Scottise Th h campaign Septimiuf so s Severus by Nicholas Reed

yeare 208-1th D sA n emperoe I 0th r Septimiu Caracalln sso Severus hi ad wersan e cam- paigning in Scotland. Agreement on even this point has only been reached in the last few years, virtualld an y everything else abou campaigne th t stils si l matte r disputefo r t archaeologicaBu . l evidence has been rapidly accruing in the last ten years, and is apparently now sufficient, when taken in conjuction with our other evidence, for us to erect a framework for the campaigns and drao t w some wider conclusions r evidenc campaigne Ou . th r efo four-folds si : literary, numis- matic, epigraphi archaeological.d can f these O 1 epigraphie th , s verci yanciene slightth d tan , historian infuriatingle sar y imprecise stary consideriny tma b e W . implicatione gth coinse th f o s. Perhaps the most intriguing are two coins which depict bridges and which are dated to 208, though tha f Caracallo t belon(Mattingly 9 20 ma a o gt y 1950, 269, 353, 390). Both bridgen sca comparee b d with those appearin coinn go previouf so s emperors. Tha Caracallaf to , which shows a bridge of boats with the legend TRAIECTVS underneath, is similar to a bridge on coins of Marcus Aurelius of 172, showing his crossing of the Danube (Mattingly 1940, 624; cf Oman 1931, 137), s alsi od depictean e Columth n do f Marcu no s (Caprino 1955, fig , 10)e othe9 s Th . r coinf o , Severus, show permanensa t bridge, whic identicahs i thao t l t issue Trajay db commemorato nt e his crossin Danube th f go e (Mattingly 1936, 178-9; Strack 1931, 127; Cichorius 1896 f IxxiiTa , , Bild xcix). It is normally suggested that these coins commemorate bridges over the Forth or Tay, or both therreasod o n an , s e i disput o nt e this. Where then were they built? Discussio focusses nha d roun e boat-bridgeth d n 193i d 1 an Oma, n argue s build wa that t i acrost e Forthth s . This suggestion has since been discounted (e.g. Whittaker 1969, 359), and there are certainly topo- graphical reason doubtinr sfo gcrossina Alloat ga , where Oman placeadmite H . positioe dit s th n would involve crossing a broad expanse of soft tidal mud. Moreover, no Severan - and hardly any Roman - finds have been made at Alloa, and the Forth is in any case not wide enough here for impressivn a e bridge. More recentl suggestioe yth s beenha n made thaboat-bridge th t s wa e situateJosept (S y ht CarpodTa a 1969,118 e th wn o ; Birley virtuall1971e b n , 258)ca ye certaiW . n there was some sort of bridge here, because air-photography has detected a bridgehead camp on the N bank of the river, facing the fortress on the south. A bridge of boats here is certainly possible, an t wouldi d follow that Severus' permanen Forte , sayth at t hn ,bridg- o Oman' s ewa s sitf eo Alloa. But there are just as strong objections to a permanent bridge as to a boat-bridge at Alloa. In any case, Stirling would be the natural crossing-point. However, we cannot place the permanent bridg t Stirlinea g either. This follows fro discovere mth 63-acra f yo e (and therefore Severan) marching-cam t Craigarnhallpa , whice th f ho Keird 'lie e for e Ol norte son ,t th d th jusa f ho o t t lowest fords on the Teith, and the importance of the discovery lies in the clue it affords to the Roman crossin rivere th marshee f sgo th Fortf o Teitd sd h an betwee han n them t Josep(S ' h 1973, 218). This camp clearly belongs to the double set of 63-acre camps N of the Forth, which REED: THE SCOTTISH CAMPAIGNS OF SEPTIMIUS SEVERUS | 93 mus datee b t d betwee 210d existencs nan .It A8 D20 thingso e tw implie f o : e eitheson r they forded Fort Teitd han h near this point ther o , y built small field-bridge croso st s these rivers. Whichever the thed ydi y woul t havdno e buil permanena t t bridge worth f commemoratioyo coina n no . If Craigarnhall belongs to 208 we could not explain the coin, but if it belongs to 209 or 210 they cannot have rebuil permanene th t t Antonine bridg t Stirlinea 208n g i ther o , y would have reused it in 209 or 210, and would not have crossed near Craigarnhall. Stirling, then, may also be ruled out. Nor could we imagine a permanent bridge east of Alloa in 208: Kincardine can be ruled out e samfoth r e reason s Alloaa s belod an , w tha a Romao dee t r to poin pfo r e rivefa n th ts i r permanent bridge. The hypothesis of a boat-bridge at Carpow, then, while unobjectionable in itself, obliges usupposo st permanenea t bridge somewhere acros Forte sth orden hi explaio rt othee nth r coin. But that leads int impassen oa , since there seem nowhere b o st e alon Forte gth h wher mighe ew t even hypothetically assum permanenea t bridgworth8 20 n sucf ei y o h commemoration muse W . t therefore go back one stage, and examine the alternative view, which is to suppose a permanent bridge across the Tay. Given the virtual certainty of a bridge at Carpow, this might be the site of a permanent one. The road which fronts the principia in the fortress, and the NE/SW axis of the bridgehead camp at St Madoes both point to almost exactly the same point of the river, just before the river starts to widen to embrace Mugdrum Island. The width of clear water just before the river widens is only 300 yards, leaving 400 yards of marsh and reeds on either side to be negotiated before firm ground is reached. Erosion of the S half of St Madoes camp shows that the distance would have been even less in Roman times. Thus, the siting of the permanent fortress indicates that when buil this bridgy swa an t e would have crosse t thida s western pointobvioue thid th s san ,i s position for a bridge in AD 208 also. The making of a roadway through 800 yards of boggy ground might well call forth comment in our sources: certainly, it ought to indicate some of the prepara- tions which we are told were involved. Herodian's pontes are not ordinary bridges over water, because the description he gives is clearly of the technique used to cross marshes by log- causeways (Herodia ; Ree5 n , iiid14 , 1975; Uslar 1971). Such causeways must have been uset da Carpow to approach the bridge across the river. There seems no reason to doubt the practicability o 300-yarfa d long permanent wooden bridg t Carpowea rivee neve.Th s ri r more than twenty feet deep at this point, and in fact, compared with the 1100-yard length of Trajan's bridge (and Severus' coin invites the comparison), the bridge itself was perhaps not ail that impressive. On the other hand, the causeways through the marshes would indeed have been an impressive tech- nativee nicath complete o lt sth feat d an , e structure should have been jus over-awins a t thas ga t built by Julius Caesar across the Rhine to reach the Sugambri. Here the natives were terrified in looking upon 'Rhenum suum sicponte quasi iugo captum' (Caesar BG, vi, 9; Florus i, 45,15). 