 14 The implication of this would be that only a detachment of the Ninth was present (as it was building at York in 108) and that the whole of the Ninth was back in Britain by the early years of Hadrian's reign, legio XXX moving to Nijmegen around 119.
Among the traces of the legion on the continent is a dedication by its praefectus castrorum (and sometime primus pilus) at the healing sanctuary of Apollo Grannus at the spa of Aachen (see appendix 2). The presence of this camp prefect, unlikely to have been commanding a vexillation, is sometimes taken as indicating that the entire legion was on the continent. However, there are other possible explanations. L. Latinius Macer might have been commanding a detachment as primus pilus and the altar might celebrate the news of his promotion. Alternatively, Macer may not have been on active duty when the dedication was made; he may have visited Aquae Granni on his way home after retirement from the legion.
More recently A.R. Birley has tentatively suggested that part of the Ninth was with the vexillatio Britannica in Nijmegen between 104 and 117, but returned to Britain after Hadrian's accession, and that the legion was there until removed by Julius Severus in the 130s, to travel, via the Jewish War, to the east and eventual destruction at Elegeia.Footnote 15
COULD THE NINTH STILL HAVE BEEN IN BRITAIN IN THE MID-120S?
The evidence for the presence of the Ninth in Nijmegen has raised the possibilities that part of it was withdrawn from Britain in c. 103/4 (to plug a gap on the Rhine left by legions moved east to the Dacian War) or part or the whole of it from 114 (ditto the Parthian War), and never returned. The record of samian stamps from York has been interpreted as indicating a fall-off in occupation in the period 110–25.Footnote 16 In this view, the Sixth Legion arrived (traditionally with Hadrian in 122) to restore Britain to a three-legion province.Footnote 17 It has always seemed difficult to believe that the northern frontier of Britain was deprived of the entire Ninth Legion, despite Trajan's preoccupation with other parts of the empire. Haalebos’ suggestion offers a possible solution to the problem: under Trajan a mere vexillation might have gone to Nijmegen, making up (with auxiliary units) the vexillatio Britannica. A draft from the legion in Britain closest to the northern frontier is not as surprising as it may seem: York was the most conveniently situated of the British legionary fortresses for sea transport between Britain and the Rhine. Also relevant is the possibility that a detachment of the Ninth lay in this period at or near Carlisle, where its tile stamps, some locally produced at a works depot at Scalesceugh, have been found.Footnote 18 These, in common with the Nijmegen examples, use the numeral form VIIII rather than the IX used in the earlier tile production of the legion at York; but they are not from the same die. A vexillation of legio XX was probably active at Scalesceugh at the same timeFootnote 19 – so withdrawing part of the Ninth from Britain would not have left the northern frontier unprotected. Haalebos points out that if only part of the legion was active near Carlisle in the period 105–20, the rest being at Nijmegen, the apparent absence of a full-size fortress in the Carlisle area would be explained.Footnote 20 The presence of detachments there and at Nijmegen would help explain the drop in the samian stamp record at York, if that is not a figment of the reduction in the supply of samian in the Trajanic period. It is entirely plausible that with renewed Hadrianic interest in Britain and the move of XXX Ulpia to Nijmegen the vexillatio Britannica and the detachment of the Ninth Legion returned to the island. This is in fact suggested by some who favour a later loss of the legion outside Britain.Footnote 21
One argument against the Ninth still being in Britain under Hadrian is the absence of evidence for its involvement in the building of the Wall on the Tyne-Solway isthmus ordered by that emperor. It has often been said that if the legion was still in, or had been returned to, Britain, its involvement in the building of the Wall might have been in the turf western sector, without stone building inscriptions. This is true, but there is also no reason to dismiss out of hand its participation in the earliest stages of stone building of the Wall, before a decision was taken to add full-sized forts to the line of the Wall. This early work on the stone Wall includes structures and Broad Wall foundation laid for a few miles east of the river Irthing. Here legio XX Valeria Victrix are the builders named on a Hadrianic inscriptionFootnote 22 from near (and probably from) Milecastle 47, an unusually large and wholly broad-gauge construction, probably very early in the building programme. No governor is named; it has been thought possible that this belongs in an interregnum between Pompeius Falco and Platorius Nepos, or the governorship of the former.Footnote 23 In the central sector work on several structures (milecastles and turrets) pre-dates the fort decision, for a turret (36B) is overlain by Housesteads fort and a milecasle (43) by the fort at Great Chesters. Inscriptions dedicated under Platorius Nepos at Milecastles 37, 38 and 42 associate II Augusta with the same distinctive milecastle type as at Milecastle 43. These dedications might actually refer to completion following the fort decision, but, given the correlation between the Second Legion dedications and what is a distinctive building plan, there is a strong possibility that II Augusta was the originator of this milecastle type. Work in Wall miles 7–27 was advanced before the fort decision: Halton Chesters fort (mile 21) was built over the works and Chesters fort overlies Turret 27A, which had been built and occupied. The differing styles used in apparent 5-mile building allocations throughout this stretch suggest that three different legions were involved, conventionally thought to be II, VI and XX. One of these ‘legionary signatures’ corresponds to the work finished by II Augusta in the central sector, but otherwise none of the centurial stones or other inscriptions from structures gives a firm link between the commencement of curtain or structures in Wall miles 7–27 and an individual legion. Thus, only XX Valeria Victrix and II Augusta can be connected epigraphically with the earliest work on the Wall.
If IX Hispana is unattested in the building of the first plan, so equally is VI Victrix, of which no inscription can be firmly attributed to the pre-fort decision phase of Wall building. The only possibility is an undated Sixth Legion building stone from Turret 33B, but there is no recognised type of early Hadrianic building inscription from turrets, and this stone is widely regarded as commemorating a later second-century repair.Footnote 24 Altars erected by VI Victrix to Oceanus and Neptune, found on the site of the Hadrianic bridge at Newcastle (presumably built or started as part of the first plan), have been variously connected with the arrival of the legion from across the sea and the building of the bridge.Footnote 25 These dedications do not show that the Sixth Legion was the original builder: the altars undoubtedly came from the uppermost superstructure and the legion might have completed an unfinished bridge whose building had been interrupted.Footnote 26 If the altars are really to do with the transfer of the Sixth to Britain, it is just as likely that the ceremony took place on a structure that was already there, or they could be contemporary with an inscription of c. 158 found in close associationFootnote 27 and record a mid-Antonine restoration of the bridge.
