Proc Soc Antiq Scot, 115 (1985), 59-66

History, archaeolog Romad yan n Michae JarrettG l *

SUMMARY littleTheretoo is Roman documentation adequatean for history Romanof Scotlandbe to written. There is a consequent desire, and perhaps a need, to use archaeological material as a support substitute.or Some aspectsconsequentthe of problems examinedare light the recentof in discoveries interpretationsand on Romanthe of frontiers.

e las0 year2 Th t s have see a vasn t increas e amounth n i f earchaeologicao t l evidence available for all periods and for all regions of Britain. The quality of the evidence has also improved materia w e characte ne f variou Th o .e s th i l f o sr kinds site w e constantlar sNe . y being discovere fieldwory b daeriay b d l kan photography e numbeTh . f excavationo r s increaseha s d greatly d excavationan ; s hava muc n eo h tendee largeb o t dr scale than formerly, greatly enhancing the potential reliability of the conclusions drawn. These factors have all played their part in changing our understanding of Roman Scotland. e mustW , however e awarb , e evidencthaw ne t s stili e l appearing annually e outsideth o T .t i r seem syeaa tha t r no tpasse s withou discovere th tfort w temporard ne san f yo o t ys i camps t i d an ; credie th f thoso t e workin thin gi s field that man site w subjectee yne ar s excavatioo dt n withina very short tim f theieo r discovery. Admittedly, witmaio htw n period f Romao s n occupatiof o n very different character it is often easy to guess at the likely date(s) of a Roman fort, but it is still importan havo t e confirmation. Withou mighe w t i t easily dra wtotalla y false picture. In contras Romao t t n military sites , e assume whicb limitea y f o h ma belono e dt on o gt number of patterns of occupation externally ordered, civilian 'native' sites present much greater difficulties of interpretation. While a site may be datable on general grounds to the first five onln centurie relatee ca Romae y b t i th o D dt n sA occupatio r abandonmenno f Scotlano t d after extensive excavation coupled with considerable good fortune. Many similar sites will be excavated before a convincing pattern emerges, if it ever does. This means that any account of Scotland durin e centurieth g f Romao s n Britai e s heavillikeli b n o t y y biassed toward e Romath s n evidence. This is of course compounded by the (random) survival of a certain amount of documentary evidence fro e Romamth n world littla ; f thieo s relatenative th o t se population writtes i t i t nbu , by Romans for Roman readers. Little of what survives suggests that Roman ethnographers were carefu r reliablo l e observers; most t wer collectort observerbu no el al f reportt o sa sd an s travellers' tales at second and third hand, writers more at home in the library than in the field. It is unrewarding nigd an ,h impossible relato t , e their statement archaeologicae th o t s l evidence. When we turn to military and political history we might hope to find archaeological and *Dep Archaeologyf o t , University College, Cardiff | 6SOCIET0 Y OF ANTIQUARIES OF SCOTLAND, 1985

documentary evidence blendin produco gt singlea e coherent story. Suc hopha illusorys ei t i f I . wer t therno e e woul e littlb d e scop r excavationfo e , whic a remarkabl s i h y expensivf o y wa e obtaining information, justifiable only when there is no cheaper alternative. The first problem is that Roman historians were Romans, writing at Rom r otheefo r Romans detaile th :distana f so t province were irrelevan could an tomittede db . Moreover, histor para f literaturs o t ywa e wits hit own rules which did not include a need to seek objectivity or to admit to ignorance. It is of interest tha anciene th t t write probablo wh r y comes closesr idea historiaa ou f so o t t Luke S s ni . Becaus f theieo r background attituded an s e Romath s n historian mentioo swh n Britaie nar excessively brief and infuriatingly vague. To this may be added a propensity for including material whic f scano s i ht interes e moderth o t t n scholar e classiTh . c exampl s probabli e y Cassius Dio, admittedly abridged for a medieval Byzantine readership. He tells us of a serious disaster to the Roman army durin e reiggth f Commodusno whicn i , governoe hth f Britaio r killes nwa d wits hhi accompanying troop barbariany b sd crosse ha e walo th d lwh s which separated them froe mth Roman forts. Commodus sent Ulpius Marcellu deao t s l wit situatione hth ruthlessle h d an t , ypu dow e barbariansnth . Seventy-five percen f wha o tsayo