sign in sign up
<<
Home , Antonine Wall, Auchendavy, Balmuildy, Caracalla, Petriana, Portgate

The Development of the Mural Frontier in Britain from to Caracalla* by David J. Breeze and Brian Dobson

In 1936 in a short paper in the Journal of Roman Studies,1 Sir lan Richmond effectively demonstrated tha Antonine th t frontiery t 'ato e no t actuallWals bu ' wa l y embodied structural advances ove s Hadrianiit r c predecessor somd an , e year sRomans latehi n i r Britain gave h 2 e briea t concisbu f e accoun improvemente th f o t s introduce Antonine th y db e planners inte oth later developmene Th . resula s ta experiencf to murae th f eo l barrier from Hadria Carao nt - calla has, however, never been examined in detail: the purpose of this paper is to study this development. Hadrian's Wall, from the start of building in the early 120s to its abandonment when the Antonine earl e Walbuils th ywa n l ti , underwent several modifications. When first planned the units closesfrontiee th o t Romaf ro n Britain appea havo rt e been accommodate seriea n di s of forts base Agricola'n do s road from Carlisl Corbridgeo et Stanegatee th , lina ,f com o e - munication through one of the few convenient east-west gaps.3 The forts were at half the normal interval of one day's march, allowing the rapid assembly of a force to meet an attack on whatever scale or to go on the offensive, striking up from or . Beyond the road to the west Kirkbride - and possibly Drumburgh4-may have housed units to control the localpopulation guard an Solwaye d th ease newle th th t o t ;y discovered for t Whickhaa t havy mma e formed part of a system of protection of the south bank of the Tyne, unbridged below Corbridge. Between fortcertaith e n th so f no e from Corbridg Carlislo et e fortlets have been discovered- Burn and Throp as yet are the only two certainly known - and the system, such as it wascompletes wa , numbea y db lookouf o r t posts, most placed forwar line fortf th ed o f dso an fortlets. The fortlets may have housed units which were more concerned with control of movement, tha patrollins i t -mannind an g g duties functioe th , n later assume Wale d th an ly db it watchtowerd san s (). This syste mf deploymeno f unitso t , however satisfactor r dealinyfo g with major attacks, could not deal with the problem of infiltration by raiding parties or cope with the whole business of contro movementf o l . Hadrian decide construco dt onle th t y for f effectivmo e controe th n i l absence of a natural barrier, an artificial barrier which could only be crossed at controlled points. This gave securit provinc e possibilite th th o yt d ean economif yo c developmene th t o right p u t frontier. This barrie originalls a r y planne consiso t s stona f dwa o t e wal Roma0 1 l n feet wide and perhaps 15 high to a parapet walk for the eastern 45 miles, and a turf wall 20 Roman feet wide at base and probably 14 high5 to parapet walk for the western 31 miles. In front of the Wall

* This article is based on the paper read by the first author to the Society of Antiquaries of at their one-day conference, 'The Two ', 1st May 1971. I PROCEEDING 0 11 SOCIETYE TH F SO , 1969-70

1 :T~«—""tr- • i f u» • » J —L • • u • TJ-TT

LJ• • LJ • • LJ • • U *^ r LJr • LJ

L- ______'::::;' J------_-_ 1^3——I———————————' '.-INjji . .——"~~~T ~t:-;=.;-.:.-:;|.-J-~ ~ Li'lJ^~ ---/ - ——- -- UJ^- - —T — __._--

CD—————CD—————HC3——-——CJ d

"FI 1 GHadrian' Antonine s th Wal d an l e Wal diagrammatin li c form . Hadrian'a . s Wal originalls la y plannet dbu not completed. Main features: wall and with turrets, pierced by milecaste gateways; pre-existing forts behind the line . Hadrian'b . s Wall original pla modifies na d under Hadrian. Main features: walditcd an l h with turrets, pierce milecastly db e gateways; forts astrid walle eth , behind with crossing t fortsa s only . Antonin.c e Wall. •Main features: with ditch; forts, with annexes, attache rampartdto ; fortlet beacon-platformsand s where -necessary; Military Way . Hadrian'd . s Wall from early third century. Main features: wall with ditch; some turrets; pierced by reduced gateways; forts; Military Way

there was to be a ditch wherever necessary. This ditch varied in width from 26 to 34 feet, with emphasie th lowee th abous n r o sfigure wa fee9 td tan , deelittla r po e more t regulaA . r mile intervals along the Wall gates were to be placed, each guarded by a fortlet or milecastle BREEZ DOBSOND EAN MURAL E TH : FRONTIE BRITAIN RI N FROM HADRIA CARACALLO NT | A11 1 containing one or two small barrack-blocks, and each apparently provided with a lookout ove norte th r h gate.6 Between each milecastle therturreto tw e r lookoue werso b o et t , regularly spaced. Along the exposed west flank of the Wall, the Cumberland coast, milecastles d turretan s (for convenience distinguishe s milefortletda d towersan s ) wer ee built b als o ot , though no continuous wall was provided, the sea being considered a sufficient barrier. The line of frontier control was thus to extend from Newcastle-upon-Tyne, Hadrian's new bridge across the Tyne, for possibly nearly 120 miles to St Bee's Head. There is also some evidence that the south bank of the Tyne was defended, and certainly a new fort was built at , at the mouth of the river, though this may be a later addition. The troops to garrison the mile- 7 turretd an s s were probabl providee b auxiliare o yth t y db y units statione line f th eo n di forts already in existence on the Stanegate a mile or two behind the Wall; there is no evidence for a separate force. In the event of an impending invasion from north of the Wall, these units would be expected to move up through the milecastle gateways to fight the enemy in the field. purpose Wale Th th f l e Romao e itself th s a , n biographe Hadriaf o r divido t s e ne th wa tell , sus barbarians from the Romans. The military units had no special part to play in the control of movement unles militarsa y threat8 developed; they remaine thein do r natural lin communicaf eo - tion. The fortlets were presumably abandoned as the milecastles were occupied and the lookout towers were either incorporate systew d ne int r abandonedme o oth . In the course of putting this scheme into operation, when it was something over half- completed importann a , t chang frontiee mads th ewa o et r complex. decides wa t abando9I o d t n the forts immediately south of the Wall and replace them by new forts actually on the line of Walle th . This decisio clearls nt wa undertake yno n lightly t involvei r fo , t onl abandone dno yth - ment of permanent forts and the erection of new ones but also the demolition of milecastles, lengthd san Walf so l already constructe infilline th d ditche partf gdo an th f so . Thesw ene forts, placed about seven miles apart astridy la , Wale eth l whereve locae rth l topography allowed, with three twin-portal gates providin equivalene gth milecastlx si f o t e gates Walle nortth f h.o The purpose behind this change seems to have been to allow unrestricted access for major aree forceth a o Wale t norts th f l ho whic h would hitherto have been difficult with milecastle gate negotiateo st extremels i t I . y unlikely tha milecastle th t e gate kine s werth use dw r deno fo of sorties that have sometimes been envisaged,10 with the enemy rolled up against the Wall;