HOW EARLY COULD THE NINTH LEGION HAVE BEEN DESTROYED?
The latest datable evidence for the existence of the legion is set out in appendix 1. This shows that there is no proof of the existence of the legion after the earlier 120s, the latest date at which officers whose careers are recorded necessarily served in the legion. There is an apparent outlier, a tribune of the Ninth named Numisius Iunior, who has been identified with a man of the same name who was consul in 161. Normally the consul would have been senatorial legionary tribune 20 or so years earlier, so c. 140. However, as explained in appendix 1, if we accept the cogent argument of Keppie that Numisius Iunior, consul in 161, is not necessarily identical with, but might be the son or grandson of the Numisius Iunior who was tribune in the Ninth, there is no compelling reason to see the legion as surviving up to c. 140, and the tribune might have served under Trajan.Footnote 28 The legate Sextius Florentinus definitely served before 124 and possibly, with a gap between appointments, at the opening of Hadrian's reign. L. Aemilius Karus could conceivably have been tribune under Trajan. The career that provides firmest evidence for the extended life of the legion is that of L. Novius Crispinus Saturninus, consul in 149 or 150, who should not have served as tribune in the Ninth any earlier than the early to mid-120s (proposed reconstructions of his career that have placed his tribunate nearer 130 are discussed in appendix 1). If we set Numisius Iunior aside as an unproven case, the remaining careers rule out the traditional loss of the legion in 117–19 but are compatible with loss or disbanding at some point in the 120s.
As we shall see, an episode of serious warfare in Britain almost certainly occurred at this time, and it is at least as reasonable a possibility as any other that the Ninth was still in Britain then and was in some way compromised. That this is the context for the disappearance of the Ninth (or a defeat that led shortly to its disbanding) is as deserving of consideration as either of the other, eastern, scenarios.
To examine these briefly: the Ninth could have found its way to Judaea and have been destroyed in the second Jewish Revolt of 132–35, but there is absolutely no evidence that it participated in that war or any trace of its presence there.Footnote 29 There is the complication that another legion vanished from the record in these years, its disappearance traditionally explained by the Jewish War: XXII Deiotariana, last attested in Egypt in 119, although there is no clear evidence for its involvement. Transfer of a whole legion from western to eastern extremity of the empire in an emergency would be quite unusual, but not wholly unparalleled (I Minervia travelled from Bonn to participate in the Parthian campaigns of Lucius Verus). The available evidence has been taken to suggest that three legions fought the Jewish War in their entirety (the two already based in Judaea plus one from Syria) and that there is evidence for the participation of detachments from six others, all drawn from the eastern provinces or from the Middle and Lower Danube.Footnote 30
The other idea is that the Ninth was destroyed in 161, when the governor of Cappadocia, the ‘foolish Celt’ Severianus, advanced against the Parthians, who had invaded Armenia: ‘Vologeses had begun the war by hemming in on all sides the Roman legion under Severianus that was stationed at Elegeia … and then shooting down and destroying the whole force, leaders and all’.Footnote 31 The word in Xiphilinus’ eleventh-century epitome of Cassius Dio translated as ‘legion’ is stratopedon, undoubtedly used in places by Dio in the technical sense of ‘legion’, but elsewhere in the sense of a military camp, an encamped force or garrison, or an army of occupation.Footnote 32 The passage in fact occurs in material that Xiphilinus was forced to compile from unknown sources other than Dio because, as K. Juntunen has shown, the relevant books of Dio had been lost by early Byzantine times.Footnote 33 In any case, it is far from certain that the destruction of a specific single legion at Elegeia is meant.Footnote 34
TRACES OF KNOWN LEGIONS IN THE EAST: IS THE ABSENCE OF EVIDENCE FOR THE IX LEGION EVIDENCE OF ABSENCE?
Another line of enquiry is to ask whether we should expect to have recorded traces of the Ninth Legion in the east if it had existed somewhere there through the reign of Antoninus Pius until destruction in 161. Keppie has drawn attention to the problem of finding an eastern home for the NinthFootnote 35 and has commented that ‘British scholars have shown considerable enthusiasm to be rid of the legion, without adequate regard for the need to find it a posting in other provinces where the occupants of various fortresses were well known, or the provincial garrisons fully subscribed.’Footnote 36 Is the absence of positive evidence for the legion's presence in the east evidence of absence, especially when we consider, for example, that the exact site of the legionary base of XII Fulminata at Melitene in Cappodicia is unknown and there is not a single inscription of that legion from the place? We can assess this by seeing what evidence we would have for any of the post-Hadrianic eastern legions if, for the sake of argument, they had ceased to exist in 161. Appendix 3 shows that there is evidence for the existence of every eastern legion in the period 138–61, and in every case the fortress is known or approximately located from either literary (Melitene) or archaeological evidence. There is really no reason to doubt Keppie's contention that if the Ninth had been based in the east until 161 we should expect to find some evidence for it.
HADRIANIC WAR IN BRITAIN AND THE EXPEDITIO BRITANNICA
A war or wars, involving serious Roman losses, occurred in Britain during Hadrian's reign. For the scale of the losses, we have the well-known testimony of Fronto, who reminded Marcus Aurelius that ‘when your grandfather Hadrian was emperor, how many soldiers were killed by the Jews, how many by the Britons’.Footnote 37 An early second-century tombstone found at Vindolanda in 1997 commemorates a centurion apparently of cohors I Tungrorum, the unit based at Vindolanda before its move to (probably) the newly constructed Hadrianic fort of Housesteads on the Wall, in bello … interfectus, ‘killed in the war’.Footnote 38 But other than the statement in Historia Augusta (Hadrian 5) that the Britons could not be kept under control on Hadrian's accession (117), formerly used to explain the disappearance of the Ninth and itself quite possibly a literary cliche with no relation to real specific events, there is no extant literary attestation of a war in Britain under Hadrian.