concernes Di tsi d wit ecentricitiee hth f so Marcellus. In the remainder of the passage we are given no indication of the date, of the scale of Roman casualties, of the native tribes involved or the area of campaigning. Much modern controversy would have been avoided if Dio had indicated whether the Wall was Hadrian's or tha f Antoninuso t . fillee b gapSome n dth sf ca froe o m other evidence inscription A . n from suggests most strongly that, at a date between c 175 arid 190, legio VI Victrix received a detachment of experienced soldiers from Noricum (RIB 2148): the most likely occasion is the disaster recorded by Dio. Coins of 184 record victory in Britain, presumably in that year or in 183 (RIC Commodus 437; 440). Marcellu s associatei s d wite buildin th hn aqueduc a f o gd anothean t r structurt a e Chesters on Hadrian's Wall (RIB 1463; 1464). Another inscription, from Benwell, indicates that the fort was occupied, and has been interpreted to show that Marcellus was appointed in 180, thougl scholaral t hno s accept this interpretation (RIB 1329 Jarretf c ; t 1978). Opinions diffeo t s a r whether this governorship marke e rebuildindth f Hadrian'go s Wal e Antoninl th (Perio r o ) deII Wall (Perio r neithero ) dII . John Casethae m pointes t o ytherha t stront a s deou i attractivd gan e case (whic e planh publiso t s detailn hi r thinkinfo ) g thae activitieth t f Marcelluo s havy ma se extende dnorte welth f Hadrian'o ho t l campaignins sshoulwa e Wallw e h 3 f dI .18 go t fro1 m18 certainly expect some majo rt recorchangesno d ddi them o ,r problemeveDi Ou .f i n e ar s compounded by the fact that the memory of Commodus was condemned after his assassination and inscriptions bearing his name were cast down. They are rare throughout the empire and none survives in Britain. With their destruction we have doubtless lost other information about the activities of Marcellus. When we turn to the Severan campaigns of 208-210 Dio is little more precise. Two tribes, Caledonee th Maeataee th d namesan e ar , eved dan n located afte fashiona rvaguel o s t bu ; y that some modern scholars plac Maeatae eth norte Antonine othersoutd e th th th ho an f e t ho o st e Wall. No place-name survives in Dio, though this may be because British place-names meant even less in 12th-century Byzantium than third-century Rome. Beyond this we have only the reported intention to conquer the whole island and a reference to astronomical observations to indicate that som e campaignin t leasth ea f o t g took plac e Forth-Clyde nort th welth f ho o t l e line. Once we have accepted that we can point to temporary camps which may belong to these campaigns e stile proposeh W .l w hav ho ideo o hol n det f o ad 'the whole island'; possibls yhi permanent dispositions were never made. The only sites in Scotland which seem to have been occupied are Carpow, Cramond and possibly Old Kilpatrick. JARRETT: HISTORY, ARCHAEOLOGY AND ROMAN SCOTLAND | 61

e earliesTh t wor f Tacitusko Agricola,e probablo th , wh biographe n th s yi ma spena f yo a t doze r morno e year Britain si t variouna s stage careea f so r whose governorshiapogee th s ewa f po province th t recordei t 83/577/o n bu t ei 8 ; d only seven place-name fivd esan tribes. Thre f theseo e twelve cannot be located with any certainty. This geographical uncertainty makes it difficult to interpret Agricola's campaigns, while providing plentiful material for speculation and controversy. The Laird of Monkbarns was neither the first nor the last antiquary to be obsessed with the site of Mons Graupius, which remains as elusiv s evera e . Recently Professo t JosepS r a s stronmad t ha h ou eg casr placinfo et a t i g Bennachie. On present evidence this looks the most likely contender for the site of Agricola's great victory, though only discoverie site th e t wilsa l prov pointe eth shoule W . d remember that locatioe th f morno e temporary camp f Flaviaso n type leay otheo dst ma r suggestion futuren si 0 3 ; years ago Bennachie would probably have seemed impossibly far north. e searc Th r Monhfo s Graupiu perhaps si s most significan symptoe th r struggls a ou t f mo o et relate historical documents to the evidence of archaeology. There is too little of either for us to feel any great confidence in any proposed solutions. Great efforts have been made to identify different group f campassociato so t d san e them with campaigns recorde Romay db n writersd an ; some succes bees ha s n achieved e perhapW . s forget that ther havy ema e been many campaigns which escaped such record but