Proc Soc Antiq Scot, 110, 1978-80, 230-85

Excavations on the fort of Rough Castle, Stirlingshire, 1957-61 lain Maclvor*, M Clare Thomas and David J Breeze* with contribution BarberW J y Charlesworthsb ,D DickinsonM B , , HartleyBR Hartley F HodgsonI K ,W G , SlaterE , WilP d J ,d an D F Williams

SUMMARY The N and W ramparts of the fort and part of the praetentura were examined in 1957-61. The 'additional margins' of the W rampart were shown to be an ascensus on the inside by the W gate and an accumulation of three successive road surfaces interleaved with rampart tumble on the outside. The earliest structures in the praetentura appeared to be temporary and were followed by two timber buildings, presumably barrack-blocks l evidencAl . putativr efo e Antonin I buildingeI beed sha n removee th y db 1903 excavator t evidencbu s e survive suggeso dt presence th t f lean-teo o buildings agains e insidth t e faces of the W and N ramparts. The occupation of the site continued into the early 160s.

INTRODUCTION The 'entireness and magnificence' of Rough Castle (NS 843798) was remarked by Alexander Gordon (1726, 59)fore notices i tTh . antiquariae mosn di th f o t n writing 18te 19td th han f h so centuries which deal with Roman r JohSi : n Sibbald (1707, 30), John Horsley (1732, 173), Willia (1793y m) WilliaRo 61 , m Nimmo, (1777 , Rober8) , t Stuart (1852, 334). Gordon, Horsley providy Ro , e plans; Roy' s usuaa s beste th l , Horsley' e worst th srampare Th .s wa t sectione e Glasgoth y db w Archaeological Societ n 1890/9i y 3 (GAS 1893). Forannexd an t e were excavate Societe th y db Antiquarief y o f Scotlanso 190n di 3 (Buchanan 1905, 442-99)e th , last of the Romano-Scottish operations which the Society began in 1895: Birrens, Ardoch, Birrenswark, e Gasth , k Ridge, Inchtuthil, Lyne, r GeorgSi . e Macdonald carried out a limited excavation in 1932 (Macdonald 1933, 243/96). The results of both excava- tions were used by Sir George in his account of Rough Castle in the 1934 edition of the Roman Wall in Scotland (Macdonald 1934, 217-38). Recent accounts are those by Professor Anne Robertson (1979, 53-6) and the entry in the Stirlingshire Inventory of the Royal Commission on the Ancient and Historical Monuments of Scotland (1963); modifications to them as a result excavatioe ofth surved nan f 1957-6yo suggestee 1ar d belo 233-42)p w(p . The earthworks of Rough Castle are still the best-preserved among the Antonine Wall forts and they lie on the longest surviving stretch of good rampart and ditch, some 2850 m (3,100 yd) from Lime Roa857798S (N d ) throug Tentfiele hth d Plantatio Bonnysido t n e HousS e(N 834799) rampare th fort e d ditcd Th .an ,an t h westward froBonnysido t mt i e House (included

Inspectorat* Ancienf eo t Monuments, Scottish Development Department Athol7 1 , l Crescent, Edinburgh Lilia

O

H I

ta >

aI o 50 . 100 FIG 1 Rough Castle fort and annexe

K) OJ j PROCEEDING 2 23 SOCIETYE TH F SO , 1978-80 within a generous surrounding area), made up one of those lengths of the Wall given in guardian- ship to the Minister of Works by the National Trust for Scotland in 1953. The adequate display of these lengths was a long-term task; as with most field monuments, the greatest end of preser- vatio assures nwa d simpl guardianshipy yb forte Th ., however, needed separate attentiot pu o nt t evei n provisionall ordern yi excavatione Th . s carrie 190 n i beed t 3ha n dou immediately back- filled over most of the fort interior, but were left open over the defences, the pits or lilia to the N, e commandinth g officer's house e bathhousinsidth forte d eth an , e insid e annexeeth . Hala f century late scarree rth d monumen pars twa covere densy db e bracken, vigoroupara y tb s growth birchd an k .ooa f decides 195n I wa t 6i d that, while clearin brackee gth thinnind nan treee gth s ove forte rth , excavationd ol e mosth f o t s shoul infillede db . Amon mose gth t unsightl confusind yan e th f go latter, the W rampart section as cut back for photography in 1903 offered a rare opportunity for re-excavation. A trial excavation in 1957 redefined this rampart section and opened a small praetenturae are th f ther i n ai e ese wer o tracey t ean f timbeso r buildings firse Th t. operation was profitable secone th , d promising 1958-6n :i 1 further area spraetenturae oth f were stripped.