However, we hear from two epigraphic sources of an expeditio Britannica during Hadrian's reign. This event occurred at least a few years later than the Historia Augusta troubles of 117–19, for this must be the case with the expeditio Britannica mentioned on the career inscription of a participant, Pontius Sabinus, a senior centurion who brought legionary reinforcements numbering 3,000 across to Britain.Footnote 39 This cannot have been the ‘war’ on Hadrian's accession (so 117–19) deduced from the Historia Augusta, because Sabinus’ career included centurionates in two legions between being decorated in Trajan's Parthian War (possibly as early as 115) and achieving the primipilate in a third and then being sent on the expeditio Britannica. A second participant was Maenius Agrippa, personally chosen by Hadrian and misso in expeditionem Britannicam; while in Britain he was tribune commanding cohors I Hispanorum, based at Maryport.Footnote 40
We know Hadrian was in Britain in 122, and consensus has strengthened in recent years that the particular terminology used on these inscriptions means that the expeditio took place in that year, as in general the adjectival form of the name of a people or country – as in Britannica – is only coupled with expeditio in cases where the emperor participated in person.Footnote 41 The expeditio was not merely an exercise in strengthening the island garrison faced with the huge task of building Hadrian's Wall: the 3,000 legionaries brought to Britain by Pontius Sabinus surely arrived at a moment of acute military crisis, when severe losses had taken place. In Hadrian's biography there is no mention of him conducting a war in Britain, but this means little in such a patchy and selective source as the Historia Augusta.
In the past, much later contexts, in the late 120s or even in the mid-130s, have been suggested for the expeditio Britannica. Apart from the need to link the term with Hadrian's presence in Britain, the careers of the two known participants make a much later date improbable. For Pontius Sabinus, there is a kind of fixed chronological point in his career. When decorated in Trajan's Parthian War he was in his second equestrian militia, when we would normally expect him to have been in his early 30s; he was probably born in the early 80s. Rather than going on to a third equestrian militia, Pontius Sabinus transferred to the legionary centurionate, serving in two legions before becoming primus pilus of III Augusta and leading the vexillations to Britain. He might just have been 40 by 122; the usual minimum age for a primus pilus was 50, but promotion could be accelerated in time of war, especially for a directly appointed equestrian who had been decorated by the emperor.
But if the expeditio was as late as the mid-130s, Sabinus would have served for nearly two decades in the centurionate before only then going (as he did) through the standard sequence of Rome tribunates, primus pilus II, and becoming procurator of Gallia Narbonensis at some time in his 60s. This would not have been the rapid rise the decorated equestrian hoped for, and, as Brian Dobson pointed out, this would hardly have been an advantageous transfer from an equestrian career, in which a procuratorship could have been reached more rapidly. For senior centurions hoping for such posts, ‘It was of decisive importance to achieve the Primipilate at the minimum possible age and to complete the sequence of Rome tribunates in as short a time as possible.’Footnote 42 For these reasons, Dobson doubted that Sabinus’ expeditio could have been as late as the 130s; while not making this an outright impossibility, the career suggests that a date before 130 should be sought.Footnote 43 Even a date in the later 120s has been seen as over-extending (to over 12 years) the likely length of Sabinus’ first two centurionates.Footnote 44
In his expeditio Britannica – reasonably assumed to be the same as Sabinus’ – Maenius Agrippa commanded cohors I Hispanorum, and we know from inscriptions that this was when the unit was stationed at the fort of Maryport. We do not know where in the sequence of attested Hadrianic commanders there Agrippa is to be placed, but it is usually assumed that the known fort at Maryport was built with the Wall forts, not before about 124 (if we follow conventional chronology for the building of Hadrian's Wall) or at earliest about 122 (if we follow Erik Graafstal's dating of the fort decision to Hadrian's visit – see below). There might be an undiscovered pre-Hadrianic fort at Maryport, but it is still difficult to accept suggestionsFootnote 45 that Agrippa came to Britain with Pompeius Falco in 118.Footnote 46
A man, L. Tusidius Campester, believed to be the son of Maenius Agrippa (who also has the names L. Tusidius Campester and is styled ‘father of a senator’ on his career inscription), is now known to have been consul in 142. He would have been born at least 40 years earlier, c. 102, suggesting that Agrippa must have been at least 40 at the earliest possible date (122) he could have come to Britain. If the identification of the son is correct, this practically rules out a date for the expeditio in the 130s, when Agrippa would have been over 50 – barely credible given his extended later career including the posts of praefectus alae, prefect of the British fleet and procurator of Britain. But at present this indication that Agrippa's expeditio must have been in the earlier part of Hadrian's reign does not amount to proof, for despite the rarity of the name, the identification of the consul of 142 as Agrippa's son is not certain.Footnote 47 The combined evidence of the careers of Sabinus and Agrippa therefore is suggestive of a date early in Hadrian's reign for the expeditio Britannica, but stops short of certainty.
It has been argued that the warfare of the expeditio Britannica might have occurred shortly after Hadrian's visit, in the period 123–26, something that David Breeze proposed in 2003, opting for 124.Footnote 48 He went on to suggest that such a war might explain perceived evidence for interrupted building work in the forts on Hadrian's Wall added by the recently implemented ‘fort decision’. In support of such a view it is said that the term expeditio does not invariably denote the personal presence of an emperor;Footnote 49 Sabinus could, as Jarrett demonstrated, have held his two centurionates between c. 115 and 122, but it has been argued that a slightly later date for his participation (124) is preferable.Footnote 50 Maenius Agrippa is stated to have been sent by the emperor to the British war, not to have accompanied him.Footnote 51 None of these is a decisive objection to the obvious 122 date. Alexandrian victory coin issues of 124/5 and 125/6 have been seen as referring to Britain on the basis of the belief that no wars are known anywhere else at this time, but the relevance of these to Britain has been strongly disputed.Footnote 52
A difficulty for a serious war in 123–25 lies in the steady flow of troop discharges from the British army throughout the period and into the 130s, attested by diplomas of July 122, September 124, September 125 (possibly), August 127, 130, December 132 and April 135. It might be thought that this makes a major war unlikely, especially between 122 and 127 – this on the assumption that discharges would not occur during periods of heavy fighting. For a war in the mid-120s, if we go along with the idea that discharges would not be made in time of war, then either between 122 and September 124 or between September 124 and September 125 (if that date is correct) we would have to compress all of the events connected to the expeditio. This is not impossible, but it is a very tight timetable. The great discharge made in the emperor's presence in July 122 could, on the other hand, have been made on the conclusion of a war; so the diploma evidence sits most comfortably with the idea that the expeditio Britannica be dated to 122.Footnote 53
It has also been observed that, according to the diploma lists, the British army was at the peak of its strength at the time of the great discharge recorded in July 122, with some units apparently being transferred out of the province in the following years, rather than the garrison of the island being bolstered by new arrivals.Footnote 54 The transfer of some units out of Britain could have followed the conclusion of a war in 122.
