Proc SocAntiq Scot, (1988)8 11 , 131-137

The histor Antonine th f yo e Wal reappraisaa l- l J C Mann*

ABSTRACT Literary, epigraphic numismaticand sources analysedare orderin provideto frameworka by which archaeologicalthe material from northernthe tentativelyfrontierbe may dated.

Consideratio histore northere th f th no f y o n frontie Britain ri n confirms tha have tw e onlo ytw absolutely fixed points, when it comes to the attempt to date archaeological material. The first fixed point is the building of 's Wall. The combination of literary and epigraphic evidence which enable identifo t groune s th su n yo d sites built between aboud tan A2 D12 130 need not be recapitulated here, but that dating is quite secure. The pottery from the lowest levels of turretsd san sam e buill th al , t eta time, shoul identifiable db Hadrianics ea situatioe th : n is consolidated by the fact that Hadrian's Wall was given up soon after construction, so that it is readily possibl distinguiso et h level materiald san latef o s , whicr be date y h .ma are Withir o , n this short perio 122/130dc archaeologe th , sectoe th f yro milecastl milecastlo t 9 e4 enable1 e5 o t s su identify milecastle 50TW as a short-lived site dating to c AD 122 and a very few years after - before it was replaced stony b e wall milecastl northe th o t , whic0 e5 h produced very similar pottery. This must be the most closely dated site in . The vital point to note is that it is only because we are dealing with sites which are being built on for the first time, and because the date of that building can closelo s e date b suro yth thif s e o f fixed ee so b grou n , ca Hadrianif tha po e w t c pottery. The second fixed point is the building of the Antonine Wall. Again, the combination of literary and epigraphic evidence which enables us to identify on the ground sites built in the early need not be rehearsed. Since there was manifestly no Hadrianic occupation of any site on or near the Forth-Cl isthmue yd s (an possiblo dn e occupation fo rreig e mosth Traf f n o to either) n ja , most siten so Antonine th e Wal counn ca l sites a t s occupie firste th r time.dfo (Flavian larg d materiaan ey b s i l sufficiently different that any Flavian occupation does not really affect this view.) In theory then, it shoul possible db dato et e material fro lowese mth t level occupatiof so Antonine sitef th n o n so e Wall reige th Antoninuf no o t s Pius. Again must i , stressee tb d thaonls i t ti y because Antonine Wall sites bein e firste ar th g r time buil fo dat n n ca o tthae e theiw t r earliest potter thio yt s reign. Hadrianic levels on Hadrian's Wall, and levels of the reign of on the Antonine Wall, are the only levels which can be firmly dated. At no later date do we have an example of a site, occupie firse th tr timedfo , whose dat constructiof eo absolutelns i y certain onct .I e seeme thougs da h Carpow might turn out to be such a site, but in fact its date of construction is quite uncertain. All later levels have, so far, been dated by guesswork. There is simply no way at all of dating levels afte firste reitheth n o , r Hadrian's Antonine Walth r lo e Wall. Pottery afte middle e th r th f eo second century cannot be given a firm date. The situation is compounded by the fact that, when the "Institute of Archaeology, Gordon Square, London 132 SOCIETY OF ANTIQUARIES OF , 1988