'captureA d Tayhavy ma e' roused similar feelinge inhabitantth n i s f Scotlando sy an t A . rate, the existence of this bridge is a hypothesis we can test: excavation could well reveal traces either of the causeways buried among the reeds, or remains of the piles of the bridge in the river- bed. permanenA t bridgTaye th ,n theneo certainls ,i possibilityya thid san , hypothesis requires boat-bridga lowee th n re o Fort 208iden e hi aTh . that Caracalla could boat-bridg a hav d eha e built where the Forth Railway Bridge now stands, between North and South Queensferry, did indeed occu Omano dismissee rt h t variour bu , fo t di s reasons concluded an , d tha sucf ti hfeaa t beed ha n carried out, 'It would indeed have bee nTRAIECTUa S worth f commemoratioyo a n ni first brasa medallion r o s t merelno , ya modes t dupondius' s ironici t I . , therefore, thae h t continues, 'It seems hardly possible to doubt that since this coin was struck for Caracalla, there 94 | PROCEEDINGS OF THE SOCIETY, 1975-6 must have fathes beehi r nsimilaa rfo Severuse onlf rspecimeon i e t yon Bu . Caracalla'f no s issue survives, ther evers ei ychance reasoth y f fortuneo nwh e should have obliterate s father'dhi s corresponding coinage altogether.' As we have seen, the corresponding coin has survived - it merely cam ligho et t after Oman wrotthit sbu edrive- s hom face eth t tha cannoe tw t argue from absence th coia f eno (more particularly medalliona , t existno )d . thaTherdi vert y i t yema well have been first brasse r medalliono s s commemoratin TRAIECTUSe gth e morth t e valuablbu , ea coin, the more likely it was to be melted down and the metal reused later. Oman's other objections have rather greater force. 'The width of the Forth, though so much smaller than at any other point till one reaches Alloa going up stream, is 2,765 yards - a mile and two-thirds - ... and I think that such a width makes impossible any idea of a bridge of boats, when one considers the siz Romanf eo tide e vesselsswae sucth n th s f o enormou yn o d h a an , s breadt waterf o ht o t no , speak of the danger from storms.' Before dealing with these objections, it may be as well to draw briea p u f conspectu f ancienso t boat-bridge theid san r potentialities. bese Th t known boat-bridg antiquitf eo thas yi Xerxesf o t 'yokedo wh , Hellespone th ' o t t cross into Europe. His bridge was seven stades - that is, about 1400 yards - long (Herodotus viiboate ,th 36)d s supportin,an wergt i e anchored agains steadta y five-knot current partls wa yt .I to outdo him that Caligula had his own bridge built, the longest known in antiquity; built in stretchet i , AD39 d from Puteol Baulio it distanc,a ovef eo r three miles. Recently, what seeme b o st a mooring-block for the bridge was found near Bauli, and it is fully consonant in size (Paget 1971, 33, 37; Dio lix, 17; Suetonius Cal, 19). Now in deep water, it is a square block of Roman concrete 10 m high and 30 m square at the base, tapering to 20 m square at the top, which would originally have been level wit surfacee hth bridge Th . e doubl a itsel s fwa e lin shipf eo s joined togethern o , whicf o p hmouna to eartf dheapeo s hfashionewa d dan resemblo dt Appiae eth n Way, complete with resting-places and lodging-houses along it. It only lasted two days before Caligula grew t therbu e, tire it wer f do e more permanent boat-bridge Empiree t Zeugmth a n si e on : Syrin ai a (Pliny NH, v, 86), and another at Aries, which is apparently that depicted on a mosaic at Ostia. Temporary boat-bridges to attain a military objective were much more commonplace. The equivalen modere th f o t n military pontoon bridge, their thas advantagi d t thean s y eenablwa e larga e bod troopf yo croso st s much more quickl safeld yan y tha ferry y nthed b an y, also enable heavy equipment to be transported across water without having to be dismantled first. Dio says, 'River bridgee Romane sar th y db s wit greatese hth easef o t , sinc soldiere eth alwaye sar s practis- ing bridge-building, which is carried on like any other warlike exercise, on the Ister, Rhine and Euphrates'. He goes on to describe in detail one method of constructing such boat-bridges.2 Oman' sFortwidte e figurth th 2,76f f r ho o efo 5 yard grosa s si s overestimate actuae :th l width is about 1,800 yards, not that much wider than the Hellespont. He also fails to mention the equally important fact tha estuare tth neatls yi y divide thit da s smal e pointh y l tb rock y islanf do Inch Garvie, so two bridges of much shorter length would have been required. From North Incf Queensferro hp ti Garvi e th somo s e yt i yards 0 e55 , while from Inch Garvi nearese th o et t poin southe lanf th o t o dt , Long Craig Gate juss i , t under 1,000 yardyard0 40 ss- less thae nth lengt Xerxesf ho ' bridge itselfn I . , then lengte th , smalha t woulbu l probleme db . A more substantial difficulty would be the depth of the water. Thus, between North Queens- feet 0 ferr averaged s Incmucd a 18 an , an y e hs hb a Garvi sn abouca t ei t 150. Between Inch Garvi Lond ean g Craig Gate southere th , yard0 n 60 onls si y nine