this type of warfare was not in keeping with normal Roman practice. The new forts were only 1

part of the Wall complex for convenience; the conversion of a mobile1 field army into a frontier police force had not yet begun, though the temptation to make this conversion now existed. All the forts would seem to have been planned to project north of the Wall wherever possible. However, it was soon realised that, with the forts on the line of the Wall itself, so many gates opening north of the Wall were unnecessary. Hence at Halton Chesters both portals of the west gate were apparently blocked before the erection of the gateway was completed, and at Housesteads, with only one double gate opening north of the Wall, the east portal was built up at a similar stage.12 The fort of Greatchesters, one of the last of the primary series of forts completede b o t buils wa ,t wholly sout , thoughof h still attache Wale th , l dto althoug t coulhi d perfectly well have projected. The same is also true of , added at a late stage in building to plug the long gap between Chesters and Housesteads. At Birdoswald too modifica- tions, probably lat Hadrian'n ei s reign, reduce numbee dth f portalo r s Wale nortth f lh o fro m six to two. same th t e A tim thess ea e forts were constructed r perhapo , littlsa e later, thre fortw ene s were also built on the Cumberland Coast which hitherto seems to have been guarded only by pre-existine th g Maryport. Further modifications resulte strengthenina n di eastere th f go n flank 112 | PROCEEDINGS OF THE SOCIETY, 1969-70 by the extension of the Wall curtain four miles down-river to Wallsend, where a fort was built, same th t e a t earliertime d no f an i , , anothe moutre rivee sites forth th t Sout t a rf wa dta h o h Shields. Finally three fort t Bewcastlea s , Netherb Birrend yan s shielde westere dth n approache Wal e alsd th an olo st probably tha Brigantee tribe tth parth f eo f o t s left nortf ho the Wall. Hadrianie Th c garrison Wale th f l so fort s were apparently predominantly infantre th s ya following table demonstrates:

SUGGESTED ORIGINAL GARRISON HADRIAN'E TH F SO S WALL FORTS13 South Shields a/a Sabiniana? Wallsend cohors quingenaria14 equitata? quingenaria?a a/ Benwell Rudchester cohors quingenaria equitata? Halton Chesters cohors quingenaria equitata? Chesters cohors milliaria equitata?? Carrawburgh* cohors quingenaria equitata? Housesteads cohors milliaria peditata Great Chesters cohors VI Nerviorum quingenaria peditata?15 Carvoran* cohors I Hamiorum quingenaria peditata16 Birdoswald cohors I Tungrorum milliaria peditata?" Castlesteads cohors IV Gallorum quingenaria equitata?1* ?a a/ 19 Stanwix Burgh-by-sands cohors quingenaria equitata milliariar o peditata? Drumburgh? * Bowness-on-Solway cohors milliaria equitata? Beckfoot cohors quingenaria peditata Maryport cohors I Hispanorum equitata Burrow Walls cohors quingenaria peditata 20 Moresby cohors II Lingonum quingenaria equitata?21 *Carrawburg Carvorad han late nar e Hadrianic addition seriee th fortsf o so st . The date of Drumburgh is uncertain.