Although a war in 123–24 or so has its supporters and cannot be disproven, 122 provides the most straightforward, least problematic context for the conclusion of the conflict that culminated in the expeditio Britannica. Whether this episode is to be equated with the serious warfare attested by Fronto and the Vindolanda inscription remains an open question, but must be regarded as a strong possibility. If it is possible that the Ninth Legion was in Britain at this time, a serious defeat for this legion, precipitating the reinforcements brought in during the expeditio, might seem the best explanation for its disappearance from the record.
INTERRUPTION IN THE BUILDING OF HADRIAN'S WALL
The archaeological evidence for a dislocation and change of plan in the building of Hadrian's Wall is well known and widely discussed. It is necessary to revisit the phenomenon and its date here because it might indicate an episode of warfare and thus be connected to the end of the Ninth Legion. A first plan for milecastles, turrets and a 10-Roman-foot-wide stone wall east of the Irthing (the Wall was to be of turf to the west) was abandoned while incomplete. It is during this first phase of building that three distinct types of construction have been noted, each occurring in more than one stretch, some of which seem to be 5-mile-long building allocations. These have reasonably (though without proof) been seen as the work of three different legions. After the abandonment of this incomplete first phase, a second plan for major garrison forts superimposed on the mural line and an accompanying rearward earthwork, the Vallum, was decided on and implemented (the forts definitely built under Platorius Nepos, that is somewhere between 122 and the end of his governorship, on present evidence somewhere after September 124 and before August 127). When work resumed on the curtain wall, which had not got beyond foundation level in the central sector, it was to a narrower gauge and different specification.
This dislocation has been accounted for in two basic ways.
1. One school of thought believes that the Wall had been decided on in 119 and started c. 120 under the governor Pompeius Falco, and that on his visit in 122 Hadrian personally intervened and ordered the forts and Vallum.Footnote 55
2. The more generally adopted view has been that Hadrian's visit saw the commencement of the Wall in 122 and that the ‘fort decision’ and Vallum came later in 123 or 124, after Hadrian had left Britain, as a result of a reappraisal of military requirements, in particular the need to overcome the obstacle that the Wall had provided to rapid movement of the army, not allowed for in the first plan.Footnote 56
The view that the Wall was being built before Hadrian came to the province has been powerfully revived by Graafstal. It cannot be proven, but the fact that this is now known from dendrochronology to have happened with the German palisade (timbers felled winter 119–20; Hadrian's visit in 121) and Graafstal's eloquent case for the amount of survey and materials preparation that would have been necessary make it most probable that, at the very least, preparations were in full swing for the Wall in Britain under the governor Pompeius Falco in 120–21. That construction work actually began then must be seen as a serious possibility.Footnote 57
Graafstal emphasises Hadrian's visit and personal appraisal of the work in progress as the decisive factor in abandoning the first plan. Similarly, the traditional view of a later ‘fort decision’ regards it as an adjustment following an appraisal of the way the new Wall was working. Arguably, however, the nature of the dislocation that took place speaks less of measured appraisal and more of an emergency response to a crisis precipitated by unexpected events that lay entirely outside the work programme for the Wall.
When the dislocation occurred, the Broad Wall of the first plan had been raised above its foundations between the area of Milecastle 7 and some unknown point to the west, but perhaps much of the way to the Portgate, mile 22. In the 5 miles to the west of Milecastle 22, between Portgate and the North Tyne, Broad foundation had been laid throughout and sections of Broad Wall had been raised, but these were interspersed with gaps – obviously intended building allotments – where nothing had been built above the Broad foundation. These gaps were later plugged at an unknown date by a very narrow curtain wall only 6 Roman feet wide.Footnote 58 The implication is that the Broad Wall building programme had been suddenly abandoned when incomplete, the builders called away for some urgent purpose, usually supposed to be the call to build the forts of the second plan. This is not an original observation: it has long been recognised that these phenomena offer a ‘time slice’ showing where building had got to when a dislocation occurred.Footnote 59 But what was its cause?
If Hadrian, Nepos and their advisers had decided, upon inspecting the work, that forts and a Vallum were necessary, would such a dramatic dislocation in work have resulted? Would a ukase (as Graafstal terms it) have suddenly gone out for everybody to stop work on everything at once? Where would the builders go, and what would they do? For, following on from the fort and Vallum decision, sites would have to be chosen, survey work undertaken, architects’ designs produced, materials procured, quarries identified and opened. It could be argued that the evidence allows for work continuing on the Broad Wall until the planning of the forts was completed and it was time for their building to start. But in this case we might expect, during this phase of design and planning, the gaps in the curtain to have been filled and for there to have been a gradual and orderly transition of the workforce from the old project (first plan) to the new (forts and Vallum). The Broad Wall building programme was interrupted in a seemingly disorderly way.
In the central sector, Milecastles 37, 38, 42 and 43 were completed under Platorius Nepos, but some at least had been begun before the fort decision, as seen at Milecastle 43.Footnote 60 At Milecastle 37 a start was made at Broad gauge but the northern Wall tapered to Narrow gauge to either side of the Broad-gauge north gate that had been started first. Either Milecastle 37 was abandoned incomplete, and work resumed at a later date when the Narrow Wall plan was current, or the order to change to Narrow gauge came through in the early stages of its construction, and the work was seamlessly adapted to the new specification for the Wall curtain. If the former, we have another example of work suddenly abandoned and not readily explicable in terms of the decision to add forts that had not yet been planned in detail. If the change in construction at Milecastle 37 does represent a hiatus, it is possible that it corresponds to the dislocation in Wall miles 22–7. If we follow Peter Hill's reasoning that the Platorius Nepos building inscription from Milecastle 37 cannot have been put in place until after the milecastle was adapted to the Narrow gauge, then that inscription and those from 38 and 42 will belong to the post-hiatus resumption of work rather than to the work of the first plan.Footnote 61
There is also the evidence that Graafstal himself has adduced for an interval between first-plan building and second-plan fort construction, as at Turret 27A, which seems to have been occupied for a time before Chesters fort was built over it.Footnote 62 Does this not suggest that when the builders abandoned the Broad-Wall building programme immediately adjacent to this, there was a considerable interval before there was any fort building programme for them to move to?