left traces on the ground. No sites have yet been proposed as evidence for the various fourth-century campaigns against the Picts, yet we have no reason to suppose that the late Roman army did not protect its encampments in hostile territory. We have principao tw l group f campe lato s th e f firso s t century. Thes e convenientlear y assignee th o dt sixth and seventh campaigns of Agricola, in complete ignorance of the activities of his successor. Once year- Flavia upo0 o 2 y sag timna sa n e- Scotlan simpla s dwa e affai invasiof o r d nan fort-building by Agricola, followed at once by the abandonment of his northern conquests and of virtually all forts north of the Forth, and a dozen or so years later by withdrawal to the Tyne-Solway line. Archaeological evidenc multiplies eha d since thestild nan l expands apacee Th . complexity of the data has been paralleled by the complexity of interpretations on offer, despite the fact that these are liable to modification in the light of new discoveries. Two recent accounts of Roman activitie n Scotlani s d have already suffere y (Breezn thii dwa s e 1982; Hanso& n Maxwell 1983). Their interpretations of the Flavian period require substantial modification to take account of the discovery of a fort at (Maxwell 1984) and should be read in conjunction with St Joseph's very different (and more traditional) interpretations (Pitts & St Joseph 1985) t JosepS . h incidentally provide morr fa r thia se fo usefus p perioma l d than wile b l found in either of the other works. records that in his third season (79/80) Agricola reached the Tay; the fourth year was devote consolidationo dt , includin establishmene gth f forto t s between Clyd Forthd ean e th ; next campaign began with a sea-crossing and ended with Agricola's army on the coast facing Ireland, therefore either in the Rhins of Galloway or the Mull of Kintyre. The sixth season involved close co-operation wit n aree fleeta th han i e Fort,nort th f ho h where forts already existed, since they were attacke Caledonianse th y db ; subsequentl arme yth y operate thren di e divisions. Agricola's last season was dominated by Mons Graupius, evidently to be placed far north of the Forth. After it a leisurely return to his base took him through the land of the unidentifiable Boresti. In the ensuing winter (83/4 or 84/5) he handed over to his successor a peaceful and secure province. Excavations and reappraisals in recent years have established that Agricola's forts between Clyde and Forth do not underlie those of the ; Hanson & Maxwell (1983) regard Castlecary, and as possibly Agricolan, though they would admit that the case is 62 SOCIETY OF ANTIQUARIES OF SCOTLAND, 1985

far from proved. At no other Antonine Wall fort is there any case to be made. The only certain Agricolan sites are and the newly discovered small fort at Mollins, with Barochan Hill to wese Elginhaugd th an tisthmuse t enablth ease establiso no fortf o t th o to s o hed t u Tw s. e hth line chosee lookin ar e isthmu e Agricolath y nb w t g a muse f i s w d t an awai, t further discoveries. The new fort at Doune suggests that Agricola's line may have lain well to the north of that of the Antonine Wall, close to the edge of the Highlands. Maxwell suggests that it would consist of forts t Dumbartoa n (not known t reasonablbu , y postulated), , Mailing, Bochastle, Doune and near Stirling (not located); St Joseph rightly questions the outlying fort at Bochastle as part of this scheme, since it is in a weak position and produces a quite unnecessary salient in the line. Bochastle is in any case a problem, but surely makes better sense as an addition made whe advance nth e north sixte camth hn e i season whed an , schemna evidentls ewa y drawr fo p nu a road with forts leading towards Strathmore, with 'glen-blocking' forts to its north-west. The other sites of the 'Clyde-Forth line' fitted neatly into this new scheme, and may indeed have been planned with thi minn si d shoul emperoe dth r grant permissio advancen a r nfo . Opinion seem hardenine b o st viee th w o gt tha legionare th t y fortres t Inchtuthia s mosd an l t other forts beyon e Fortth d h were constructe unknows hi Agricoly b y b t t dno n bu asuccessor , thoug t hJosepS Pittd thein an shi r definitive repor Inchtuthin o t l favou Agricolan a r n origio ns strongly tha t givdetailety theno e an o yd d consideratio alternativee th o nt t deservei ; s discussion though not necessarily acceptance. Certainty is not possible, and it would be sensible if all agreed on that if on nothing else. Breeze's