Alon whole gth e frontielengte th f ho r complex there were only three solely cavalry units; at South Shields in an isolated position at the mouth of the Tyne, probably at Benwell, though the identification of the original garrison is uncertain, and also presumably at Stanwix though there is here no direct evidence for the Hadrianic garrison. Six of the remaining sixteen forts contained wholly infantry units, two of these being on the Cumberland coast, while the remaining forts were occupie mixey db d unit cavalrf so infantryd yan . Thireflecy proportioe sma th t f no mixed unit infantro st y arme unitth f Britainyn o si formee th , r being much more common,22 but it was also undoubtedly connected with the local situation.; A mixed unit could provide the infantry to man the milecastles and turrets and undertake local patrolling, and also the cavalry carro t t longeyou r range patrollin scoutingd gan furthermors i t I . e noteworthye thatth f o , units statione Wale th ln do itself , those partl whollr yo f cavalryo y wer groupel eal d aroune dth point majoso whertw re nortroade th th ho st pass throug Walle th hth t a , e just west of Halton Chesters and at Stanwix. If necessary the cavalry of these units could converge on these roads to provide equal forces of something over 1,000 cavalry to move rapidly north. same th t e A tim r shortleo y afte importane th r t decisio movo nt forte eth s ont Wale oth l was taken; it was also decided to add the earthwork known as the Vallum behind the Wall and the forts. The Vallum, consisting of a ditch with a mound set back on each side, offered no advantag eitheo et purposs rit sidt undoubtedls ebu ewa shielo yWalt e reae th df lth ro complex . earthwore Th k could onl crossee yb forta t da , wher causewaea y guarde gatprovideda s y edwa b , wherr o majoea r road passed throug Wall.e hth produco t effec s 2s 3It wa t emilitara y area between BREEZ DOBSOND E AN MURA E TH : L FRONTIE BRITAIN RI N FROM HADRIA CARACALLO NT | A11 3 the Wall and the Vallum, and to reduce the number of places where the Wall could be crossed fro originae m th r thereabout o 8 7 l mera , providino st e14 ggreatea r measur f controeo l than previousld ha y been possible. This migh takee b t suggeso nt t tha buildine th Walte d th f ha lg o met with local opposition, though there is no other evidence to support this supposition. The advanc forte th sf eo ont Wale constructioe oth th l lind Vallue ean th f no m behin Wale dth l resulte traffil al dn ci being channelled throug hsmala l numbe crossing-pointsf ro , thereby reducing usefulnese th milecastle th f o s e gateways presence Th . completf eo eWale unitth ln so mus t also have reduced the importance of the lookout towers, both the turrets and the milecastle towers. The Wall abandoned at the beginning of Pius' reign was very different from that envisaged in the early 120s. That Wall had undergone many modifications, major and minor, during the cours constructionf eo Wale Piuf Th .planne s o lexperience lighe sth wa th f o tn d i Hadrian'f eo s reign and, in contrast to its predecessor, it seems to have been built with the very minimum of modifications. mose Th t striking difference between Hadrian' Antonine s th Wal d an l e Wal thas li t while curtaine th continuoue th , s barrier formee th mainls f o wa ,r f stoneyo lattee , s th tha f wa ro t whollturw fne f walturfonls e y o fee 4 wa Th ly1 . t widture th compares feef 0 ea f 2 o t e th o dt sector of Hadrian's Wall, but, as Richmond long ago pointed out, this may not have implied a reductio heighn i unlikr tfo predecessos eit erectes wa t ri d upo nstona e base. This would allow for a greater steepness in the back of the rampart and a saving in material. Far from being a slighter barrier than Hadrian's Turf Wall, e Roma'ipossibls i tth s a n thino t e t i army f ko - engineer's most up-to-date pattern in turf-work designed to achieve the same effect as Hadrian's Turf Wall t savin materia e bu ,thir e th gf on d o l while becoming much less easily accessible from

behind'. however, is curtait e I Antonin e th cleat y th f ,no n o r wh e Wal buils turln wa t i f apparently 4