The structural evidence can be read to suggest, therefore, that some serious emergency unconnected with the building programme or revisions that Hadrian or anyone else wished to make to it was what precipitated a sudden abandonment of work, and it must be a possibility that this was the revolt or war that led to the expeditio Britannica. In that case, the second plan, the fort and Vallum decision, could be seen as a military response to events, designed and executed after an abrupt cessation of work on the first plan and an interval during which fighting – and reconsideration – took place.
Certain aspects of the second plan are explained if seen as a reaction to war and rebellion: there is not only the addition of forts for auxiliary units all along the line, of a newly invented type, integrated with the Wall curtain and designed for the rapid deployment of military force,Footnote 63 but also the Vallum, a rearward protection which can be seen as a response to internal rebellion that had threatened the Wall from behind and as a measure to prevent collusion between internal rebels and enemies to the north. Even the decision to extend the curtain 4 miles to Wallsend, which on a straightforward view appears to be contemporary with the fort decision (for Wallsend seems to fit into the scheme of regularly spaced Wall forts) could be read as a response to some attempt, possibly successful, to bypass the eastern end of the Wall. It must also be a possibility that some or all of the Cumbrian coast fortlets and towers were an addition at this time, perhaps an overreaction to recent traumatic events, since they were not retained for very long.Footnote 64
THE POSSIBLE HISTORICAL CONTEXTS FOR THE LOSS OF THE NINTH LEGION IN BRITAIN
Having established that there are no compelling grounds for believing that the Ninth Legion was permanently transferred from Britain or was in existence after c. 125, and having surveyed both the evidence for Hadrianic war in Britain and the structural mark this may have left on the building programme of Hadrian's Wall, the conclusion must be that it is possible that the expeditio Britannica, the dislocation and change of plan in the building programme of Hadrian's Wall and the substitution of the Sixth Legion for the Ninth were associated events. Indeed, this is the most economical (though of course not the only possible) interpretation of the evidence, and there is a good case for dating this hypothetical combination of events to 122.
That in turn necessitates accepting Graafstal's contention that the building of Hadrian's Wall had begun before 122 and that the fort decision took place following the events of that year. The evidence we have reviewed for interruption in the building of the Wall could be explained if the first plan was commenced under Pompeius Falco and interrupted by war brought to a successful conclusion in 122; the great troop discharge of July 122 could have followed on from this. The addition of the Wall forts would then follow under Nepos. This model gives a neat explanation for the wholesale transfer of the Sixth Legion to Britain: to replace the Ninth, irreparably damaged or lost in the events leading up to the expeditio. That the Sixth simply replaced the lost Ninth was the natural view when it was believed that the latter was destroyed in 117–19.
If a pre-122 date for the beginning of the Wall is entertained, the possibility must also be considered that, if we are correct in seeing the ‘signatures’ of three legions in the first Broad-Wall phase of Hadrian's Wall, one of those was the Ninth. Working on the assumption that the Ninth had been transferred from Britain under Trajan, Graafstal suggests that part or all of the Sixth was brought to Britain before the traditional (but not attested) transfer date of 122. As his Wall-building timetable starts before 122, when in his view the fort decision took place, the Sixth Legion would have to have been in Britain by 121 at latest in order to be one of the three legions whose structural signatures have been detected in the first-plan Broad-Wall construction in Wall miles 7–27.Footnote 65 This is to leave a question unasked and unanswered: if an additional legion was needed in Britain c. 120, either in preparation for the building of the Wall or to make up numbers following losses in war, why was the Ninth (generally assumed to have left for Nijmegen at this stage) not returned to Britain? Is it not more probable that the Ninth Legion was reunited in Britain early in Hadrian's reign and that the Ninth was one of three legions engaged in the first plan of Wall building? As we have seen, there are no inscriptions of the Ninth from this phase, but equally there is none of the Sixth, whose earliest dated inscription is post-fort decision, recording the building of Halton Chesters under Nepos.Footnote 66
One possible answer to the question of why the Ninth was not reunited in Britain when needed might be that that part of the Ninth Legion in Britain was more or less wiped out sometime in the period 115–20 – that is, at the traditional, pre-Ritterling date – leaving a vexillation at Nijmegen in existence for a time before a decision was finally taken to dissolve it, perhaps as late as the mid-120s.Footnote 67 The Sixth Legion could conceivably have been moved to Britain as early as 120 to make up for the losses. However, this reconstruction of events divorces the loss of the Ninth from the expeditio Britannica, not to be dated any earlier than 122, and the legionary reinforcements brought in no earlier than that point and certainly following heavy losses – in other words, forcing us to postulate two distinct wars or episodes involving heavy casualties. But if we do take the most economic approach and date the disaster affecting the legion to c. 122, associating it with the expeditio Britannica, it remains possible that a detachment from the legion still lay at Nijmegen at that time and lingered on after the destruction of the larger part in Britain.