fee r les e o tdepthn si th t bu , other 400 yards can be as much as 200 feet. In one way, of course, this depth is an advantage : t meani s tha currene tth mucs ti h less strong tha mighe non t otherwise expect shipd ,an s coule db anchored even in 200 feet of water - ancient cables could be far longer than this. Thus, the bridge might have been constructe usuae th n ldi way sendiny b , boate gth soney d b dow an ,e non REED: THE SCOTTISH CAMPAIGNS OF SEPTIMIUS SEVERUS | 95 anchoring the mturnn i . Alternatively, Caligula's mooring-block indicates that, where water was exceptionally deep, a big mooring-block at the end could suffice for an immense boat-bridge. Across the Forth, such blocks would not have been necessary: there is plenty of room on the shor placinr efo winchee gth bollardr so s necessary Italyn i ; sheee th , r havcliffy ma es prevented endS f e thiI . thith s t techniqusa fasteninf eo bridge gth e were adopte Forthe th n do , they yma have followed the practice attested by the Anonymus Byzantinus, of building the bridge to the required length alongsid riverbanke eth thed an ,n towin t intgi correcs oit t position. They could then have let down one or two large anchors (or the baskets of rocks described by Dio) into the deep water, for extra stability. I would therefore suggest tha la t£ mile bridg boatseof , divided nea middlthe r Inceby h Garvie constructes wa , d unde guidance th r f Caracalleo a wher Forte eth h Railway Bridgw eno stands. Thi s lesi s s than three miles fro importane mth t Severan harbou stores-basd an r t a e Cramond, so that it would have been a simple matter for supplies to be brought by sea to Cramond and then carried across the bridge. This was the TRAIECTUS Caracalla celebrates in AD 208, showing both Severu himseld san f marching acros bridgee sth . Unfortunately, ther littls ei e hope rindinf o g archaeological evidenc t nowi r efo . However t weri f ,i e tru pointo e tw tha e s tth wher arme eth y crosse werA8 n di D 20 e Queens- ferr Carpowd yan mighe w , t hop fino et d some routtracee th t f tooei so k between those points, sucd an h traces coul suppord crossin a case d ad r th e fo o t t Queensferry ga route Th . e taken i facn i s t beeha 8 Anmajoa D 20 r sourc puzzlementf eo layo Di .s stres road-buildinn so g generally: 'Severus encountered numerous difficulties in cutting down the forests, levelling the heights, filling up the swamps and bridging rivers'. As Richmond says, 'Much preparation of roads was undertaken in marshy country, a description suitable to Angus and the Mearns or to Fife, but hardllong-establishee th o yt d land-route between Fort Tayd han ' (1955 hav,e 58)w ef I .correctl y crossing-pointlocateo tw e dth s concerned mose th , t natural assumptio roanw n wouldne i a e db Fife, running in a reasonably direct line between Queensferry and Carpow. And indeed, if Severus was already plannin campaiggo t Scotlandn i woult ,i sensible db builo et dpermanena t roadway at least as far as Carpow. There are several finds which support this hypothetical route. First is the recent discovery 63-acra f o e (and thus Severan) marching-cam t Auchtermuchtypa , fiv f eCarpo o mile E wSS s t Josep(S h 1965, 82). Second, ther repora s 'rectangulaea i f o t rectilinead an r r inclosure, within which was found a silver coin of Pertinax'; this was seen before 1812 and (if we can trust the report) must surely have been another Severan marching-camp, situated about two miles east of Lochore, near Auchterderran (Macdonald 1917, 169; Crawford 1949, 146). Lastly, a hoard of more thacoin0 n60 s ranging from Galb Severuo at discoveres swa 185n di t Portmoa1a e th n ko east side of Loch Leven (Macdonald 1918, 264). Most naturally interpreted as hidden by a soldier Severusn o ' campaigns, these coin line r neaf march.sth eo o shoule o n t r th lino e f e 3eo Th db Severan roa Af do D 208, then wely froe li lma , m North Queensferr Auchterderrana yvi , then east of Loch Leven, and via Auchtermuchty to Carpow. If the road was here we could also explain why it has not so far been detected. Many stretches would lie through marshy ground, so that, instead of the normal stone-paved or gravelled agger, one would only expect to find the log- causewa excavationy yb : eithe grounn ri d whic remaines hha d marshy detectiny b r o , telle gth - tale chang soin ei l colour which would attest decayed wood. beginnine th t campaignse A th f go , Herodian stresse preparatione sth warr sfo , including 4 the building of log-roads. He continues, 'Once the preparations for war seemed to be satisfactorily completed, Severus summoned Geta, and left him to ... carry on the civil administration of the empire. Antoninus he took with him, and advanced against the barbarians.' The impression 96 | PROCEEDINGS OF THE SOCIETY, 1975-6 substantiaa givef o s ni l time spen sucn i t h preparations before advancn a e agains enemye th t . These preparations were apparently those which we have deduced were taking place in Fife, which suggest t hostileno s Fif8 s .thaewa 20 n i t At this point we can turn to the dating of Severus' campaigns, which are generally accepted o havt e taken plac 210d ean .betwee 8 Sinc r 20 sourceD eou nA s suggest there were onlo ytw campaigns, the second has been dated to 210, and the first to '208-9'. Now, recent writers almost all agree tha Maeatae tth e suppos e werw f Fifei n e i concerne s o eS . tha wa yea e 8 tth r20 d only with marca h through their peaceful territory preparationd an , advance th r sfo e then northca n e w , neatl rightino ytw slo e gtth campaign s210mosd e int yearan e .Th o9 tth conveniens20 t assumption would thus be that the Maeatae did not resist in 208. Perhaps a bargain was struck, whereby the Romans agreed not to take revenge on them for their hostility in some previous year, on condition tha Maeatae th t e allowed the mbuilo t d their road bridged san s without continual sabotage. considew no campaigny We rth e ma 209-10f so