extreme view that 'it is probable that no forts were constructe Agricoly db a e Forthnortth f s unlikelho i ' meeo yt t with much support. Something here depend date th en preferreso Agricola'r dfo s governorship arrivee h ; dsummee latth n ei f o r d lefan t , afte furthex 78 si rr eithe o r7 7 rsummers , therefor eithen i e r 83/ r 84/54o . Scholarly opinion remains evenly balanced. Inchtuthi demolishes wa l d before completion; mint coin6 8 f so demolitioe inth n levels provid terminusea post quern. Nothing certainly late coms ha r e froy man e single-periooth f d forts t InchtuthiA . e labouth l r camps exhibi phaseo tw t f useo s . Whils it e internal buildings were of timber the fortress was given a stone wall, perhaps a modification to the original scheme. Demolitio t Inchtuthia n e th ld musan , t 87 surel r o e assigne b y6 8 o t d construction work must have occupie t leasseasonsda o tw t ; three season t inconceivableno s si . A post-Agricolan date depends on two linked assumptions, neither of which can be proved: that fort-building would only occur after campaigning was finished, and that it could not occur in the same season as campaigning. But such evidence as we have suggests that the campaigning season probably lasted onl o Septembert y y fromMa , giving ample tim r othefo e r activities; building in timber is possible in times of frost; and clearly Agricola built forts at times other than his fourth year. Having described this appropriate activity for his ideal governor under that year Agricol neeo n mentioo dt d aha t againni mory an , e tha needee nh mentioo dt provisioe nth f no civic amenities after its record in the second year. accepte Iw f I believ s a , shoulde ew latee th , r dat r Agricolefo possibls i t ai e that building begasixts hi r sevent ho n i h season. Even site t begusno r completeno d under Agricola might still have been planned by him. It is improbable that Agricola had made no plans for garrisoning the area he had conquered. The account of his sixth season clearly implies that there were forts north e Forthoth f ; Breeze's plea that Tacitus doe t specificall s unwontedlno si o s y sa y y perversef I . forts existed they were presumably intende 'permanent'e b o dt remaio t e i , lons na thes ga y were t requiredunreasonablno s i t I . o suppost e e that they formed par f somo t e overall scheme, whethe t thano tr t completedschemo r ye s ewa . There are two main elements in that scheme. First is a road with regularly spaced forts running in a virtually straight line from Doune up Strathallan into Strathmore at least as far as JARRETT: HISTORY, ARCHAEOLOG ROMAD YAN N SCOTLAN3 6 | D

Stracathro near the North Esk. This marks the main line of advance and presumably the principal supply route. A short distance north-west of this is another 'line' of forts on the edge of the Highlands, placed at the mouths of the major glens; in the Tay valley is the most important of these. This line is more apparent on the map than on the ground. There is no evidence for a road linking these forts and on topographical grounds none is to be expected. They are traditionally seen as blocking the exits from the glens, whence hostile highlanders might debouch to threaten the supply route. This was probably one initial function for them, but Breeze is surel ythee righ se advancn bases m a a o glen e t tr th s fo p se u whic presumabls hwa y plannes da e nexth t major conquest t woulI .naturaa e db l consequenc f Moneo s Graupius t InchtuthiYe . l suggests more than this. Like Cheste suitabls i t i rbasa r advancs a edirectionso fo tw n ei p u : Killiecrankio t Strat y hTa beyondd an e alsd an o, continuin e maigth n line throug Mearne hth s towards Aberdee Morae th d yn an Firth . Flavian marching camps indicate thalattee th t r ares awa know Romanse th o n t occupatiod an , f territorInverness no a r fa havy s ya sma e been intendedf i ; it was, a road across the Highlands would seem to be essential. Exploration of the area to determine the best routes across may be a further explanation of the advanced forts: while the dispositio f fortno s indicates thaRomane th tclea a d rgeographe sha ideth f ao f Scotlanyo d this can only have been achieved by a considerable amount of scouting in the years before final conquest. Whatever Agricola's plans were they came to nothing. Crises nearer home compelled to withdraw troops from Britain, though we need no longer suppose that in 83 all Agricola's legions were below strength (Kennedy 1983) withdrawae th : f vexillationo l s bettei s