One objection to associating the destruction of the Ninth with an expeditio Britannica of 122 might be the career of L. Novius Crispinus Martialis Saturninus, which has been taken to suggest his service as tribune in the legion no earlier than the mid-120s (see appendix 1 for full discussion). This is not an insuperable obstacle: rather than praetor at the latest possible date (135), from which the date of his tribunate, approximately ten years earlier, is derived, he might have held that post several years earlier, allowing him to have been in the legion in 122; it is by no means an impossibility that he served in the Ninth so early. This would not have been a model career, with a period of unemployment after the praetorship and the fasces not held until his mid-40s, but some careers were simply interrupted or retarded.Footnote 68 It also has to be said that, although we know of several cases where it happened, we understand little about the process by which legions were disbanded after defeat. Perhaps rather than being annihilated, the Ninth suffered heavy losses but continued to exist for a short period; dissolution may only have been decided upon after an interval. Conceivably, Crispinus might have been sent out to the legion at a time when it was still thought possible to build up its numbers again.Footnote 69
Crispinus’ career might seem more consistent with the attempts to place the war in 123 or in 124, but this is not necessitated, any more than it is by the career of Pontius Sabinus. Difficulties for this proposed date have already been discussed (that the term Britannica as an adjective attached to expeditio should mean Hadrian was in Britain at the time and the flow of troop discharges following 122). Nevertheless this later date cannot be disproven, and it is worth noting that there seems no way of disproving that the Sixth Legion was transferred to Britain (replacing the Ninth?) as late as 124, rather than at or before the conventional 122 date.Footnote 70 Those who have argued for the later war see a dislocation effect on Hadrian's Wall in the building histories of its forts; here, as argued above, it is seen as much more likely that if any dislocation signals war, it is the major interruption between the first plan and the revised wall-with-forts plan. If that were the case, there seems insufficient time for the Wall to be started in 122, interrupted, war (possibly involving the loss of the Ninth) to be fought and the Wall (in its second-phase plan) resumed, all within the governorship of Nepos. But a war in 123–24 could just conceivably have interrupted work on a first plan that had begun some time before 122.Footnote 71
Arguments for a later date for the expeditio Britannica, around 130, have been made. Sheppard Frere argued that Maenius Agrippa came second in the sequence of known commanders at Maryport and that his appointment indicates the expeditio Britannica occurred in 128 or (as Frere preferred) 129. He could adduce no known reason for an expeditio Britannica at his preferred date except for the presence as governor from 130 of a formidable general, Julius Severus.Footnote 72 We have seen that, besides the adjectival form Britannica pointing to 122, there are objections raised by the careers of Maenius Agrippa and Pontius Sabinus to their expeditio Britannica being conducted any later than the mid-120s, especially if the former's son has been identified correctly as the consul of 142 with the same name.
None of this absolutely disproves that the Ninth Legion was still in Britain so late and was destroyed in some unrecorded war of the later 120s or 130s that was separate and distinct from the expeditio Britannica. A much later date than 122–24 would remove any objection based on the career of the tribune Novius Crispinus Saturninus, but it would spoil the neat substitution of the Sixth Legion for the Ninth, which is naturally explained by a war in the earlier 120s. It would mean, as some writers have proposed, that the Ninth was returned to, or reunited in, the island early in Hadrian's reign and then supplemented by the Sixth, certainly present under Platorius Nepos, in preparation for Wall building, Britain becoming once again a four-legion province.Footnote 73 However, while there may not be a statistical basis for using the absence of inscriptions to say that the Ninth was no longer in Britain after c. 125,Footnote 74 the absence of any fortress or any kind of archaeological site we can connect with the legion for the remainder of the Hadrianic period is surely telling.
CONCLUSION
Ever since the old story of the Ninth's destruction around 117–19 was challenged in 1925, the scene of the legion's final fate has drifted further from Britain. Ritterling, who first voiced doubts about its loss in Britain at the opening of Hadrian's reign, suggested that there was a second British rising in the middle or the second half of the 120s in which the legion was lost. E. Birley in 1948 kept an open mind on whether the legion was transferred from Britain before its loss and whether the expeditio Britannica might be placed at some later time in Hadrian's reign. The emergence of evidence for the legion's presence in Nijmegen seems to have tipped the scale, and by 1971 Birley was less equivocal, emphasising the lack of any evidence for loss in Britain. We have seen that this material in no way disproves that the Ninth Legion was in Britain in the early 120s, but there seems to have followed an eagerness to write the Ninth Legion out of Hadrianic Britain. Since the 1970s, and the suggested dating of the tribunate of Numisius Iunior in the Ninth to c. 140, the possibility that the Ninth was in Britain when it ceased to exist has disappeared so thoroughly from the educated public consciousness that many believe its departure from the island to be a matter of fact.
The problem with the removal of the legion from Britain before its disappearance is not that this is necessarily wrong or impossible, but rather lies in the sweeping away of the possibility of finding an explanation in unrecorded events in Britain, which on any reasonable grounds must be regarded as equally, perhaps more, probable, with all its implications for the hostile climate in which Hadrian's Wall was built and the purpose it was intended to serve. The assumption that the Ninth Legion had no role in the first phase of building on the stone part of Hadrian's Wall should be reconsidered. The prosopographic evidence does not yet prove that the Ninth was still in existence after the mid-120s. The case for the Ninth being at Nijmegen under Hadrian no longer seems strong following the re-evaluation by Haalebos, and Keppie has reminded us that ‘no inscriptions recording the building activities of the legion or the lives and careers of its members have come from the east’.Footnote 75 That the legion was dissolved after losses in Britain is arguably a less problematic reconstruction. Among various possibilities, the hypothesis making most economical use of the evidence and most consistent with the subsequent epigraphic silence is to associate the end of the Ninth with the attested expeditio Britannica, the gap between the two plans for Hadrian's Wall and the transfer of VI Victrix to Britain – whether these events are placed in 122 or a short time after.
ACKNOWLEDGEMENTS
I am extremely grateful to Paul Bidwell, Anthony Birley, Lawrence Keppie and Kai Juntunen, as well as to Britannia's two anonymous referees, for reading versions of this paper and suggesting corrections and improvements. None of these necessarily agrees with the conclusions reached here and all errors are the responsibility of the author.
APPENDIX 1: THE EPIGRAPHIC EVIDENCE FOR THE LATEST DATABLE MEMBERS OF THE NINTH LEGION
(A) NOTED BY RITTERLING AS UNLIKELY TO HAVE SERVED AS EARLY AS 117–19
L. Aemilius Karus: tribunus laticlavius of the Ninth Legion (CIL 5.1333 = ILS 1077). Recently published diplomas show that Karus was consul in March 144.Footnote 76 In a model career he would have served his tribunate some 20 years earlier, so 124 or so. A.R. Birley concludes: ‘He [Karus] may have left IX Hispana in 122’.Footnote 77 But note that a pre-119 date is possible: ‘his service with the Ninth might well have come before the death of Trajan … as witness the case of Claudius Maximus, suffect consul in 144 who while tribunus laticlavius of … IIII Scythica had been decorated by Trajan, evidently in his Parthian War’ (i.e. 27 years between tribunate and consulship).Footnote 78
L. Novius Crispinus Martialis Saturninus: tribunus laticlavius of the Ninth Legion (CIL 8.2747 = ILS 1070). Consul in 149 or 150. Crispinus’ career inscription has been taken to show that the Ninth Legion still existed as late as c. 130. Barbara Levick suggests the following chronology for his posts:Footnote 79
If the praetorship was held at the usual age of 29, this would place Crispinus’ tribunate in the Ninth around 130. Duncan Campbell proposes a similar chronology, when he writes that ‘Scholars have generally assumed that Crispinus was consul in AD 150. He would then have been born around AD 110, so that his legionary tribunate could not reasonably be placed any earlier than AD 128.’Footnote 80
However, it has been argued that Crispinus cannot have held the juridical role in Spain immediately after a praetorship in c. 139, as others are known in this office between 138 and 142,Footnote 81 and it seems impossible that he was iuridicus in 142–44 and that his following three posts (legionary legate, proconsul and governor of Africa) were all compressed into the period 144–49.