whicr fo ,mosr hou t important evidencs ei e marching-campsth majoo . Thertw e r ear set f campo s s which havbeew eno n fairly securely dated to the Severan period (St Joseph 1969, 105f; 1973, 216f). One set, 63 acres in size, has a few example maie Forthe th th nt f o seriebu , sS roughlso runtw n si y parallel lines northe th , - western as far as Keithock, the other up to Kinnell, but almost certainly aiming for the Montrose Basin secone Forthe Th . th acre 5 t thef do 16 ,bu serie s nS f acre 0 dropo continue d s s13 i s an o st s Kais a r rt leas fa a House s p a tu mileMontrosef 2 o 1 , sN shoule W . d most naturally expece th t more northerly (130-acre) set to belong to the final campaign of 210, and the other set to 209; archaeolog s offereyha d suppor r thi showiny fo tb s earlier.e b e 63-acr go th t t Josept S 5 ese h suggests that the 63-acre camps, in two lines, mark a 'round trip' by one force. But 63 acres is almost exactly half the size of 130 acres, and it is most unlikely that half the army was not being used during the first advance into hostile territory. It is much more likely that in AD 209 the army was divided into twoled d doubto an n , Severuy b , Caracallad san . We may be able to go further, if we consider the question of numbers. It is normally assumed that Severus had three legions for these campaigns - the three stationed in Britain - and that when brought together 'with a substantial number of auxiliaries or other troops' they occupied the 130-acre camp Richmonf s(c d 1967 SmitE shows ,R 62)t hha Bu .n that Severus also too t leaska t part of the praetorian guard with him (1972, 488, n 41). Its strength was now 10,000, which is virtually the equivalent of two legions. We can also be almost certain that Severus brought the new legion II Parthica with him as well.6 Thus, Severus probably had the equivalent of six, not three, legions campaignin Scotlandn gi . Bearing this in mind, it is interesting to consider the size of the marching-camps. Normally it is assumed that a 20-acre camp would have accommodated one legion. Thus Frere, building on an ide Richmonf ao d (1934, 50f), 20-acrrefere th o st eCros y cam Re holdins st a p a legione gon , an61-acre dth e camp near Neath sayse h , , 'would theoretically hold three legions' (1974, 121). This is also the most plausible interpretation of the Severan 63-acre camps - but it follows that 130 acres should represent a force equivalent to six legions, which fits nicely with the force detailed above. Now t mighi , e objecteb t e dSevera th tha y b t ncommoa s periowa t di n practicr efo detachment legionf s o take e b campaignn o snt o , rather than whole ones reasoe :th n presumably being to avoid any frontier being excessively weakened if whole legions were taken from it. But this coul t appldno Britainn yi , where temporary remova legione th f o l s from Caerleon, Chester and York could not conceivably endanger the frontier.? Let us return to the suggestion that the army was divided in two, and led by the two Augusti advanceo wh , paralleln di , planting 63-acre camp thes sa y went. Severus would presumr ably take with him his Praetorian Guard and II Parthica, while Caracalla would lead the three REED: THE SCOTTISH CAMPAIGNS OF SEPTIMIUS SEVERUS | 97 legions stationed in Britain. He was thus in charge of legions who were used to fighting in Britain, and he was also quite fit, unlike his father, who suffered from gout. Thus Caracalla will have taken potentialle th y more hostile western route, following Agricola' s Keithocka r routd fa ol s s ea . Severus will have taken the eastern route, crossing the bridge at Carpow. And by the end of the year, matter gond sha e sufficiently wel r thefo l mclaio t title mth e BRITANNICVS. We turn now to the function of the navy in these campaigns. Its activities are attested in various ways. The most evident is the establishment of stores-bases at South Shields, Cramond and Carpow. All these were clearly turned into such depots because of their proximity to good harbours. Coins of Neptune and Oceanus dated to AD 209 must also attest fleet activity. That other fleets were involved as well is shown by an inscription set up to a Prefect of the Fleets of Britain, Germany, Moesia and Pannonia, all apparently combined for the expedition.8 Presumably they were taking part in a joint campaign with the army, something of which we have a splendid descriptio 'Agricola'e chapteth n ni f o 5 2 rthin I . s year (82) Agricola use flees dhi exploro t t e harbours 'quia infesta hostili exercitu itinera'. Severus, wise to the same considerations, followed safa e route havthrouge w e0 observeh21 Fifn i acre0 208t n droe i 13 dth s Bu . po t fro5 m16 once the Forth is crossed. We could link this with an inscription from Corbridge (RIB 1143) mentioning someon chargn e granarieei th f eo s during Severus' expeditions. This must imply that Corbridge held supplies, which were presumabl sen e Derp b u t o t ye StreeDerd an et- Street was used in 210, as the 165-acre camps show. Thus, the extra 35 acres are likely to designate an enormous baggage train, and indicate that much of the grain went by ship north from the Forth.9 This is also suggested by the many coins of ANNONA in 209-10. Agricola succeeded in bringing the natives to battle at Mons Graupius, and the resultant massacre must have crippled the yearsr mfo . Severus presumably intende dsimilaa r result. What he did not anticipate was the natives' unwillingness to provide him with a second Mons Graupius. By this time thelearned yha lessoe dth n take hearo nt modery b t n guerillas: that guerilla tactics are the only, but highly effective, defence of primitive peoples against forces vastly superior in equipment and organisation. As Dio says, 'Severus fought no battle and beheld no enemy in battle array alse h ' o ; (Ixxvi make2) , 13 ,t clea si r that Roman army losses were alarmingly high. Nevertheless when Severus finally returned from the north, 'he