r connectee placeb y ma d e d latedth an wit, n abandonmenri e hth Strathmorf o t e rather thas nit e fouth rf o legionse lates e th occupation t on ta / / Adiutrix, , 2 9 y d beeB ha . n permanently transferre Danubee havth y o dt ma e earliese t . Tha lefI earl.s th a 86 t s i s ty a t possible th dat r efo abandonmen r frof Inchtuthilo fa t ms i certait i t bu ;n thalegionare th t y fortres t Wroxetea s s wa r given up as early as this. Current opinion is that for a brief period after the retreat from Strathmore occupation continue e Earnr nortth fa s s ha .da Ardoc Strageatd han h both sho periodo wtw f Flaviao s n occupation thoughs i t i d tan , thasecone th t f thesdo e followe withdrawae th d l from Strathmore. Strageath seems to be the pivot of a system of closely spaced wooden towers extending south to Ardoch and north-east along the towards Bertha (Perth). These, it is suggested, represent a short-lived frontier protecting the (possibly) pro-Roman folk of Fife from their anti-Roman neighbours. They seem pretty inadequate to thwart more than small raiding parties, and it is difficult to see that the garrisons of two forts can have been regarded as a match for Caledonian forces even after Mons Grapius e evidencTh . r thiefo s particular phasr frofa ms ei conclusive: the 'Gask Frontier' depends more on our difficulty in fitting it into our model of Agricolan activity than it does on firm evidence for a separate existence. We may note that Doune soutth ho t e appear havo t s e onle perioyon f occupationdo d alsan o, thae towerth t s t necessarilno e ar typeo d tw ysame an s f th paro appea f ee o tb syste o t r m (Frer t JosepS e& h 1983, 135-6). The towers and the first period at Ardoch and Strageath could well belong to the middle year f Agricolao s , wite seconth h d period corresponding wite advancth h e into Strathmore. e latesTh t relevant coin from eithe. Breez86 rf o esit s i egoe s beyon e limitth df o s reasonable speculation whe usee seconne h sth d perio t thes da siteo suggeseo tw st t tha 'roade th t ' forts as far as Stracathro may have been occupied after the abandonment of Inchtuthil; when we have some evidence fro foue mr morth o r time th siteadvanco et e questiob n s i y ema suct ni ha hypothesis. | SOCIET 4 ANTIQUARIE6 F YO SCOTLANDF SO , 1985

New discoveries have come as fast in the Lowlands as they have further north. They have, however, merely been adding detai knowa o t l n patter f occupationno e tim r reappraisaTh .efo l will presumably come wit discovere hth postulatee th f yo d sites wes f Gatehouso t f o Fleef eo d an t Loudoun Hill; there can be little doubt that these sites remain to be discovered, since neither fort make strategiy san c sense without other sites beyonaftes i withdrawae t I th r . dit l from Strathmore thaLowlande th t s engag r interesteou . Now that we no longer believe in Agricolan sites below those of the Antonine Wall we no longer have caus o carrt e y Flavia I occupatioI n e Forth-Clydr nortth fa s s a ha n es lineha t I . become increasingly clea recenn ri t years that fro late Romas mth e80 n occupatio t extenno d dndi beyond a line from Glenlochar to Newstead, though there are suggestions of activity at Loudoun Hill (with no less than four Flavian building periods) and perhaps at Castlecary. The new forts at Newstea Dalswintod dan n were both large mord an r e strongly defended than their predecessors. Hanso Maxweln& l suggest that these havforty ma es been base r long-rangsfo e patrols alsd an o, that some of the troops withdrawn from Scotland may have been located in the Lake District which does not seem to have been garrisoned before about 90. In this brief period nothing suggests that there was an organized frontier - still a new concept to the Romans; nor does the Glenlochar-Newstead line seem to take account of tribal boundaries. It is indeed difficult to see tha t representi t s more tha furthese nth t lin whico et available hth e troops coul stretchede db . t held no casy s e lin. an eth Probabl ewa n I resula s ya f enem o t y action forts e nortth f ho Tyne-Solway line were abandone earle th yn d i second century. Agricola's road between Carlisle and Corbridge, the Stanegate, became the effective frontier. Extra forts and fortlets were added t undei o t r Trajan , t cleathougno s ri that hi overaltn a ther s ewa l plan recenn I . t year have sw e learned that the Stanegate was extended west of Carlisle to Kirkbride close to Moricambe Bay [not Morecambe, as Hanson & Maxwell 1983, 48]. East of Corbridge the route of road