pottery from Hadrianic level Hadrian'n so s Wal compares li d with pottery fro lowese mth t leveln so the Antonine Wall, the two groups are found not to be totally distinct. There is a considerable over- lap. Potters unfortunatel benefie notth d r yarchaeologistsdi f ,fo o t , change their style potterf so y with each new reign. Much of the pottery can only be labelled 'Hadrianic/Antonine'. f somo Potter e eb valuen yca somn i , e cases providinn i , g relative chronologies t cannoi t bu , t give absolute date turs sit unlesnn i bases i t sfiri n do m literar epigraphid yan c evidence dates i r ,o y db such absolute scientific means as may become available. It must be emphasized that attempts to elucidate the history of the Antonine Wall which in any way depend on pottery have no value. A firm literary and epigraphic basis has never been available for the period after the middle of the second century. The attempt must be made to see whether it is now possible to supply such a basis. In particular, in default of closely dated pottery, we must also look to see what can be gained from a study of other material, particularly coins: coins are after all dated. But first we must look carefully at literare th epigraphid yan c evidence. vitae Th l startin inscriptioe th e b g o poin t whic Af 8 ns o D 15 ha t h records reconstructioe th f no curtai Hadrian'f no s Wall, somewhere between milecastle (RIB3 1 d 1389)an 1 s1 thif I . s stone really came fro mfora t (eg, Rudchester claime s wese it th f o to t d o mil a ,findspots r eo valus it ) e mighe b t questioned (perhaps improperly e accep e w below) f se ,i tt thabu ,t cami t e fro e curtaimth f no Hadrian's correctl s Walli d an , y dated, the have n w accep o et t that Hadrian's Wal reoccupies lwa s da frontiea r lin An ei D 158. Wit reoccupatioe hth Hadrian'f no s Wall Antonine ,th e Wall must have been abandoned. Not only is there no precedent for, or likelihood of, two lines being held at the same time, bupoine bees tth tha n supporte wore th Bria f ky o d b n Hartley differenstude s 1th Hi . f yo t dies useo dt stamp the work of different potters is what Richard Wright has correctly called 'pure epigraphy'. He demonstrated tha overlae th t stamp-dief po pottern so Wallo ysmalo s tw fro s se i lm th tha t then yca never have been held as frontier lines simultaneously. In other words, when one frontier is manned, the other is not. In passingnoty furtheema a e w , r consequenc Briaf eo n Hartley's work placee H . s and same th n esi categor forts y a Hadrian' n so s appeart Walli d an , s tha muse tw t abandoe nth view that individua Hadrianie l fortth n so c line could have been held, even whe frontiee n th o y la r Antonine th e line. This will suggest that inscription Antoninuf so s Pius from Benwell (RIB 1330d )an Chesters belong to the years 158/161, at the beginning of the renewed occupation of Hadrian's Wall.

In 2 theory evidence th , stamp-diee th f eo s (non f whiceo absolutels hi y dated) e woulb t dno violate Antonine th f di e Wall were finally abandone 158Hadrian'n dd i ,an s Wall were occupied down to the end of the second century and beyond. But that chronology cannot be correct if there is any material at all which implies an occupation of the Antonine Wall after 158. No dated inscriptions were set up on the Antonine Wall after that date (although several are in a style which could be dated to the late second century earl e ever ,o th yo nt third century) thert numbea bu ,e ear coinf ro s which havo et be taken into account. The following coins, dating after 158, have been reported from Antonine Wall sites:

Date Metal Site Publication Other references (1) 160 Marcus bronze PSAS 68, 28; Clarke, RIC 1354 - tr pot xiiii Cadder, 82 (2) 164/183 Lucill aKilpatricd silve Ol 78C 6RI r- k VenuMiller; PSAS 28 d , sOl ,68 Victri x granary Kilpatrick 33-4 (?) (3) 174 Marcus bronze PSAS 94, 134 RIC 1107, Cohen 433 (4) 176/192 ? PSAS 52, 224 (5) 176/192 ?bronz r HilBa le 4 PSAS22 , 52 Commodus MANN: THE HISTORY OF THE ANTONINE WALL - A REAPPRAISAL 133

In assessin value gth f theseo e coins have w ,tak o et e into accoun greae th t t decline th n ei quantity of coins produced after about AD 160. As Sutherland remarked (1937, 33), there is everywhere a sharp drop in the site-finds of coins, 'beginning with the reign of Marcus'. In other words late th e n seconi , d century, people were mostly firse coinforcee th us tf shalo o dt f tha o f t century. Since there seems in fact to have been a catastrophic decline in coin production - the result of the economic crisis of the late second century - there were few coins of Marcus to circulate, and even fewer of Commodus. We may note in passing that the only coins sealed by the second floor in the barrack-block at Birdoswald, in 1929, were two coins of Trajan (Trans Cumberland Westmorland Antiq Archaeol Soc, 2 ser, 30, 174). The paltry number of coins listed in the table above is thus evidence, not of a lack of late second-century occupation of Antonine Wall sites, but merely of the predictable lac coinagf ko thosn eo e sites. construcw no e Ifw chronologicaa t evidencle basie tablth th f sn o eo e discusse have faro w ds , e the following (using the conventional HW la and Ib for Hadrian's Wall, and AW 1 and 2 for the Antonine Wall, and leaving out of account the ephemeral third occupation of the Antonine Wall):