had forced the Britons to come to terms condition o , n that they abando na larg e par f theio t r territory' reasoe Th r .thei nfo r surrende concessiond an r obscures si ; possibly they surrendered becaus f lace o suppliesf ko t A . any rate, Severus regarded this surrender as sufficient to claim a Victoria, and the SHA, referring returns hi o t , shows tha regardee h t Scottise dth h proble finallms a y settled: 'Post murum apud Luguvallum visu mBritannian i proximad a m cu , m mansionem redire solun no t m victod se r etiam in aeternum pace fundata . . .' (Vita Severi 22, 4). However, whatever the details of this settlement, it must have been invalidated by later events, since it was also in 210 that the Maeatae, followed by the Caledonians, revolted. Dio tells us that this revolt of the Maeatae occurred after Severus' return from the north, so probably in the second half of 210. The reason for their revolt has often been put down to the severity of the terms imposed on them earlier. That is possible, though the terms mentioned above were apparently impose e dCaledonians th onl n o y shoul e w o s d, hav o imagint e e harsher terms impose e Maeatath n do s wellea spitn i , f theieo r quiescence fro e reasomth y 208t nma Bu . alternativel thae yb t Severus' very failur inflico et militara t y Caledoniane defeath n o t s persuaded e Maeatath adopo et t similar guerilla tactics. Severus retaliate sendiny db arme gth carro yt y polica t wholesalf ou yo e extermination. Given tha settlemene tth t with Caledonia seeme havo dt e settle e Scottisdth h problem n welfinallyca l e imaginw , e exasperatioth e n which must have prompte thio t s m policydhi . | PROCEEDING 8 9 THF SO E SOCIETY, 1975-6 Dio continues, 'When this had been done, and the Caledonians had joined the revolt of the Maeatae, he started preparations to make war on them in person. But while he was thus engaged, his sickness fourtcarriee th ofFebruarf n m hfo o d hi y (211). othen I ' r words Caledoniane ,th s only joined the revolt once Severus had embarked on his brutal policy towards the Maeatae. There can be little doubt that that was the reason why they revolted as well: they were not going to stand idl whily yb e their fellow countrymen were being systematically massacred alsn ca o e W . deduce tha Maeatarevole e th th t f o t e starte yeard e Caledonian e latth n Th .ei s must have joined in fairly soon after the massacre started, but Dio implies that by that time it was too late in the year for Severus to deal with them also. If we also take a rescript of the Emperors datelined York, 21y evidencs 0a Ma h 5t e that they were stil Yorn li t thaka t time (Cod , 1 Just; 31 Birle , 3 , y 1971, 267), we may reasonably conjecture that the major campaign, now led by Caracalla alone (as Herodian III, 15, 1 implies), started in June, was over by September at latest, and the Maeatae revolte Septemben di r Octobero r . Returnin evente th o gt s precedin occupatioe gth Fiff n o 208 n earchaeologicai e ,th l evidence consists in the series of 63-acre camps S of the Forth, of which only three have been discovered so far: Kirkpatrick, seven miles north of Burgh-by-sands on Hadrian's Wall, and probable example Castlecrait sa Eskband gan k nea Forthe rth . Their size should indicat forcea e equivalent to three legions advancing from the Wall, and if assigned to AD 208 or 209 we should have to suppose that a more easterly set, to contain the other half of Severus' army, has not yet been detected. Thi possibles si , though unlikely rulee b y darme outth ma . thi n ys i A0 a , sD21 yea s ri accounte165-acre th y b r e dfo camps . This campaig territore th Selgovaee n nth i f yo , theny ma , rather belong to 207, and may have been undertaken by the three British legions, not by the forces brought by Severus in 208. If there were no further evidence we should most naturally assign this campaign to the governor Senecio. But there is other evidence. It is still sometimes assumed that wit8 coin20 h f PROso F AUGG indicat departure eth Emperore th f eo s from Rom than ei t year (Jarrett 1970, 199). This will not do: the coins are repeated in 209, when they can only indicate departure from their British base to the Scottish front, and the coins of 208 may indicate the same.1 ° t coinBu s clearly indicate that Caracall somn i 7 activs e 20 province awa n ei . Ther severae ear l Caracallan showin 7 issue20 f so g Mars, others with Virtus, i.emilitars hi . y prowess mosd an , t significant, some showin witm ghi h captive river-goa d an s d (Mattingly 1950, clx). These just might refe somo t r e mysteriou badly-documented san d Continentevente th n so . The more yar e commonly thought to refer to Britain (Richmond 1955, 57f). To add to them there is another coin, or medallion, on which stress can be laid (Cohen 1880, vol. IV, 3; Gnecchi 1912, II, 73). Dated to inscribed an 7 20 d ADVENT(US) AVG(usti) GALL(iae) t showi , s Severus alon horsebacn eo k arriving walle citya th f t so a . Ther reasoo n s ei disput o nt genuineness,s eit singulae th d an r AVd Gan 11 depiction of Severus alone strongly suggest that he was not accompanied by Caracalla on his arrival Gauln i , whatever impressio historiane nth contraryse givth o et . Give prominence nth e whic hBritise Caracallth n o h s coinagaha 209-1n ei s inconceivabli t 0i e tha arrivee h t d after Severus, and much the simplest view is that he was already in Britain in 207. In that case, it was Britise th d hprobablle legiono wh e syh throug territore h th Selgovae th f yo 207n ei captivee ;th s cois hi n n wil o froe Solwab le mth thae b y rivee ty Firthtribeth rma d .an , f CaracallI alreads awa Britain yi 207n Severund i an , they Gauln sni ma e e turth w , o nt rescripts from 208. The first two, of the 12th and 18th of February, were issued by Caracalla alone. That of the 10th of March, and four of the other five rescripts of this year, were issued by the joint Augusti. naturae Th l conclusio thas ni t Severu Britain i s n joineso n s betweedhi n 18th 12 February and 10th March. We may now offer an explanation for a puzzling coin of 208; this is a rare type depicting Severus seate magistrata s da curula n eo e chair, being crowne Victory db y REED: THE SCOTTISH CAMPAIGNS OF SEPTIMIUS SEVERU9 9 | S and resting his elbow on a kneeling figure, and the legend is VICTORIAE AUGG (Mattingly 1950, clii). Since both Emperors were in Britain in 208, this is not likely to refer to Africa. Rather, it might commemorat resumptioe eth Romaf no n administrative power over Fifenon-militare Th . y character and attitude of the figure of Severus suit the diplomatic agreement deduced above, and kneeline th g figur than ei t case woul membea e Maeataee db th f o r .