and frontier remains uncertain, thoug e discoverhth eartn a timbed f han yo r for t Whickhama t , west of Gateshead, indicates tha shoule w t lookine db g rivere sout th shoule f W .ho d have expected this, since Corbridge marks the lowest crossing of the Tyne until the reign of Hadrian. Presumably at least one more fort is to be expected between Corbridge and Whickham, with another eas f Whickhamo t , since South Shield knows si Hadrianice b o nt . If these deductions abou Stanegate th t correce ear have w tadequatn ea e explanatioe th f no early move of the fighting garrison of Hadrian's Wall into new forts on the line of the Wall itself. With bridges onl t Corbridga y d Newcastlan e e this stretce Stanegatth f o hs thoroughl wa e y impractica basa r troopss a elfo same havy th ; ema bee nwestere truth n ei n sector, where eth Irthing canno fordee b t d whe spaten ni . Recent excavations have greatly increased our knowledge of Hadrian's Wall, notably with the recovery of most of the plan of Wallsend fort. Also increased is the amount of the Wall accessibl visitorso et . Whil egreaf o the e ty ar interes t thes discoveriew ene t necessitatno o sd y ean serious reinterpretatio Wale th f ln o durin builders reige it gth f no . Breeze (1982) makes mucf ho the fact that we have no firm evidence for a walkway on top of the Wall, though its existence is dust-jacketn ow usualls hi n yo . s assumeAbandonini t i s a d- walkwae gth y leave withoum shi n a t adequate explanation of the width of the curtain wall. More important is his recognition that at onlmilecastleso ytw hav evidence ew causewaa r efo y acros Wall-ditche sth ; ther similaa s ei r lack of a causeway at many of the forts on the Antonine Wall. Since all interpretations of Phase I of Hadrian's Wall assume that milecastles provided gate whicy b s h civilian armd an s y units might cross the frontier line there is clearly an urgent need for excavation outside several milecastles to determine whether a bridge was provided or whether a primary causeway had been removed at a later date. If there is evidence for neither it will be necessary to think again about the way in JARRETT: HISTORY, ARCHAEOLOG ROMAD YAN N SCOTLAN5 6 | D

which the Wall operated. Any such reappraisal will need to take account of the functions of the Vallum. Establishe Phasn t createi di , II e a dmilitar y zone through which civilian passags wa e only fortw permitte f thine o s se Phaseth t defensivda e Vallua t Th . no ms r evei (o e n defensible) earthwor havn ca ed nothinkan wito d ho g t (unproven) Brigantian hostility. Over the last decade our understanding of the Antonine Wall has developed enormously, even though we are no closer to determining the reason for the return to a Forth-Clyde frontier. plannins It constructiod gan seee b relato nt n nca e more closel thoso yt f Hadrian'eo s Wall than was once thought. It is now clear that the Antonine Wall went through considerable changes between its original planning and its completion. The first intention seems to have been to build in stone, for the fort at has stone walls with projecting wings as though a stone curtain was expected. On the other hand the stone walls of Castlecary suggest that the builders expected f turf simpl y t o thie curtai e bu th ma ;s b yo t nmea n thae defenceth t f Castlecaro s y were built rather later than those of Balmuildy, when turf was already being used for the curtain. Both these forts are amongst the six or so primary forts, each large enough to contain a complete auxiliary unit; to this primary stage also belong a number of fortlets, comparable in size with the milecastles of Hadrian's Wall. Breeze suggests that they may also have been regularly spaced, perhap t milsa e intervals: this coul dusefua prove b lo et hypothesis moro n s i e t thai t nbu , thao s t far. Followin discovere gth fortletf yo searce sth begu s h ha turretr Antonine nfo th n so e Wallt bu , so far without success. Before the building of the Wall was completed the scheme was modified by the addition of another ten or twelve smaller forts. No clear reason for this modification is discernible onls i t i y t recentlbu , y thae modificatioth t n itsel s beeha f n recognized. Breeze argues that building started in 142, the year in which was saluted as imperator for the British victories, and that the intention was to complete the work by 144. Delay would have been occasione e buildin th secondare y th db f go y forts. Breeze suggest s muca s s ha three or four years 'if experience on Hadrian's Wall is a sound parallel'. The parallel is not sound, for virtually as many troops will have been available for the same number of smaller forts, built mainly of turf and timber rather than stone. We may suspect that Balmuildy at least was built befor inscriptiono e tw 142 r fo , s fro e formth t recor e governodth r Lollius Urbicus (RIB 2191; othet y nameno an rs n i 2192d inscriptioo o wh ) (RIBritai9 n i n 13 s B fro y Walle wa nb m th e H . 