This difficulty has compelled an alternative reconstruction of his career in which Crispinus holds the Spanish office in the last years of Hadrian, 136–38, which places his praetorship in 135 at the latest and suggests service in the Ninth around ten years earlier, in the mid-120s. On this basis, A.R. Birley concludes that ‘His tribunate of IX Hispana may have been in the mid-120s.’Footnote 82 Crispinus certainly served in the Ninth well into Hadrian's reign, and his career is the one that (setting aside Numisius Iunior, below) extends the life of the legion the furthest, but the tribunate is not necessarily to be dated any later than c. 125 and could be somewhat earlier if his praetorship, not dated, was actually earlier than 135. If Crispinus served in the Ninth aged 18–20 in 122–24 (praetor in 133), he would have been 45 on being designated consul in 149 – far from impossible. See further discussion in main text.
(B) SUGGESTED BY ERIC BIRLEY TO HAVE SERVED IN THE 130s
M. Cocceius M. f. Pol. Severus: primus pilus of the Ninth Legion (CIL 5.7159). The name Cocceius suggested to E. Birley enlistment at earliest under Nerva in 96–98 (the idea being that the recruit was granted Roman citizenship on acceptance for legionary service, taking the emperor's praenomen and nomen); this man became primus pilus of legio IX before going to be praefectus of X Gemina. Birley cites Juvenal to the effect that the traditional age for a primus pilus was 60, and, reckoning on 40 years of service leading up to this, placed his primipilate in the Ninth around 136.Footnote 83 Dobson, however, showed that the centurion came from a part of northern Italy where there were several long-established and distinguished families of Cocceii and concludes that the occurrence here of Nerva's praenomen and nomen was probably entirely coincidental. On other grounds, the inscription is dated no earlier than c. 100; Severus could therefore have been in the Ninth Legion in pre-Hadrianic times.Footnote 84
(C) SUGGESTED BY ECK TO HAVE SERVED IN THE 130s
Numisius Iunior: tribunus militum of the Ninth Legion (CIL 11.5670). A diploma published in 1972 and discussed by EckFootnote 85 shows that a man of this name was consul in 161. If this man is correctly identified with the tribune of the Ninth, this would show that the legion was still in existence in the 130s: ‘A consul of 161 with children and a favoured start cannot have been born later than c. 120, so the tribunate was probably held c. 138–40.’Footnote 86 The identification has been widely accepted.Footnote 87 However, Keppie has suggested that the consul of 161 might be the son or grandson of our legionary legate, who seems to have had a son called Iunior.Footnote 88 In that case, the military tribune Numisius Iunior might have served in pre-Hadrianic times. The possibility of this is accepted by A.R. Birley, who maintains, however, that ‘on present evidence it seems reasonable to identify the consul with the legate’.Footnote 89 Although the identification has been vigorously defended,Footnote 90 at the very least it must be concluded that there is an element of uncertainty in this single piece of evidence for the existence of the Ninth Legion after the 120s.
(D) AN EARLY HADRIANIC LEGATE OF THE NINTH LEGION
Sextius Florentinus held posts in the following order:Footnote 91
Legate of legio IX
Proconsul Narbonensis
Governor of Arabia c. 126–29
Therefore, Florentinus commanded the legion either immediately before 124, if there were no gaps between these appointments and his proconsulship was in 124–25, or, if he had a break or breaks between posts, which is possible,Footnote 92 some years earlier. A.R. Birley has judged: ‘He is unlikely to have left the legion much before 124’ and ‘His command of IX Hispana can be dated to Hadrian's first few years.’Footnote 93
(E) AN EASTERN RECRUIT OF THE NINTH LEGION GRANTED CITIZENSHIP BY HADRIAN?
CIL 10.1769 is the epitaph of Aelius Asclepiades, who was from Cilicia (south-east Turkey) and buried at Naples at the age of 42 after serving eight years in the Ninth Legion: D(is) M(anibus) // Aelius Asc/lepiades nati(one) / [C]il(ix) mil(es) leg(ionis) IX / vix(it) ann(os) XXXXII / mil(itavit) ann(os) VIII / Aelia Seleria / b(ene) m(erenti) f(ecit). It has been suggested by A.R. Birley that he was a sailor in the Misenum fleet (although this is not mentioned on the inscription) who was drafted into the Ninth Legion in an emergency and returned to his home after death; other men from the Misenum fleet are known to have been transferred to legio X Fretensis under Hadrian, probably in connection with the Jewish Revolt of 132–35. It has been suggested that the Ninth was in the east and similarly reinforced at this time.Footnote 94
If, as has often been supposed, the name Aelius indicates enlistment under Hadrian, this would indicate the Ninth Legion was still in existence in c. 125/6 (117/8 being the earliest possible date for transfer under Hadrian, plus eight years of service in the legion), and possibly much later. However, Keppie suggests that the name might have no chronological significance: ‘If he had recently been given citizenship by Hadrian, the form P. Aelius Asclepiades might have been expected.’Footnote 95 Nor, if it is possible that he was transferred into the legion from the Misenum fleet, does Asclepiades’ origin demonstrate that the Ninth was ever based in the eastern provinces. All the stone indicates is that a man from the Misenum fleet might have been transferred to the Ninth Legion at a time when it needed reinforcement.