Summary campaignsthe of 207 Caracalla come o Britaint s . Conducts campaign Britis 3 wit e hth h legions from Hadrian' Forthe th s o Walt , p througu l h territor f Selgovaeyo . Severus arrive Gauln si . 208 Uncontested advanc f boteo h Augusti through Fife (Maeatae), establishing bridges at Queensferry and Carpow with a road between. 9 Caracall20 a advances wit British3 h legions against Caledonians ,Agricola d closol o et n route up to Keithock. Severus with Praetorian Guard and II Parthica advances via Carpow to Kinnell (and probably Montrose); both leave 63-acre camps. Fleet, drawn fro mnavies4 , brings supplies north from South Shields. 210 Advanc l forcesal f eo , with supplies, from Corbridg Inveresko et , planting 165-acre camps. Army, under Caracalla, continues nea Agricolad rol n Kairouts a r t rlease a fa House s ta , planting 130-acre camps. Fleet sail frosN m Cramond, taking supplies. Extensive casualtieso n ; battleg bi ; concessions mad Caledoniansy eb . Fleet circumnavigates Britai Victord nan y claimed. Late in year, Maeatae revolt. Severus' attempted genocide provokes Caledonians to revolt as well.

The Purpose of Severus' Campaigns In the case of Agricola, no historian tells us his purpose in invading Scotland, but no one doubt intend s di tha e occupo dt h t ScotlandE yS t more evidencer no f i , Ou . , normally accepted conclusives a buildine th permanen e s ,i th f go t fortress, intende 20te th hr Legiondfo Inchtuthilt ,a . If a force was going to be stationed here permanently, they must have intended to keep a hold on lane th d around than i t t cas Bu .establishmene eth permanene th f o t t for t Carpoa t w indicates similaa r intentio Severusn no ' par holo t surroundine dth gdemolitios it land d an , n soon after no more negates this intention than the similar demolition of Inchtuthil. One argument has been given against this view: 'Ievidens ti t from Severus' neglec garrisoo t lowlande nth s s thawa e th not intending a permanent occupation of Scotland' (Frere 1974, 201). But if Severus felt confident of victory in the far north he may have thought garrisons unnecessary in the lowlands. speciao s Wha s wa lt about Fife, sufficien warrano t t t Roman occupation? Birley (1936) dismissed the possibility that coal there should have been sufficiently important to make Fife valuable shoulr ;no expece d w mor y tan esal itsele y importantb tb o ft : there were plent saltf yo - pans elsewhere in Britain. On the other hand, Fife cannot have been so marshy that corn-growing farmind an g were impossible mighe w d t linan , k thi witp versu e hth y high frequenc f coino y s depicting ANNON especialld an A9 An i D 20 y 210. Thes simply referencea ma e y importancb e th o et e of the fleet in bringing supplies north. But we might also see in them a reference to the fact (if it was one) that the Maeatae were now paying tribute in corn, like any other state in Britain or the Empire. Indeed, tribes through whose territory the Romans were marching normally had to contribute more to the annona than in time of peace. Also, their subordinate or neutral position towards Rome must have produced a useful buffer state against the Caledonians, and, when they were peaceful, the risk of attacks on the E half of the Antonine Wall must have been greatly reduced. Any one of these factors in isolation was perhaps insufficient to make their subjection seem worth while. Nevertheless, the combination of buffer-state, and its resources in coal, salt 100 I PROCEEDINGS OF THE SOCIETY, 1975-6 especialld an y grain, would probably have appeare extrae th r d troopfo enoug y pa s o hinstallet d . Wheit t occupiedn i nno , their friendly disposition would itsel usefue fb maintaininn i l g security alon Antonine gth e Wall. Severae Th n plan, then, involve occupation da t intene leasna h f Fifed o t dDi . anything further? According to Dio (Ixxvi, 13, 1), he intended to conquer the whole of the island. It has sometimes been suggested that Agricola' occupo st plas garrisod nwa yan e whole th nth f eo Scottish Highlands. Analogie e drawar s n wit e mountainouth h s area f Waleo s d Spainan s , which were both occupied. However vitao , thertw l e differenceear s between these countried san Scotland. Both the Welsh and Spanish mountains contained goldmines, and such mines ranked ver ylis e desirabl f higth to n ho e attribute prospectivf so e provinces. They might offe retura r n no the investment of troops in their locality, and provide a considerable profit as well. Mineral resources were alway importann a s t consideratio Roman ni n aggrandisement s showna , r fo , example, by the extraction of lead from the Mendips as early as AD 49. But Scotland had no such resources, as far as they knew. Secondly, unlike Scotland, neither Wales nor NW Spain could be easily cut off from the pacified areas by a limes - indeed, a limes cutting off Wales would have to run for more than 100 miles. Scotland was far greater than Wales in area, but it could be, and for some tim frof e province of wasm th t limesa cu , y eb only forty mile concludy s longma e W .e that the government in Rome is very unlikely seriously to have considered occupation and garrisoning of the Highlands.