Inner face of fort rampart

FIG 2 Rough Castle fort: location of excavated areas 1957-61 MACIVOR, THOMAS, BREEZE: ANTONINE WALL FOR ROUGF TO H 3 CASTL23 | E re-surveyeds fore wa Th t . Durin afte d excavatioe gan th r restoratioe nth defencee th f nd o an s the interior of the fort was carried on, and the fort and the area round it cleared of bracken (a surprisingly difficult task) 1964-n I . attempn 6a mades wa t , with only limited success cleao t , r some of the silt from the Wall ditch where it forms the northern defence of the fort. The results of the excavation are modest. A few certainties emerge concerning the defences; a few points may be tentatively advanced concerning the interior. lain Maclvor supervised the excavation and wrote the excavation report. Clare Thomas wrote the finds reports other than those separately acknowledged, and drew the finds. David Breeze wrote the discussion and guided the whole to publication.

THE EXCAVATIONS Rampart Background 190n I 3 seven sections wer acrost ecu defencese sth acros e ramparN on ;e sAntoninth d tan e Wall ditch; four acros rampare sth doubld an t e othe e ditcth f rho fort e acros sideo th tw f ; so s the rampart only. All of the sections were probably begun as trenches, like those of the Society in 1890-3. Three of them were then enlarged for photography by cutting back one side trence o th gentlfa o ht e slopef whico p ,hphotograpa to fro e m th remainine th f ho g vertical face could be taken without distortion. The most laborious preparation was involved for the photograph of the N rampart and Antonine ditch section. The other sections enlarged were the ramparN junctios it t a t ramparn W rampart E wite e th h th t d abouan ; t mid-way betweee nth gate and the N rampart. The photographs do little justice to the preparation, partly because of poor photograph reproductiond yan , partly becaus f incompleteo e preliminary cleaning-upn I . s decide195wa t 7i re-examino dt sectioe eth n ramparacrosW e th s t (Buchanan) 190511 g fi , because this was the only simple rampart section subsequently enlarged; the other two were at the T-junction Antonine th f so e Wall-N rampartsramparW d an t E wit. e hth excavatore Th 190f so 3 commente peculiaritien do rampartse th f so generaln .I cobblee ,th d foundations delimited by kerbing had under the N rampart the normal Antonine Wall width of unded 1 t (4-755 f an othee , rth 0m) r rampart 'tt o t (6-1Bu generalisef . widtsa 0 02 m) y f ho ma t ,i be affirmed that underneath all the ramparts of the fort there are stone foundations of an average width of not less than 20 ft (6-10 m), supplemented by varying margins, adapted to suit special requirements, and increasing the width so that it is nowhere less than 30 ft (9-14 m); and amounts ease th t n rampari t (10-6f ' (Buchana5 ) 3 7m o t n 1905, 460) charactee Th .externae th f o rd an l internal 'varying margins' differed in the several sections. 'The 20 ft (6-10 m) stone foundation junctioe th f o n (the rampartsT-junctio N wit e W he th th f )no become bottoe sth channela mf o , eacn o h sid f whiceo h additional margin formede sar , risin abov) , (15in 0. gThamm 6 e it n o t the outer side increases the width by 10 ft (3-05 m), and is composed of large rough boulders firmly bedde clayn di , whil e inneeth r margihighee th t na r f flaleveo ts i lston e like paving' (Buchanan 1905, 460). Sir George Macdonald, accepting 15 ft and 20 ft (4-57 m and 6-10 m) as averages for the breadt e originath f ho l foundations, thought thaidee f 'varyinth to a g margins prioria s wa ' unsatisfactory, and that a uniform breadth for the whole would be a far more likely arrangement. He carried out a trial excavation on the E rampart S of the gate to check a variation of the outer margin allege 1903n di , foun substantivo dn e evidenc sucr efo hvariationa 'fairlys felwa d t i tan , safe to say that on all three sides the breadth of the (outer) extension had originally been about 234 | PROCEEDINGS OF THE SOCIETY, 1978-80 9 ft (2-74 m)' (Macdonald 1933, 264). Sir George checked the inner margin at several points, including the W rampart adjoining the 1903 major section; although 'the evidence regarding the insid less se wa unambiguou s than could have been wishe t seemi de m ..s o certai.t n that inside f o d en 3-0 d (2-7t widee ) f an 5fortth m 0 e 41 t m th r a r , juso rampar e t outsidf s th ta 9 , s it f etwa o the occupation than it had