sooo s n afte rebuildine th r2 ture parf th gf o f sectoo t Hadrian'f o r s Wal stonen i l t seemI . s that rebuilding on Hadrian's Wall, which included the rebuilding of the turf curtain and turf and timber milecastles in stone, the construction of new turrets and the digging of a new ditch where one stretc Wale re-aligneds th wa f l ho started ha , t Rivee milecastlda th n r o Irthin9 e4 d gan proceeded westward abouo st t milecastl , before 54 decisio e eth advanco nt e into Scotlans dwa taken. It has recently been suggested that the move into Scotland brought this work to a halt. Eve provisioe nth turstona w f n ne o Wal ee basth l o ecannot t have provide same dth e degree 25 of permanenc stona s ea e wall. If the Wall curtain itself is of less permanent material than Hadrian's Wall, the ditch to the north is more substantial. In contrast to the average dimensions of the Hadrianic ditch, 27 feet wide and 9 feet deep, the Antonine ditch has an average width of 40 feet and an average depth of 12. The berm, the space between the Wall and the ditch, on the was usually abou fee0 2 t stonte wideth esamn e o Walth , s ea Hadrianf o l t mucbu , h wider thae nth six feet on his turf Wall. However, it is doubtful if this extra width was, as has been considered, corallinn sallien i i d d orden i sai an eneme go th rt y withi ditche nth , which have w es a ,seen , were contrar Romao yt n military practice. More likely perhap structuraa s si l purpose connected wit greatee hth r widt ditch—the deptd th h an f ho e collapsture th f n sectof turreo e o a f o r54 t Hadrian's Wall int ditche oth , probabl Hadrian'n yi s reign, demonstrated that her t leasea e th t six foot berm was too narrow.26 In the spacing of forts on the line of the Wall the Antonine organisation differed radically from the Hadrianic, but was at the same time a logical growth from the earlier. Once the forts had been built on the line of Hadrian's Wall, many milecastles and turrets must have become superfluous plannine th Antonin e n I .th f go e Wall recognisedthis swa milecastle .Th turred ean t syste largels mwa y abandone forte th sd werdan e placed much closer together, being only about | PROCEEDING 4 11 THF SO E SOCIETY, 1969-70 two miles apart. Moreover none of the Antonine Wall forts are known to project north of the Wall, but are placed on the south side of the curtain. This is to be expected in view of the measures taken on Hadrian's Wall to reduce the number of portals opening north of the Wall by blocking portals, re-aligning curtaie th r constructinno g forts wholly Walle soutproximite th f Th .ho y of the forts to each other was not considered by itself sufficient for between certain forts small enclosures or interval-fortlets, a little larger than Hadrianic milecastles, have been found. At present three are known, at , Bridge and at , where one purpose of the fortlet was clearly to guard the gate on the road through the Wall, otherd awaiy an sma t discovery though, is t whethe I .y earlo sa ,to o yt r ther sucs intervaln ehwa a - fortlet between every fort or whether they were only provided in special circumstances. No lookout towers hav t beeeye n discovere Walle th line f th ,eo thougn do h beacon-platforme ar s known. These occur in pairs and three pairs have been found, one pair on either side of Rough Castle and one just west of Croy Hill. These are a feature completely unknown on Hadrian's Wall and their purpose appears to have been connected with signalling. The new forts are usually considered to differ from their predecessors in their materials of construction forte Hadrian'n Th .o s s Stone Wall Birdoswald Ture an , th easterfe f o th t d da nen Wall, all appear to have been constructed in stone. The fort-walls and the principal buildings were undoubtedl othee f stoneth yo t r buildingsbu , , barracks, store-house likee th , d weran s e probably of timber on stone sill-walls. Indeed at South Shields timber barracks of a Hadrianic date havbeew eno n discovered forte Hadrian'n Th .o s s Turf Walhavy ma el been originally built in turf and timber apart from Birdoswald, east of the Red Rock fault. With the exception of two forts, and which have stone walls, all the forts on the Antonine Wall follow the same pattern: turf walls, stone principal buildings and timber barracks and store-houses. Allowin diiferene th r gfo t materials employeWallso tw curtaie e e th th , th n di f no construction of the two series of forts is not remarkably different. The turf walls of the Antonine keepinn i fort e ar s g wit ture hth f rampartimbere th d an tbarrack t slightlbu e ysar inferioo t r their stone Hadrianic predecessors; timber buildings of even later date have been recently un- covered in certain of the Pennine forts.27 The building of the forts of the two Walls, and indeed of the whole of the frontier complexes, was in the main carried out, or at least directed, by the same units, the legions. All the available evidence points to the Hadrianic forts being built by with som detachmen a ease gived y th eai b t n ni Britise th f o th fleet. Howevers i t i , not impossible that some work was done by auxiliaries under the direction of the legionaries just as they helped in the digging of the Vallum. On the Antonine Wall, part or all of .Croy Hill, , and Balmuildy are known to have been constructed by legionaries,28 the headquarters buildin possibld gan y mor Rougf eo h Castl buils auxiliaries,y ewa b t 29 while both legionaries and auxiliaries seem to have been at work, at Bar Hill and possibly Castlecary. Certainl auxiliariee yth s see havmo t e playe greateda buildinre parth n i t thif go s Wall than they 30 . buildine th predecessors n i it f d go di . surprisins i t I g that ,othe e wheth l rn al know n fort turd sfha ramparts, .two forts, Balmuildy and Castlecary, should have been surrounded by stone walls. It is usually considered that this was becausfort o importand tw sha e eth t function performo t s , Balmuildy guardin crossine gth g of the , while both forts may have been at the ends of roads leading to the Wall dow Clyde nth e valley. t coulI arguee db d connectes thai t i t do wit tw positioe he th th f no 31 buildine fortth n i s g sequenc Walle th f e.o Balmuild constructes ywa d befor ramparte eth , with stone wing-walls projecting fro norte mth h corner anticipation si .arriva e ramparte th th f nf o o l , and interestingly the only, building inscriptions on the Antonine Wall mentioning Lollius Urbicus ;have been found here othee th 3 2r- building inscriptions recor governor'o dn s name situatioe .Th n BREEZE AND DOBSON: THE MURAL FRONTIER IN BRITAIN FROM HADRIAN TO CARACALLA 115 at Castlecary is not so clear, but this fort too probably antedates the rampart. However, other fort alse ar so know havo nt e been constructed befor rampare eth t hav thesd no ean to estoned walls: , , and where work had started and was apparently in progress when the rampart builders arrived.33 These four forts, with the exception of Duntocher fortlet but probably with the addition of Carriden and Bar Hill, seem to belong to a series of large, fairly evenly spaced forts constructed, or at least started, before the arrival of the rampart- builder n theii s r vicinity d possiblan , y even earlier—other forts wher e relationshith e p with the rampart is known, Rough Castle, , Croy Hill and Cadder, are structurally later tha ramparte nth , though probably only slightly later stonee Th . wall f Balmuildso Castled yan - cary would therefore appear not to be related to their early date in the Antonine Wall building- programme, unles f courss originallo s wa t ei y planne givo dt l fort al e s stone walls, theso etw being constructed before this plan was altered.34 The reasons for the peculiarities of the two forts must as yet remain unanswered. instructivs i t I comparo et Wallso garrisonforte e tw e Hadrian'th f e fort.Th so e th f th so f so s Wall were all built to receive whole units, but those on the Antonine Wall are very different as the following table demonstrates.

SUGGESTED GARRISON ANTONINE TH F SO E WALL FORTS acreage35 Antonine I Antonine II Carriden large Kinneil small? small

Mumrills I Thracum7-I h 3 co / Tungrorum/ a ega/

7 3

Falkirk 36 Rough Castle 1-5 h VIpar co Nerviorum f o t 3* Seabegs small Castlecary 3-9 coh I Tungrorum pedmil 391 coh I Vardullorum eq4mil "/ legion f VInd o 4an 1 I sI Westerwood 2-6 —vexillatio legiof no n VI?42— Cro2 y Hill —vexillatio f legiono n VI??43— r HilBa l 3-6 coh I Saetasiorum44 coh I Hamiorum45 Auchendavy c 3-3 —vexillation f legioo n II46— Kirkintilloc? h Cadder 3-3 coh quingenaria peditata? Balmuildy 4-3 coh quingenaria peditata r parunia o f o t Kilpatricw Ne k 2-6 Castlehill c 3-4? —part of coh IV Gallorum eof— Duntocher 0-66 — unitparta f o — Old Kilpatrick 4-7 coh milliaria peditata? I Baetasiorum coh 4B (Bishopton 5-2 quingenariacoh equitata?) r froFa m being buil wholr tfo e unit Antonine sth e Wall forts, varyin sizn gi e from tiny Duntocher to 7-3 acres Mumrills, were clearly designed for forces widely differing in strength. Mumrills probabls wa yalae I Tungrorum,builth I r fo t Kilpatricd Ol cohorsa r kfo milliaria peditata,r Ba Hill, Cadde possibld ran y Balmuild r cohortesyfo quingenariae peditatae Bishoptod an n probably for a cohors quingenaria equitata. Rough Castle and presumably one or both of the forts on either side were garrisoned under by cohors VI Nerviorum, and parts of the two milliary cohorts, cohors I Tungrorum milliaria peditata and cohors I Vardullorum milliaria equitata, attested at Castlecary must have been housed in one or both of the neighbouring forts, Seabegs Westerwoodd an similaA . r situation apparently existe t Castlehillda t larg, whicno es enoughwa h to hold the whole of cohors IV Gallorum equitata, assuming that unit to have been in garrison at one time, and at Duntocher, its western neighbour, which interestingly, it appears, may not

have contained a headquarters building. Finally at one time detachments seemingly