HW AW T-122

la

-Lcl42

-r c!58

Ib

-L X -r

-r Y

II

The occupation of Hadrian's Wall which began in 158 (conventionally, period Ib) continued dow data o nt e whic have hw e labelle whe, d X secon e nth d perio occupatiof do Antonine th f no e Wall began (AW 2). This occupation ended at date Y, when Hadrian's Wall again became the frontier. The 134 | SOCIETY OF ANTIQUARIES OF SCOTLAND, 1988

coins show tha musY t aftee li t r AD 176 earliese th , t possibl latese th dat r te fo coi n knowe th n no Antonine Wall probabld an , y well after, give paucite nth coinf yo f Commoduso s anywhere. furtheA date th r e o clut whe s secone a nth d occupatio Antonine th f no e Wall occurre, dis however, provided by the text of Cassius Dio, or rather by the epitome of his history by the rather unintelligent 11th-century monk Xiphilinus heae W . r (Dio 76,12,1) that 'the Maeatae live closo et the wall which divides the island in two, and the beyond them'. Cassius Dio was a membe Septimiuf ro s Severus's consilium, vern ougho y ma centrclose a wh th t e thingeo f t e on o d san hav o accest d eha accurato st e information about imperial affairs begae H . colleco nt t materias hi r lfo history probably in 197, and began writing probably in 207 (Millar 1964,28-40). There is no reason to believe thafabricateo Di t statemens dhi Maeataee th n o tfin e dW . traceMaeatae th f so e justh o t Antoninnorte th f ho e Wall t Myoa , t Hill north-wes Falkirf Dumyat o a d kan t north-eas Stirlingf to , and apparently stretching through into Strathearn (Maxwell 1975). This is important in showing not only thabelieveo Di t d tha Antonine th t e Wal stils northere th llwa n frontie reige th Severusf n o i r , t alsbu o that ther soms ewa e perio Dio'n di s caree gleane d whicn ri ha e hh d this information. This period can hardly pre-date AD180, when Dio became a senator, and presumably began to have some access to official information. In the 180s, it would seem, the frontier lay on the Antonine Wall. If coin Commoduf so Antonine rarth e sn ar eo e Wall, inscriptions whic htotalle namar m yehi absent. But then so they are on Hadrian's Wall. (We can leave out of account a small fragment from Netherby3, a dedication which is apparently dated imp. Comm. cos., ie AD 177.) Yet Marcus and his co-emperor wele sar l atteste Hadrian'n do hinterlands it sn i Wal d lan : the namee yar inscriptionn do s from Great Chesters and Stanwix, and (in the hinterland) from Corbridge, Ribchester and Ilkley

(RIB 1149, 589 and 636). What ha4 s happened to inscriptions naming Commodus? The answer lies in the manner in which damnatio memoriae affected inscriptions. The last empero suffeo rt r this, before Commodus Domitians ,wa cases hi n , I .whil e sometime nams s hi ewa simply erased, often the whole stone seems to have been destroyed: very few inscriptions of