The Aftermath We turn e resulte campaignslastlth th f o t yo s . Whatever concession havy ma se been extracted from either Scottish tribe by Severus in AD 210, the terms of the agreements cannot have lasted, since in 211 both tribes were in revolt, and Severus was planning to lead a further expe- dition, when he died in February. But actually the situation was worse than that: at no point previously during the campaigns had both tribes been actively hostile. From 208 to the summer of 210 the Maeatae stayed quiescent; the Caledonians seemed to have been conquered before the Maeatae revolted later in 210. Such was the Emperor's exasperation at this setback when every- thing seemed settled tha lashegenocidee n i th t dou , thus bringin Caledoniane gth s into revols a t well. Naturally, afte l thial rt woul sintolerabli n a e db e los facf sSeverur o efo withdrawo st , even though three years of campaigning had now in practice merely worsened the situation. However, it is hardly surprising that once his father was dead, Caracalla gave up all the grandiose schemes of conquest decided an , withdrao dt l majowal r force Hadrian'o st s Wall, however costly that line was to maintain. His new frontier policy involved patrolling N of the Wall and apparently a check on meetings of the Scottish tribes. But in the end, Severus' attempt to solve the Scottish forgottens problewa l northernmose al mr Th . fo onc d ean t frontie Empirf o r e staye t Hadrian'da s Romae th f Walo nd loccupation untien e lth .

ACKNOWLEDGMENT I am grateful to several people, in particular Professor S S Frere, for criticisms of an earlier versio thif no s article.

NOTES principae Th 1 l reference foune b y Birlen di ma s y 1971, chap xvi; references here wil confinee b l d largely to those not already in Birley. The other major recent discussion is in Frere 1974, 194f. REED: THE SCOTTISH CAMPAIGNS OF SEPTIMIUSSEVERUS | 101 2 Dio Ixxi, from Suidas s.v. £euy|M. Another account, perhaps from firsthand experience, in Arrian Anonymue Th Anab. . 7 , v s Byzantinus Trepl oTpa-n^/i/oj give9 1 . a generas?c l accoun f suco t h bridges ; cf. also Tac. Hist, ii, 34. For discussion of their representation on Trajan's Column cf Richmond 1935, 5-6cannoI t .Bu t credi 'pontoons hi t s lying between each pai boatsf ro - thes' e are surely small cabins on the boats, visible on single boats elsewhere on the Column. 3 For this interpretation, see the evidence of Robertson 1971, 134 n 10, which militates against the traditional view (repeate r texthe )n d i tha t these hoards were hidde notn nativesy nb ca e e thaOn . t two of the other three such hoards were also found near the known line of march. 4 However, parts woul normaf o e db l construction sugges.I t tha well-preservedta , stretch, later used in medieval times, survives running from Easter Lumbennie (NG 238163O RN ) past Macduff's Cross straigh Carpowo t . 5 Note also that the western line of 63-acre camps (close to the line of the 130-acre ones) is more closely spaced, as is natural if it indicates the first campaign through hostile territory. 6 Smith 1972. Unknow himo nt , ther direcs ei t archaeological evidenc f thieo s legion's presencea : standard depicting a centaur found at Spennithorne in Yorkshire (cf Taylor 1944, 24-5). Its findspot ditce marchina th f hn suggesto i s gwa t campsi . 7 Only one legionary detachment from abroad is specifically attested for this campaign: CIL xiii, 3496. There may have been a few others, or there may not. The only other evidence for the size of the expeditionary force is Dio's statement (Ixxvi, 13, 2) that Severus lost a full 50,000 men in Britain reliabilite Th . f thiyo judge se b tota n reflectinn do ca l g tha t equivalenwoule i t th e db f o t more than eight legions. For similar wild overestimates of army or casualty totals in Dio, cf Townend 1964, 479f. 8 CIL vi, 1643. The wording seems to imply - and I see no reason to doubt - that virtually the entire fleet thesf so e four countries were being used difficulto N . obtaininn yi g enoug hboate shipth r -sfo bridge, either. 9 Another possibilit thas yi representt ti bodsa auxiliarief yo wero sewh sen guaro t Antonine dth e Wall (cf St Joseph 1973, 231). This is less likely, in particular because they would surely have been lef guaro t t Wale dth l durin previoue gth s winter thud an ,s woul t havdno e forme dforce parth ef to remarkabls i springe t i th t n i senBu .t auxiliare ou te on tha t tno y detachmen specificallys ti attested as taking part in these campaigns - though several have been suggested as possibilities. compary ma e ADVENTVe eOn th 10 S AVGVSTI coi Severuf no 210f so , whic onln hca y returs refehi o t rn base th t Yor ea o t k (wher died)e eh : Mattingly 1950, clxxiv. onle Th y1 1reaso n ever give r suspectinnfo thas i f Cohent go i t saio dwh , that Severus' journeo yt Britain through Gaul could only have been in AD 209 or 210. On the contrary, we now know he t latesta welthud 8 y waan ,20 ls ma Britaihavn si D A ey beenb Gaun i 207n i l . Gnecchi merely followed Cohen in his doubts, and Birley, Gnecchi; neither had independent reasons for doubt. 12 Cod Just, 8, 25, 2 ; 2, 1 1, 9; 3, 28 r othe, 4.Fo r rescripts from 208-10 Birlee ,se y 1971, 267 rescriptl .Al s wer0 21 e joind issuee an th fro t9 y dAugustib m 20 , excep froe Januar3 mton 1 y 209a s . wa Tha e ton judgement delivere Severuy db persona,n si t Caracallbu wely alma have been present.