9 4 11 | 6PROCEEDING S OF THE SOCIETY, 1969-70 garrisoned four forts in the central stretch, Castlecary, Westerwood, Croy Hill and Auchendavy. One reason for the diverse strength of the fort garrisons, the dividing of the auxiliary units and the use of legionary vexillations, was undoubtedly the overstretching of the army of Britain reoccupatioe th du o et Scottise th f no h Lowlands. However reasoe , parth alsy f o ntma have realisatioa o t beee ndu n thainflexible th t e approac forto hfort d an st garrison Hadrian'n so s Wall was unnecessary. As Dr A S Robertson has pointed out, the divergencies between the forts 'may rather reflect an intelligent, flexible, economical response to local conditions and exigencies'.50 This would help to account for the variety in the defences, the turf rampart ranging widtn i h from 12-J-13 fee t Mumrilla t fee0 t Roug2 a t o st h numbee Castleth d f ditchesan o ,r , fouro t e , on varying fro forsida e anotheo f mt eton o r dependin nature approache th th n f geo o , the position of the bath-house to which we shall return, and the placing of the internal buildings. t CroA y Hill exampler fo , granara , situates yi praetentura, e th n d i t Cadde a d an substantiaa r l area of the praetentura was apparently left free of buildings. This greater flexibility can also be seen in the relationship of the internal buildings to the forts' axes. On Hadrian's Wall the central range of buildings, wherever known, lies along the short axis of the fort facing the porta praetoria in the centre of one of the short sides. Most forts faced north with their long axes north-south but some, Housesteads, Greatchesters, Bowness-on-Solway and probably Stanwix, had their long axes running east-west. In this case the central range still lay along the short axis and the fort face Antoninde th east n O . e Wall, however, witexceptioe hon n (Cadder which faced east), knowe th l nal forts faced approximatel ymatteo n nort d rhan whethe lone th r g axinorthy sla - south, as for example at Rough Castle, Bar Hill and Balmuildy, or east-west, as at Mumrills and Castlecary, the central range was aligned east-west so that the fort faced north. In this way forte th s closely relate locae th lo d t topograph stild yan l faced north, tha towards ti enemye sth . greae Th t variatio numbe e fort e parn th i siz th e als s o se f i t t th eunit o f o rdu o n i s employed lace standardisationf th buildink e o o t th Wale d n th i f an lg o forte Hadrian'n Th .o s s Wall itself, tha excludins i t outpose gth Cumberlane tth fortd an s d producfortse th l appea al e , b f o to t r two legions and sizdimensione d t leas barraca th ,e an n th i tfort e d th ksan f sblockso e th d ,an 51 depth of the defences (usually two ditches), a certain degree of standardisation seems to have been involved: for example, 580 feet appearing m the dimensions of four forts, Benwell, Chesters, Birdoswal Stanwixd Scottise dan th n t possiblO .h no Wals i determino t et i l attempy ean t a t standardisation. Thi probabls si t jusresule yno th t f threo t e legion numbea d san f auxiliar o r y units being involve thein di r constructio t alsnbu o becaus forte eth s wer estageso builtw n i .t Hence Balmuild Cadderd yan , probably constructe same th r e dfo siz t leasunitf a eo r to , witha similar complement of buildings, by the same legion, // Augusta, but at different stages in the building programme sizen i , , dimension barracd san k blocks, bea resemblanco rn eaco et h other. In one way the garrisons of the forts of the two Walls are similar and that is in the scarcity of cavalry units Antonin.e Onl th for e n yo ton e Wal definitels li y know havo nt e house dcavalra y regiment at any stage, namely Mumrills, which appears to have been built for the ala II Tungrorum. This unit is placed three miles from the road running north and passing through the Wall at Watling Lodge fortlet. Only three cohorts wit hcomplemena cavalrf to attestee Wallye ar th n :do two quingenary units, II Thracum at Mumrills and IV Gallorum at Castlehill, and one milliary, I Vardullorum at Castlecary. It would appear that by far the majority of soldiers were infantry- men, a much higher proportion than on Hadrian's Wall. The reason for this is uncertain but it might be taken to imply that the main purpose of the Antonine Wall garrison was local patrolling. It certainly does not necessarily imply a more defensive attitude than was current twenty years before evenWalattace e n issua th e th f ln tn th o k I . eo would stil decidee lb field e th ,dn i preferabl y north rather tha othene Walle th sout th n r f hanO .ho Antonine dth e fort thein i s r positions, BREEZE AND DOBSONI THE MURAL FRONTIER IN BRITAIN FROM HADRIAN TO CARACALLA | 117 relationship ramparte th o st garrisond an , s hav mucea h closer connection wit Wale hth l than their Hadrianic predecessors. The forts of Hadrian's Wall were only incidentally part of the Wall complex, added to the line of the Wall as an afterthought, but the Antonine forts were planned from the beginning to be an integral part of the Wall. There was possibly an increasing tendency to use the army units for what were in effect frontier police duties. The army of the Wall was not spread evenly along the whole length but the main weight of the forces appears to have lain towards the western half. Along the eastern part of the Wall the only known large forts were Mumrills, Castlecary and presumably Carriden, but in the west eighe th sif t xo know n forts were ove facre th 3-t 3o tha t acree e th t sizen du si e b . Thiy sma wester Walne halth over-lookes f i l o f Kilsyte th Kilpatricd y db h an k Hills while fro ease mth t viee th wmucs i h more open. Moreove forte f th rStrathmor o s e would have screene ease dth t weste th t . Walno e t th en f bu l do The garrisons of the interval-fortlets were probably supplied from the forts on the Wall, wered an sometimes notha s a , s been supposed, separate troops whic previousld hha y serven di the milecastles of Hadrian's Wall. It seems probable that the milecastle garrisons were provided first by the forts on the Stanegate and later by those on the Wall and were not a distinct force. Most, perhaps all, of the forts on the northern Wall were supplied with annexes, often as strongly defende forte th ss da themselves oftes i t I .n supposed that these annexee s werth r efo camp-followers, the civilians who quickly congregated outside Roman forts. Excavation, admit- tedl t veryno y extensive t Mumrillsa , , Rough Castle, Castlecary Hillr Ba , , Balmuildy, Duntocher and Old Kilpatrick has failed to produce positive evidence of civilians in the annexes though the excavators of the annexe at , just north of Watling Lodge fortlet, did suggest that timber buildings there were used by civilians.52 In his recent book, The Frontier People of , Dr Peter Salway discussed the annexes of the Antonine Wall and came to the conclusion that they were purel r militaryfo y use, sometimes containing r examplefo , unit'e th , s bath-house. 