Domitian survive case 5 Commodusth f eo n I . , while agai simple namb s ny hi yema erase ILSg d(e 390, 393-4), we know that inscriptions naming him might be totally destroyed. It is little more than chance that demonstrates this inscription a : n survive Tripolitaniae th n o s n frontier which recordw sho Severus ordere restoratioe dth statuea f no , wit revealinha g inscriptio accompano nt t (Annyi p E 1922, 53): (Sept. Severus, Caracall Getd aan a Caesar 201. D . A ,. ) titulum quod divo Commod[ofratr]esuoaerasum(\)fueratrestit[ue]rimtpervexil. leg. Ill Aug. p.v.. . . ['Septimus Severus, with Caracalla and Geta,. . .by the agency of a detachment of the legion III Augusta, restored the inscription to his (ie Severus's) divine brother Commodus, whic beed hha n destroyed'.] In AD 196, when Severus declared himself the son of Marcus Aurelius and the brother of Commodus, he set about attempting to rehabilitate the name of the latter, for example having his name restored on inscriptions (eg ILS 402,405), or seeing to the setting up of new inscriptions (eg ILS 403-4). The army of Britain had no cause to like Commodus. The precise nature of the disturbances recorde Cassiuy db (72,9,2-3o sDi Historiae th d )an Augusta obscures 6easi s i t believo yi t t ,bu e that in Britain inscriptions recording his name were destroyed with great enthusiasm when news was received of his damnatio memoriae and the accession of Pertinax. So far we have no evidence of any attempt by Severus to restore the inscriptions of Commodus in Britain. However numbea , inscriptionf ro s found alonAntonine line th gf th e o e Wall could well have been set up in the last two or three decades of the second century. Thus at , the vexillation MANN: THE HISTORY OF THE ANTONINE WALL - A REAPPRAISAL | 135 of II Augusta recorded in the same inscription with a vexillation of VI Victrix (RIB 2146) cannot date to the time of Lollius Urbicus, since all of II Augusta was present then in the Central Lowlands: if the whole legion was present, then a vexillation of that legion could not be there also7. There is no known occasion later in the reign of Antoninus Pius which would require the presence of these legionary detachments reasonabls i t i d an , suggeso et t that they wer t Castlecarea latea t ya r date date Th .e proposed for the men of VI Victrix in RIB 2148 (c AD 175/190) would fit this very well (Mann 1963, 487-8). The men of II Augusta and VI Victrix may have been building rather than in garrison: the garrison may have been coh. Ifida Vardullorum (RIB 2149). r HilBa l t similarlyA e vexillatioth , I AugustaI f no , e samagaith n eni inscription wita h vexillation of XX VV (RIB 2171), cannot date to the period of the building of the Wall, and presumably belong late th e secono st earlr do y third century curline th : glettee forth woulf mrG o d certainly fit that date (cf Evetts 1949, 162, 171). Again, they were probably merely building at Bar Hill indees a , inscriptioe dth n suggests garrisoe th : probabls nwa y coh. I Hamiorunf. At , the vexillation of II Augusta (RIB 2180) cannot belong to the time of Lollius Urbicus. Here vexillatioe ,th n probably was somt a , e later date garrisonn ,i . Thi suggestes si e th y db serie dedicationf so centurioe th y sb AugustaI I f no . Cocceiu,M tombstone sth Firmusy b n d a f an e o

ordinary miles (RIB 2181). 9 furtheA r point inscriptioe th : n recordin constructioe gth legionare th f no y sit t Carpowea 10 seems to record work done under a sole emperor. This could, on what coin evidence there is, as