1147, Corbridge) and was known to the Augustan History as the governor responsible for the new conquest (SHA Antoninus Pius 5,4): the normal tenure of a governorship at this period was three years, and Urbicus will surely have returned to Rome in time to participate in the celebration of the imperial acclamation in 142. As with Flavian Strathmore we may have one governor making plans and initiating construction, leaving his successor to complete the project: whether the successo f Urbicuro responsibls swa change th r efo e from ston latee turo et wels th a frs a ldecisio n to add extra forts, we shall probably never know. There is no serious doubt that the first abandonment of the Antonine Wall occurred in the mid-150s, something like 10 years after the Wall was finished. The reasons for this are from clear. Evidenc r destructiofo e y hostilb n e actio f siteo n n Scotlani s r northero d n Englan s veri d y restricted (Hanson & Maxwell 1983, 144-7). Evacuation by the governor without imperial authorization is most unlikely. Shortage of troops may be a factor, for the second Antonine occupatio f Scotlanno d seem havo t s e involved considerably fewer troopse tha w e first t nth bu ; hav evidenco en f trooeo p withdrawals from Britai e 150sth n .ni Abandonmen Antonine th f o t e Wal usualls i l y associated with reoccupatio f Hadrian'no s Wall under lulius Veru 154-c n si 8 (RIB 138 specificall9s i f 158)y o t curren bu , t wisdom would hav Antonine eth e Wall reoccupied within vera y short timfinalld ean y abandone 163/c n di 4 whe frontiee nth r returne Hadrian'o dt s Wall (still in Period I B which began in 158). A final abandonment of the Antonine Wall in the early 66 | SOCIETY OF ANTIQUARIES OF SCOTLAND, 1985

year f Marcuso s Aureliu itselfn i , sis , readily explicable. Marcupersonao n d sha l involvemenn i t the project; a new reign was the normal time for a reappraisal of old policies and the outbreak of the Parthian War may well have induced the emperor to reduce his commitment in Britain. probleme th t Thae reano s Th l.i t troubl e overalth s ei l uncertainty whic s arisehha n from e attempth writo t t e history from archaeological evidence e see W .mhavo t e Antonine WalI l endin g155c ; Hadrian's Wal beginninr earlierB o I l 8 15 ; n gAntonini e r late Walo 8 rfroI I l 15 mc to c 164; Hadrian's Wall I B resumed from c 164 and lasting until 180 (or 197 for traditionalists), to be followed by Hadrian's Wall II. This essentially archaeological interpretation, which endeavours to link itself to the sketchy and uncertain documentary evidence, may be correct - though it certainly cannot be proved so at present. There are, however, two historical questions which must be asked. Can it be explained adequately? Does it make sense? To both those question answee sth . r musno e b t To the question 'Is there a better solution available?' the answer must also be negative. The historical material alon capabls ei producinf eo g more sensibl admittee b o et solutionss dha t i t bu ; that at present those solutions do not seem to be compatible with the archaeological evidence. We ma e wronyb seeo gt k archaeological answer historicao t s l questions whe e strictlnth y historical materia categorieo e havmarro thist tw o ll trie W d r e fail al e. o ,o yth dt us sf evidence o s s i t I . possibl mato et lioea n wit tigresshresultana e th t bu ; t offspring doe t resemblsno e either parent l closelyal t a , evea handsom e nb thoug y d attractivma an e t i h e animal s presumabli t I . y also sterile.

REFERENCES Breeze, D J 1982 The northern frontiers of . London. Frere, S S & St Joseph, J K S 1983 Roman Britain from the air. Cambridge. Hanson Maxwell& S S W , 198 G , 3 Rome's north-west frontier. Antoninee Th Wall. Edinburgh. Jarrett, M G 1978 'The case of the redundant official', Britannia, 9 (1978), 289-92. Kennedy, D 1983 'C. Velius Rufus', Britannia, 14 (1983), 183-96. Maxwell S 198 G , 4 'New frontiers e Romath : n fors t possiblDounit a t d an ee significance', Britannia, 15 (1984), 217-24. Pitts t Joseph, S Lyn S 198& K nF J , 5 Inchtuthil. Romane Th legionary fortress. London.