53 No annexes were provided on Hadrian's Wall54 for the area around the forts were protected by Vallume th , which provided protectio continuoua r nfo s elongated annexe (incidentally another pointer to the contemporaneity of the forts and the Vallum). No Vallum was constructed on the later Wall, but the forts were provided with an extra defended area. Civilians were not allowed to settle in the area between the Wall and the Vallum, the Hadrianic annexe, and were therefore presumably not allowed into the Antonine annexes. The annexes at Rough Castle and Balmuildy containe unit'e dth s bath-hous t othea t rebu forts examplr fo , e Castlecar Hillr bathe Ba th , d -yan house was actually situated within the fort; Cadder seems to have had two bath-houses, one within and one without the fort. No bath-house for the men is known to have been placed within Hadriania f thoso c ford e an knowtWale th ln no one , Benwell Vallue souts th wa f ,d ho m an one, Chesters, betwee Vallum.e th Wal e d nth an l 55 The Antonine Wall forts were linked by a final element in the frontier complex, the Military Way roaa , d about 18 feet wide running paralle rampare th fee 0 o usuall5 d lt t o soutan tt p yu h of it. No such road was provided on the Hadrianic frontier but, after the Antonine experiment, it was clearly considered a necessity for one was constructed on Hadrian's Wall, probably when re-occupies wa t i d abou vere modificationw th y fe f t o 158 e .on This s swa know havo nt e been made to Hadrian's Wall at this time: another was the abandonment of the milefortlets and towers Cumberlane onth knowonle e th y on s havi o n t d 5 eCoasF beeM nt- re-occupie alsd f oo an d- fore Burrof th o t w Walls completioe th thir ;a s rebuildine dwa th f Turne o th ff g Walo stonen li . For the rest, Hadrian's Wall simply appears to have been reconditioned as it was when abandoned e earlth yn i 140s e fortTh .s were re-occupie auxiliary db y units, cohors I Hamiorum actually returning to the fort it had left twenty years before, if indeed it had ever left,56 the milecastles 8 11 I PROCEEDING SOCEITYE TH F SO , 1969-70 d turretan s were brought back into working order e curtaith , n repairee Valluth - d mre an d constituted wit ditcs hit h cleaned out. When Hadrian's Wall was re-occupied for the last time in the early third century, the lessons Antonine learnth n o t e Wall see mhavo t e bee cause nth f substantiaeo l modifications to the frontier complex, though these are as yet only imperfectly understood. Perhaps the most importan partiae th s l wa tabandonmen milecastle th f o t turred ean t system centrae th n I . l sector turrets were given up wholesale - no turret between MC35 and MC42 is known to survive into the third century - while in the five miles west of the and in the eastern sector the abandonment of turrets, if not quite as drastic, was certainly extensive.57 This modification should probabl associatee yb d wit greatee hth r control impose aree th a e n Walle nortdo th th n f I .ho third century the outpost forts became the bases for long-range patrol units and exploratores. Bewcastle, Netherby, Risingha Higd man h Rochester wer l garrisoneeal cohortesy db milliariae equitatae, thousand-strong mixed infantr cavalrd yan y units whicf o , h ther onle ear y five attested in Britai fifte th h n- bein t Castlesteadga thire th dn i scentur y- whil e exploratores wero to e apparently stationed at the latter three forts.58 Changes were also made to the garrisons of the Wall forts.59 Many fortcontainew sno d unit differena f so t size, often larger, than thos whicr efo h they were built, indeed in the third century three units, a and two numeri, are attested t Housesteadsa . Wher earle eth y third century garriso demonstratee b n nca t containeddi , with e exceptioth f Birdoswalno probabld dan y Rudchester elemenn a , f cavalryo t particulan I . r four alae werstationew Wale no th t Benwella ln do , Halton Chesters, Chester presumabld san y Stanwix as opposed to two in the Hadrianic period. This implies a greater emphasis on long- distance patrollin Wale th ly g unitb s themselves correspondina d ,an g reductio importance th n i e owatch-towea f r systeWale th ln m o line . However numbet e th ,no Wale s f th fort o rwa ln o s increase t thida s tim brino et g them into line wit developmene hth t from Domitia Antoninuo nt s Pius whereb e distancyth e between frontier forts decreasewa s d from abou mile5 1 t s through 7 to 2 miles; instead the garrisons of the existing forts were strengthened. Other changes intro- duced at this time were the narrowing of certain milecastle gateways where wheeled traffic was unnecessary apparentld 6an 0 abandonmene yth Vallue th f o m t - civilian s allowewerw eno o dt erect their house shoparee d th san a n si betwee WallVallue e nth th d .man These changes appreciatioreflecw ne a t reae th l f probleno northere th f mo n frontiere Th . ways in which a solution to the problems of the frontier were sought and the Walls were alter- natively occupied demonstrated that neither Wall was totally successful. The reason appears to have been that Hadrian's Wall was too far south and so out of touch with the centres of trouble, Antonine th nortr e fa Wal tha n ho occupatioi e to lScottisth e t th Wal e d th an f hl n o Lowland s overstretche resourcee d th arme th Britai f yn o so mucho nansweto e frontiea Th .s rwa r roughly Cheviotse linth e f th eo n o . This solutio apparentls nwa y attempted durin secone gth d occupation of Hadrian' ssecone Walth n i ld century when addition i , three th eo n t wester n outpost forts, Risingha d probablman y also High Rocheste d Newsteaan r ease th t n dwerei also occupied. In the early third century Caracalla adopted and improved this policy, occupying Netherby and Bewcastl weste th n ,ei Risingha Higd man h Rocheste easte th ,n i rwit h possibl advancen ya d base near Jedburgh and probably establishing at the same time loca or meeting places even further afield.61 The forts, the bases of long-distance patrol units, effectively advanced the frontier to Cheviote th ensured san d reasonably stable conditions well int fourte oth h century. Only after troublee th 360e Hadrian'd th sdi f so s Wall itself becom frone eth t line. Postscript. Excavatio 197n i 3 demonstrated Kilpatricthaw tNe k fort covered 2-6 acred san was probably garrisone parn d i cavalry y b t . BREEZE AND DOBSON: THE MURAL FRONTIER IN BRITAIN FROM HADRIAN TO CARACALLA | 119