APPENDIX 2: EVIDENCE FOR THE NINTH LEGION ON THE CONTINENT
PUBLISHED AND DISCUSSED BY BOGAERS IN 1967
1. Two tile fragments stamped LEG VIIII HISP found in the legionary fortress at Nijmegen in 1959 and 1966. The 1959 find is said to come from the floor level or overlying destruction layer of the latest, stone phase of the fortress, in the third officer's house north of the east gate.Footnote 96 The 1966 find was illustrated in 1977 and said to come from a robber trench of the principia.Footnote 97
2. A mortarium rim stamped G VIIII HIS found at the legionary tilery of Groesbeek-de Holdeurn, near Nijmegen, in 1938.Footnote 98
The fortress at Nijmegen had been largely rebuilt in stone when the legion based there since c. 71, X Gemina, was transferred to the Danube in c. 103/4. Military use of the site, on a much-reduced scale to judge from the level of finds, continued well into the second century, with several units attested on tile stamps: legions VI Victrix, VIIII Hispana, XXII Primigenia, XXX Ulpia Victrix, the vexillatio Britannica and the exercitus Germanicus Inferior. Of the vexillatio Britannica stamps, there are some 120 examples (compare the total of two from the Ninth Legion). The vexillatio Britannica has commonly been thought to have consisted of detachments from British auxiliary units and perhaps from the British legions, or maybe just the Ninth. Its stamps, like those of the Ninth Legion, numerous stamps of legio XXX Ulpia and the others listed above, cannot be associated with distinct stratigraphic phases and are simply associated with the latest levels or demolition contexts of the final (‘Period 5’) stone phase of the fortress.Footnote 99 There is no inherent means of dating the Ninth Legion stamps or establishing their relation to those of the other attested units. A suggestion has been made, on the basis of seeing which unit stamps are associated with each other at particular find-spots, that the vexillatio Britannica stamps predate the departure of X Gemina in 103/4,Footnote 100 but this ignores the principle of residuality – X Gemina tiles will occur with unrelated finds in contexts much later than their time of manufacture – and has been severely criticised by Haalebos, who points out that the fortress canabae, intensively occupied before c. 103/4, produced hundreds of tile stamps of legio X Gemina, but none of the vexillatio Britannica, whose presence is therefore best placed in the subsequent period between c. 105 and c. 120 when the canabae was severely contracted.Footnote 101
Bogaers’ explanation for the traces of the Ninth at Nijmegen is that detachments from Britain – the vexillatio Britannica – occupied the fortress after the departure of legio X Gemina c. 103/4 and that the Ninth Legion succeeded them in the Hadrianic period, the vexillation troops returning to Britain and the Ninth being moved out of the island c. 121 to recover from damage suffered in war there. In this view, the Ninth occupied Nijmegen until the 130s, then being transferred to the east.Footnote 102 Although this judgement that the Nijmegen evidence for the Ninth is best placed after c. 121 was long accepted with little question, it is clear, as Haalebos has shown, that there is no stratigraphic or dating evidence to prevent the Ninth Legion tiles being associated with those of the vexillatio Britannica (of which a vexillation of the Ninth might have formed a part) and belonging to the period 103/4–120.Footnote 103
OTHER EVIDENCE
3. A horse-harness pendant with punched inscription LEG HISP IX, found at Ewijk in the area west of Nijmegen: Haalebos mentions a report that it came from a villa site, suggesting that veterans of the Ninth were settled on lands outside the Nijmegen fortress.Footnote 104
4. AE 1968, 323: an undated altar to Apollo dedicated by a man who had been primus pilus of the Ninth Legion and was promoted to be its praefectus castrorum; found at the healing spa and sanctuary of Apollo Grannus at Aachen, 120 km south of Nijmegen: L(ucius) Latinius L(uci) f(ilius) Publilia / Macer Ver(ona) p(rimus) p(ilus) leg(ionis) VIIII Hisp(anae) / praef(ectus) castr(orum) pro se et suis / Apollini / v(otum) s(olvit) l(ibens) m(erito).
5. CIL 13.4030 = AE 1973, 36: a dedication from Mersch, 45 km west of Trier, to an unnamed equestrian who served as a tribune in the Ninth Legion as part of his tres militiae, following which he became a priest of the imperial cult at Trier, in the early second century: ] / flam[en] Augu[sti] / flamen Leni M[art]is quinq[uennalis] / praef(ectus) cohort(is) [II His]panor(um) eq[uitatae] / tribunus militu[m le]g(ionis) VIIII Hi[spanae] / praefect(us) e[quitum alae Augustae] / Vocon[tiorum. It has been argued that this unnamed Treveran evidently served his tres militiae on the Rhine – including the military tribunate of legio IX.Footnote 105 If this is correct, a date before 119 or so for the tribunate is implied, as the tribune went on to command ala Augusta Vocontiorum, a unit of the British army (Vocontian cavalrymen are attested on a Vindolanda tablet under Domitian)Footnote 106 which was apparently in Lower Germany as part of the vexillatio Britannia,Footnote 107 but which had returned to Britain by 122, appearing on the diploma of that year.
6. Tatiana Ivleva has documented the occurrence of 22 late first-century brooches of British origin in the Nijemegen area and discussed their possible connection with the Ninth Legion and other British units stationed there. By the nature of the evidence, no close dating is possible and it is impossible to make any specific connection with the Ninth legion as distinct from the other units of the vexillatio Britannica.Footnote 108
APPENDIX 3: EVIDENCE FOR LEGIONS AND THEIR BASES IN THE EAST c. 120 TO c. 160
In each case, any inscriptions naming the legion from a given fortress are listed by Bishop Reference Bishop2012.
XV Apollinaris/Satala: fortress site known, with (not closely datable) inscriptions of legion. Existence 120–60 attested by careers: AE 1983, 517; CIL 14.4237. Known also from literary evidence (Arrian).
XII Fulminata/Melitene: fortress not closely located, but clearly linked to Melitene by historical sources, notably Dio, and to Cappadocia by Arrian. A couple of career inscriptions show that the legion was in existence shortly before 160: CIL 02.01180; AE 1971, 00367.
XVI Flavia Firma/Samosata: fortress not closely located, but there are inscriptions from the site (not closely datable to the period 138–61). Existence of legion epigraphically attested under Antoninus Pius: AE 1903, 00252.
IV Scythica/Zeugma: fortress not closely located, but indicated by numerous (undated) tile stamps of IV Scythica and inscriptions from the area. Existence epigraphically attested under Antoninus Pius: AE 1937, 00101.
III Gallica/Raphanaea: fortress located. At least two career inscriptions show the legion in existence under Pius: CIL 10.408, 12.2230.
If a similar exercise is carried out for Judaea and Arabia, it produces an even richer epigraphic record to show the existence of the legions based there at this period: X Fretensis at Jerusalem, VI Ferrata at Caparcotna and III Cyrenaica at Bostra.