been at the beginning' (Macdonald 1933,269). As far as the N rampart was concerned: 'the Antonine Vallum alonfore frone th t g th f woulo t d appea havo t r e been similarly reinforced, presumably at the same time. There, however, the sods seem to have been laid on the natural surface without any bottoming of stones, and the precise breadth of the extensio consequentls ni y difficul determineo t t ' (Macdonald 1933, 264). West Rampart Section 1957 (fig 3) The area opened for the W rampart section measured 12-50 m by 2-44 m, the re-cut 1903 section being the N face of the area. Because of the extensive collapse of the 1903 face, which assumed ha naturae dth l angl componens it resf eo f o t t turfworksoi d sectiow an l ne e n th lin, e was about 900 mm N of the 1903 face. The rest of the area lay in the large wedge of rampart remove preparinn di sectioe gth photographr nfo 1903n yi notes a , d above. uppermose Th t190e sectioe leveth th 3s n i dumplsectioe n wa th n o ,n line very irregular, but rising to a more regular crest N of the section line. Most of the dump was rampart material, shovelle grean i p dtu clod t back sectioe sdume cu sidth whelai s S s f Th .n eo de po nwa wa n th the pre-1903 turf line, a dark band very clearly visible here and elsewhere. (In the course of infillin e 190gth 3 excavations dumpd ol e sth , could alway accuratele sb y remove followiny db g this turf line.) Beneat ture topsoie hth th f s linlewa associated wit , grey-browhit colourn ni topsoie Th . l ended wit hfairla y precise definition ove laie rrampartth de turth f fo oved turfe an , rth y material Ove. it easternmos e f rth o E d sectionan e th t W 4-8f o 6, m however topsoie ,th l merged indistinctly into grey-brown material t dissimilabelono , wit textureRomae n ri th n lattee o n ;th y levelsrla . The turfwork was brilliantly denned, with solid black and dark grey vegetation lines contrasting with yellow-orange subsoil survivine .Th g turfwork, 6-1 wid0m e immediately above foundationkerbe e th th f so , rosmaximua o et m insid heighm distancm a et 0 f 1-3a o t 61 7 m f eo the W kerb. At this point of maximum height there were 13 distinct laminations. The greatest number of laminations, 18, was at the middle of the rampart, 1-07 m high above the foundation pitching blace :th k vegetation lines were much closer togethe centrae th n r i ramparle parth f o t t e extremitiesthath t na . There werotheo etw r notable points abou e turfworkth t t 2-1A . 3m kerW fro be mmarkea therth s ewa d depressio laminatione th n ni s abovm m 0 heighea 38 f o t kerbinsid W m poin e e m th , e maximuth f 0 froo t pitching61 e mt th a d m An .heighturf e th -f o t work, there was a 'straight joint' in the turf clearly visible above a height of 300 mm from the pitching. The turfwork from this straight joint westward to the rampart face was of a lighter colour than the rest of the rampart. The W face of the rampart survived intact to a height of abovm keraboue m eth 0 b 99 twit vere hth y steep anglbeed f 78°ha neo t I .preserve thio dt s remarkable degre turfy eb y materia metallind an l g agains. it t W of the rampart turfwork, the level below the topsoil could be identified as turfy material, and - were it not for the intepretation given by Sir George Macdonald - would unhesitatingly have been interprete materias da l fro uppee mth ramparre parth f o t t dislodge weatheringy db . The turfy material was 890 mm thick. The turfy material lay on top of the uppermost of three apparent road surfaces on the berm surface .th This eswa interprete foundatioe rampar190n dth i W s e 3a th r t extensionnfo ; t unreasonablno a e interpretation then, becaus Buchanan i 1 ps 1 ie (a n 1905 showsonls ywa t )i MACIVOR, THOMAS, BREEZE: ANTONINE WALL FORT OF ROUGH CASTLE | 235 sketchily cleaned off. When thoroughly cleaned, however, it appeared as a surface of gravel and small cobbles above a bottoming of large cobbles. Below this upper surface was a layer of mixed white-grey material 50-75 mm thick, resembling the turfy material above. The heavy cobbles of the upper surface penetrated into the white-grey material. Below the white-grey material was a second surface, again of heavy bottoming supporting a cobble-and-gravel surface - very well bettelaid dan r tha othere nth ,explorator n anda s a , maie y th trenc nf o draw hW n section showed, extending right up to the edge of the berm. The E faces of both of these