NOTES

1. Richmond 1936, 190-2. . Richmon2 d 1963. ,50 generaa r 3Fo . l discussio Stanegate th f no e of Birley 1961, 132-50date th Birlef e c r .Fo y 1967, 9-11, and Hartley 1966, 14-15. 4. cf Bellhouse 1969, 77-9 for discussion of these two forts. 5. op cit, 89. evidence 6Th . sucr efo htowea r seem basee b o st d mainly upo T 0 situatioWe 5 nth wherC M t na e there wer pose5 t hole eithen so nortre sidth f heo gate passag soute opposes eth a t ha gate3 do t : Simpson Richmond an d 1935, 220-9. There doe logicasa seee b mlo t cas thir efo s equivalentowee th s ra t of a turret. Mr P R Hill has reminded us of the difficulty of constructing a tower over Type n mile- castle gateways buit by legion xx. 7. Birley 1961,157-9. 8. SHA Hadrian xi, 1. discussioa r buildine Fo . th Wal e 9 f n th o Hoolef lf c g o Breezd yan e 1968, 97-114; Breez Dobsod ean n 1972, 197-9. 10. cf Richmond 1966, 26. 11. Vegetius, de re militari HI, 21. 12. Simpso Richmond nan d 1937, 157-8; JRS, (1931)1 2 , 218. 13. cf Breeze and Dobson 1969,15-32. 14. EE m 202 and EE iv 706. 15. RIB 1731. 16. RIB 1778. 17. EE DC 1279. 18. RIB 1979 and 1980. 19. The fort is named after this ala milliaria placed there by the Notitia Dignitatum and it may have been there under Hadrian. 20. RIB 823. 798B RI , . 80021 . mixe226 3 . d unit attestee sar Britaidn i n compare infantr9 1 do t y units. 23. There appear to be no original causeways across the Vallum ditch provided at milecastles; access to the milecastles woul acrose db Vallue sth m ditc theford a t hnan ta probably alon norte gth h berm and through the north mound at a gap opposite the milecastle. cf Swinbank 1966, 86-8. 24. Richmond 1936,191. 25. Gillam and Mann 1970, 15-16. 26. Simpso Richmond nan d 1934,138-44. 27. Hartley 1966, 36-7; , i (1970), 283 alsf ;c o Halton Chesters: JRS, (1961)1 5 , 164. 2161B 28RI . , 2162, 2163, 2180, 2191, 2192. 29. RIB 2145. After Gillam and Mann 1970, 21-4; contra Steer 1964, 26-7. 2155B RI ,. 215630 , 2170, 2171. 31. Robertson 1970, 25-7. 32. RIB 2191, 2192. 33. Macdonald and Curie 1929, 214-19, and Macdonald 1934,195-7. case e . Istonfe th 34 s thisth , ewa wing-wall t Balmuildsa imply yma y tha originae th t l intentioo t s nwa construc ture th tf rampar stonen ti . acreage Th . bases ei 35 d upo dimensione nth fore th t f measureso d over rampartse th . 2140B 36RI . . 37. RIB 2142. 38. RIB 2160, 2161. 39. RIB 2155 records building by this unit under Pius. It may or may not imply that the unit was in garrison. 40. RIB 2149, 2146B RI ,. 214841 , 2151. 42. JRS, 54 (1964), 178. 12 | 0PROCEEDING S OF THE SOCIETY, 1969-70 43. RIB 2160. 2169B RI ,. 217044 Gillaf .C Mand man n 1970. ,24 45. RIB 2167, 2172. 46. RIB 2174, 2175, 2176, 2177, 2179, 2181. 2195B RI .. 47 48. Britannia, (1970)I Thi. , 31020 s. newl,no y foun annuae d th alta f o lr e dedication appearon e b o st s and thus records a unit in garrison. S N Miller, who excavated the fort, considered that it contained 10 barrack Antoninn si e I, being cohorsbuila r tfo milliaria peditata, tha d barrace an tth k accom- modation was reduced in Antonine II probably to hold a cohors quingenaria peditata: Miller 1928, 15-17. Cohors IBaetasiorum, recorded on the altar, will therefore probably have been the Antonine garrisoI I n which agrees wit evidence hth e fro Hilr mBa l (pace Steer 1964, 26-7) appearance .Th e on the stone of a prefect, Publicius Maternus, and a legionary centurion supervising its erection, lulius Candidus recen e probabls ,i th o t t e arriva formee ydu th f lo r afte temporare th r y command of the latter. lulius Candidus was in legion Iltalica. This may not be evidence for the presence of the legio promotioe th Britainn r i soldiee fo t th f nbu ,o r fro mBritisa h unicenturionata o t n ei . Beforit lefe e h appointment w Britaine s takhi no t p alta e erecteu s tth rwa d incidentally recording his new post. 49. Robertson 1957, 16-33. . Robertso50 n 1970. ,27 51. Breez Dobsod ean n 1969, 31-2. . Discovery52 Excavationd an Scotlandn i 1962 (CBA Scottish Regional Group). 45 , 53. Salway 1965, 156-8. 54. Annexes are, however, know numbea t na seconfirsd e fortf th an rto f o sd centuries exampler ,fo , Fendoch, Ardoch, Lyne, Glenlochar, Dalswinton, Milton, Newstead, Slack, Bainbridge, Gelligaer, Llystenn Pe , Caerhun, Bryn-y-Gefeiliau. bath-house 55Th . e withi fore t Haltonth a t n Chester s thirsi d centur r lateryo ; tha t Carrawburga t h is Hadriani t overliecbu obliteratee sth d south moun Vallume th f do . 56. Jarrett and Mann 1970, 191. 57. 17a, 18b, 19a, 25b, 26a, 29b, 30a, 31b, 33b, 35a, 36a, 37a, 37b, 39a, 39b, 40a, 41a, 50ao , 51bd a ,54 appeat no havo rt e been re-occupie thire th dn d i century , 12a;7b , 13a, 29a, 44b, 48a , 50b,49 , 52a, 53a seem to have been re-occupied in the third century. There may have been two phases in the abandonmen turretse th f o t , firstl blocking-ue th y e doorwayth f po secondlyd san , whee nth turret began the collapse, its demolition and the re-building of the curtain across the turret recess; cf Maxfiel Miked dan t 1973, 157-9. 58. Steer 1961, 97- Richmond 9an d 1969, 22-6. 59. Birley 1939, 211-24. 60. At MCs 36, 27, 28, 29, 40, 48, 49, 50, 52 and 54 one or both gates were narrowed. The north gate of MC22 by Halton was completely blocked. . Richmon61 d 1940. 97 ,