REFERENCES Becatti, G 1961 Scavi di Ostia, vol IV. Rome. Birley R 197 A , 1 Septimius Severus. London. Birley, E 1936 'Marcus Cocceius Firmus: an epigraphic study', Proc Soc Antiq Scot, 70 (1935-6), 363-77. Calza, G 1914 'Ostia: Piazzale delle Corporazioni', Notizie degli Scavi di Antichita, 11 (1914), 284-91. Caprino C ,195 Colonnaa 5L Marcoi d Aurelio. Rome. Cichorius C ,Reliefs189e Traianssaule.r Di 6de Berlin. Cohen, H 1880-92 Medailles Imperiales, vol IV. Leipzig. Crawford S 194 G 9O , Roman Scotland. Cambridge. Frere S 197 S , 4 Britannia, Historya of Roman Britain,. Londoned d 2n . Gillam, J and Mann, J 1970 'The northern British frontier from Antoninus Pius to Caracalla', Archaeol (1970)Aeliana,8 4 , 4 r , 1-44se . Gnecchi F ,191 2/ Medaglioni Romani, . MilanII l vo . Grenier, A 1934 Manuel d'Archeologie Gallo-Romaine, vol II. Paris. 102 | PROCEEDINGS OF THE SOCIETY, 1975-6 Jarrett, M and Mann, J 1970 'Britain from Agricola to Gallienus', Banner Jahrbucher, 170 (1970), 178-210. Macdonald G ,191 7 'Genera s "Militarhi l d Williaan yy AntiquitiemRo Romane th f o s Nortn i s h Britain ", Archaeologia,' (1917)8 6 , 161-220. Macdonald G ,191 8 'Roman coins foun Scotland'n di , Proc Antiqc So Scot, (1917-18)2 5 , 203-76. Mann, J and Jarrett, M 1967 'The division of Britain', / Roman Stud, 57 (1967), 61^1. Mattingly, H 1936, 1940, 1950 Coins of the Roman Empire in the British Museum, vols III, IV, V. London. Oman, C 1931 'The first Forth Bridge', Numis Chron, ser 5, 11 (1931), 137-50. Paget, R F 1971 'From Baiae to Misenum', Vergilius, 17 (1971), 22-38. Reed J 197N , 5 'DrusuClassie th d ssan Britannica', Historia, (1975)4 2 , 315-23. Richmond Mclntyred an A I , J , 1934 'The Roman camp t Reycrosa s Crackenthorpe'd an s , Trans Cumberland Westmorland Antiq Archaeol Soc, (1934)4 3 , 50-61. Richmond A 193 I , 5 'Trajan's arm Trajan'n yo s Column', Proc Brit School Rome, (1935)3 1 , 1-40. Richmond, I A 1955 Roman Britain. London. Richmond, I A and Ogilvie, R M 1967 De Vita Agricolae. Oxford. Robertson A ,197 1 'Roman coins foun Scotlann i d d 1961-70', Proc Antiqc So Scot, (1970-1)3 10 , 113-68. St Joseph KJ , 1965 'Air reconnaissanc Britain'n ei , / Roman Stud, (1965)5 5 , 74-89. St Joseph, J K 1969 'Air reconnaissance in Britain 1965-68', J Roman Stud, 59 (1969), 104-28. St Joseph, J K 1973 'Air reconnaissance in Britain 1969-72', / Roman Stud, 63, (1973), 214-46. Smith, R E 1972 'The army reforms of Septimius Severus', Historia, 21 (1972), 481-500. Strack, PL 1931 Untersuchungen zur Romischer Reichspr. des HJahrhunderts, vol I. Stuttgart. Taylor, M V 1944 'The Sidmouth bronze: legionary standard or tripod?', Antiq J, 24 (1944), 22-6. Townend, G 1964 'Some rhetorical battle-pictures in Dio', Hermes, 92 (1964), 467-81. Uslar, Rvon 197 deu 'Z n1 ponies longi\ Gymnasium, (1971)8 7 , 213-24. Whittaker R 196 C , 9 Herodian, Loeb. LondonI l vo , .