easily be a fort built under Commodus as one buil11 t under Severus, as John Casey points out to me. absence Th e fro Antonine mth e Wal f inscriptiono l s mentioning Commodus f dateo d dan , inscriptions of the reign of Commodus, is no more indicative of a lack of occupation than is the similar absence from Hadrian's Wall internallt .Bu y dated inscriptions (and especially importan buildine tar g inscriptions) do begin again on Hadrian's Wall or in its hinterland in 197, to continue through the reig f Severuo n a continuin n i s g seriethird mi ds e dowcenturyth o t n . o buildinThern e ar e g inscription frol Antonine al mt th sa e Wal l internalle whicb n hca y reigdatee th Severusf no o dt r o , later contrase Th . onln ca ty mean that littla r ,o e fro 7 earlierm19 frontiee th , r linmoved eha d back fro Antonine mth Hadrianie th o et c line. Thus we come to attempt to assign dates to X and Y in our table. There is one perfectly acceptable archaeological event which coul associatee db d wit . ThihX s archaeological evene th s i t destruction atteste Haltot da n Chester t Corbridga d san e- suggestin attacn ga k southwards down Dere date Street th f tha eo t t Bu .even t cannofirmle b t yye t established, unlik e acceptableth e historical event. This is the destruction recorded by Cassius Dio (72,8,1-2), which must have taken place in the early 180s, and the subsequent campaigns of Ulpius Marcellus. The coins of 185-6 record his victory reasonablw no , n whicca e yh w sugges occupatiofollowes w ne twa a Antonin e y dth b f no e Wall pertinens i t (I . enquiro t e whether 'marching-campssome th f eo southerf o ' n Scotlant no e dar to be associated with these campaigns.) probabls i o Di y incorrec writinn i t g that Commodus sent Ulpius Marcellus agains bare th t - barians. More probably, Marcellus was already governor before the death of Marcus, and it was early in Commodus's sole reign thabarbariae th t n attack took place, killin glegatea , presumable yth commande I VictrixV f o r 12. Marcellus simply reacte fashioe th n di n require governoe th f do a f o r province under attack from outside crisise th n .I , Commodus quite probabl secona r fo dyn o kep m thi term, rather than bring in a new man. Marcellus would have to have the approval of Commodus for a return to the Antonine Wall. It is worth pondering whether Commodus may have felt that annexation of territor requires ywa justifo dt takins yhi title gth e Britannicus. If X can be dated to c 184, then it follows that the inscriptions at Chesters (RIB 1463-4) must dat juso et t befor attace eth k campaigne fro norte th mth d han Marcellusf so alae AsturumI I th f I . 136 | SOCIETY OF ANTIQUARIES OF SCOTLAND, 1988

occupied Chesters in period Ib, then after the return to Hadrian's Wall a few years later, it would hardl surprisine yb t returnei f gi Chesterso dt . Afte deate th r Commodusf ho governoe th , Britaif ro n Clodius Albinus seem t firss a havo t t e allowed himsel e fobbeSeveruy b b f o t f of d s wit e titl f hth 'Caesar'e o t afte e defeaBu .th r f o t Pescennius Niger, Albinus must surely have realized the ruthless nature of the victor- and that he was next in line for suppression. Once he began seriously to prepare himself to resist Severus, and make thronethe woulfor he , bid d neehis own collec dto manas t y troop possiblhe sas y coul weld(as las recruitin ones)w thigd ne san , give versa y good reaso retiremena r nfo Hadrianie th o t c line. Sucha retirement woul accompaniee db d perforc agreementy eb s wit northtribee e hth th Maeatae o st th , e and Caledones, agreement whicn si doubo hn t these peoples promise attaco t t aree dkno th a occupied by Rome - whether or not in return for money payments or other considerations. Albinus would not worrie o possibly to an e b y db e breache thosf so e agreements serioua f I . s invasion took placs hi n ei absence fighe th tn ,wit eitherwo he h Severus f i , woule h , dresourcee havth l empire eal s th f hi s o t ea command to make good any loss, and take revenge, or, if he lost, it was not a problem he need worry about: it would no longer be his problem. (Nevertheless, it seems quite plausible that he should have ordered the cities of Britain to see to their own defence, at their own expense, of course.) In fact there is no evidence of any invasion in his absence. Dio does not describe any invasion in 197, nor does anyone else. According to Dio (75, 5, 4), all that happened was that 'the Caledonians keedit dno p their promise madd san e read assiso yt Maeatae'e tth strine .Th buildin f go g inscriptions which begins in 197 thus refers to the normal rebuilding necessary when buildings, especially if they hav beet eno n continuously occupied, approac year0 hage10 f so .t represen Theno o yd t reconstruc- tion after enemy attack. Nevertheless Maeatae th , doubo en t too opportunitye kth , afte departure th r Albinusf eo o t , consolidate their position, including effort gaio keed t s an np contro f southero l n Scotland dan . Did they succeed for a time in subverting the Novantae and the ? Whatever they achieved will have bee campaigne nth wipey b t dSeverusf ou s subsequeno e th d an , t re-establishment of control over southern Scotland. We may therefore date Y to c 195, and complete our table accordingly. This chronology has the advantage tha evidencee t doei t th violatt f o sno y literarye e ,an thal th leas f al o tf o t , epigraphid can numismatic sources. In default of such certainty as can attach to the dating of material from sites being occupie novo,e d t neverthelesi s seem provido t s reasonabla e e frame-work against whice hth