REFERENCES

Bellhouse L 196 R , 9 'Roma nCumberlane Siteth n so d Coast, 1966-67', Trans Cumberland Westmor- land Antig Arch Soc2, 69 (1969), 54-101. Birley E ,193 9 "The Beaumont Inscription Notitie th , a Dignitatum Garrisoe th d an f Hadrian', no s Wall', Trans Cumberland Westmorland Antig Arch Soc2, 39 (1939), 190-226. Birley, E 1961 Research on Hadrian's Wall. Birley, E 1967 'Hadrian's Wall and its Neighbourhood', Studien zu den Militdrgrenzen Roms, 6-14. Dobsond Breezean J D ,,196B 9 'Fort type Hadrian'n so s Wall', Archaeol Aeliana (1969)7 4,4 , 15-32. Breeze, D J and Dobson, B 1972 'Hadrian's Wall: Some Problems', Britannia, 3 (1972), 182-208. Gillam, J P and Mann, J C 1970 'The Northern British Frontier from Antoninus Pius to Caracalla', Archaeol Aeliana4, 48 (1970), 1^4. Hartley R 196 B , 6 "The Roman for t Ilkleya t . Excavation f 1962'so , Proc Leeds Phil2 . t Soc,Pt Li 2 1 (1966), 23-72. BREEZE AND DOBSON: THE MURAL FRONTIE BRITAIN RI N FROM HADRIA CARACALLO NT 1 12 | A Hartley R 196 B , 6 'Some problem Romae th f so n Military Occupatio Norte th f f 'hno o , Northern History, (1966)1 , 7-20. Hooley, J and Breeze, D J 1968 'The Building of Hadrian's Wall: A Reconsideration', Archaeol Aeliana*, 46 (1968), 97-114. Mannd Jarrett an C 197 G J , 0M , 'Britain from Agricol 'o at , Banner Jahrbuch, (1970)0 17 , 178-210. Macdonald, G 1934 The Roman Wall in Scotland. Macdonald, G and Curie, A O 1929 'The Roman Fort at Mumrills, near ', Proc Soc Ant Scot, (1928-29)3 6 , 396-575. Maxfield, V A and Miket, R 1972 "The Excavation of Turret 33b (Coesike)', Archaeol Aeliana*, 501 (1973), 145-78. Miller N 192 S , Roman e 8Th Kilpatrick.d FortOl t a Richmond 193A I , 6 "The Antonine Frontie Scotland'n ri , / Roman Stud, (1936)6 2 , 190-4. Richmond, IA 1940 "The Roman Redesdale'n si , History of , (1940)V X , 63-159. Richmond A 196 I , 3 Roman Britain. Richmond, I A 1966 Handbook to the Roman Wall (12th ed). Richmond A 196 I , 9 Archaeology. d Romanan t Ar Robertson S 195 A , Antoninen 7A Fort, Golden Hill, Duntocher. Robertson S 197A , Antoninee 0Th Wall (4th ed). Salway P , 196 Frontiere 5Th People of Roman Britain. Simpson, F G and Richmond, IA 1934 'Report of the Cumberland Excavation Committee for 1933', Trans Cumberland Westmorland Antiq Archaeol Soc (1934)4 23 , , 110-58. Simpson, F G and Richmond, I A 1935 'Report of the Cumberland Excavation Comm. for 1934', Trans Cumberland Westmorland Antiq Archaeol Soc (1935)5 23 , , 213-58. Simpson, F G and Richmond, I A 1937 'The Fort on Hadrian's Wall at Halton', Archaeol Aeliana2, 14 (1937), 151-71. Steer, K A 1961 "The Severan Re-organisation', in Richmond, I A (ed), Roman and Native in North Britain. Steer A 196 K , 4 'John HorsleAntonine th d yan e Wall', Archaeol Aeliana (1964)2 4,4 , 1^10. Swinbank ,B 196 6 'The Vallum—Its Problems Restated' Dobsond Jarrettn an ,i G (edd),B M , , Britain and , 85-94.

The Society indebtedis Civilthe to Service Department granta for towards the cost of this paper.