archaeological material, more precisely the pottery, can be studied and, at least tentatively, dated. 13

NOTES 1 , 3 (1972), 1-55, especially 22-33. His work renders out of date most of what was written on this subject before 1972. 1460-1B RI formee 2 Th . r rea Anton]ino. wil w . d. no l Aug.[Pio secone . .; th . p.p.] s dco wil e b l similar. 3 RIB 975. Commodus appears, although not named (hence the survival of the stone), in RIB 1329, from Benwell presumabld an , y also (rather disguised 946B RI , n fro)i m . 4 RIB 1737 and 2026: in the latter case again, it is a dedication dated by a consulship - that of Lucius Verus in AD 167. Verulamiue th n o s A m5 inscription Roman,J Stud (1956)6 4 , 147. 6 Commodus, 6, 2; Pertinax, 3, 5-10. Proce Antiqc Se So Scot,7 (1986)6 11 noty , 191-3ema tha2209e B W .RI t , recording worI I y kb Augusta (not by a vexillation), could well belong to the time of Lollius Urbicus. 216B RI 6 (re-rea 8 Britannia,n di , 198516 , 332), 216 2172d 7an . 9 RIB 2174-7: in 2176 and 2177, the deity is named on the capital of the altar - a late second-or third- century feature. MANN: THE HISTOR THF YO E ANTONINE WAL REAPPRAISAA L- 7 13 | L

RomanJ 10 Stud, (1965)5 5 pd i XIX=Proan ,0 2231 o n , c Soc Antiq Scot, (1963-4)7 9 , 202-5. 11 R P Wright, Britannia, 5 (1974), 289-92, correcting J C Mann, Banner Jahrbucher, 170 (1970), 201. 12 Presumably it was the legion from York which bore the brunt of the attack, and attempted to stem the tide. 13 On the evidence here presented, it seems clear that the Wroxeter gutter deposit, and the Pudding wreckn Pa whicr ,fo acceptablo hn e dating evidenc eves eha r bee tentativelne b provided w no y y ,ma reigdatee th Severusf no o dt .

ACKNOWLEDGEMENT Margareo t Thanke du e tsar Roxa valuablr nfo e criticis thif mo s discussion.

REFERENCES Ann Ep 1888 to date L'annee epigraphique. Paris. Clarke J , 193 Romane 3Th fort Cadder.t a . Cohen, H 1880-92 Description historique des monnaies frappes sous I'empire romain, 2 ed. Paris. Evetts, L C 1948 The lettering of Roman-British inscribed stones', ArchaeolAeliana, 4 ser, 26 (1948), 153-71. ILS Dessau H , 1892-1916 Inscriptions Latinae selectae. Berlin. Mann, J C 1963 'The raising of new legions during the Principate', Hermes, 91 (1963), 483-9. Maxwell, G S 1975 'Casus Belli: native pressures and Roman policy', Scot Archaeol Forum, 1 (1975), 31-49. Millar, F 1964 A study of Cassius Dio. Oxford. Miller N 192S , Romane 8Th Kilpatrick.d Ol fort t a Glasgow. B CollingwoodRI Wright& G P 196R R ,, Romane 5Th inscriptions f Britain,o I: Inscriptionsn o stone. Oxford. Sutherland, C H V 1937 Coinage and currency in Roman Britain. Oxford.