levels metallinf o g butted closel yrampare agains facth W f eo e ttth turfwork. Belo lowee wth r metalled surfac anothes ewa r laye mixef o r d white-grey material again about 50-7 thickm 5m . Below this in turn, and laid immediately on the subsoil, was a light yellow-orange gravel spread. The latter abutted on the W kerbs of the rampart foundation; it did not extend right across the berm, and was less well compacted tha highee nth r surfaces. On the S face of the trench (much reduced in height by the operations of 1903), the surviving laminations of turfwork were interrupted above the rampart E kerb, and were then resumed in turfwork built against the E face of the rampart of equally good quality, to extend along the remaining 1-90 m of the section face. On the E face of a baulk, left across the trench until a later stage of the excavation, this extended turfwork returned for 300 mm and then stopped on a sharply defined line plan I .aree nth a showed that this turfwor beed kha n raise kerbina n do g extending 2-4 eastwar9m right da t angle rampare th o st t rathef kerbso d r,an rougher construction than the rampart kerbs. On further exploration this eastward extension of tufwork was identified, within reasonable limits of certainty, as an ascensus. extensioe limie th f Th o t n turfwork turnerigha t a td S angl distanca t ea 2-4f eo fro 9m m rampare th kerE tlin n b(i e with three post-holes noted below) trenched investigato an t , g sdu e furthes it r course showe t stildi l continuin t 8-8ga beyon4m anglee dth mako T . e absolutely sure of the plan of the extension, the trenching should have been continued up to the W gate; it was laten i notd r commenan , t wil i t assumee b l d tha extensioe th t likels nwa continuo s yt a r fa s ea the W gate. The E face of the rampart was much less well preserved than the W face, preserving its original steep profile to a height of only about 150 mm above the kerb. At that point, there was a large hollow in the section of the turfwork. Extending 1 -22 m E from the kerb was a mass of turfy material, with more cohesive fragments of turf than the similar material to the W. The hollow and the nature of the fragments suggested collapse rather than weathering. This broken turfwork lay directly on top of a thin but well-defined layer of burnt wood and clay which, abutting against (and thus presumably preservin gsurvivine froth f mo p decayg to 15m e 0m )th f ramparo t rear face, extende 2-4r d fo 2-7 o 4t eastwar4m d fro reae mth r face. Wher coverins eit g of broken turfwork ended, the layer of burnt wood and clay was sealed by the greyish-brown material indistinguishable from the top soil which has been noted above. Beneat layee h th f burnro t woodiscontinuoua clad s dan ywa s laye lighf o r t brown limey materia thickl m abou beneatd m 0 an ; 5 t h thifurtheo stw r burnt layers separate smaly db l sharp rubble. Lowes lyinalld f o t an , g immediately abov subsoil e irregulan eth a s ,wa r pitchin largf go e cobbles. Approximately bounding the pitching, at a distance of 2-29 m from the rampart E kerbs paralled an themo lt , were three post-holes turn .I n beyond these, about 3-3 fro 5ramparm e mth t E kerbs, was a stone-revetted drain, only partly preserved, which curved away from the rampart, trenche parth S f e o ti n.th In the longitudinal exploration of the extension, good turfwork was found through its whole length, beneat 190e hth 2 tursoi p exploratioe fto l th lin andd f o ean ,d towarden nS e sth also beneath a 1903 dump. These levels were interrupted by a filled-in trench, 960 mm wide, 236 | PROCEEDINGS OF THE SOCIETY, 1978-80 beginning 5-94 m S of the N kerbs of the extension. The trench had been dug through 1902 and 1903 levels into the turfwork, was filled with substantial well-defined pieces of turfwork, and ended in a rounded bottom 1-09 m below present ground surface. It appeared to be the trench dug in 1932 by Sir George Macdonald to check the inner extension of the W rampart and illustrated by a photograph in his report (Macdonald 1933, fig 10). North (Antonine Wall) Rampart Section 1958 (fi) g3 In 1958 a trench 8-53 m long by 2-44 m wide was opened to take in the N rampart and the area immediately N and S of it. Since the main body of the rampart presented no anomalies centrae th rampare th l f 2-1o 3tm turfwor lefs tk wa undisturbed turfworrampare N e Th . th f kto survive muca o dt h lesser height rampartthaW ne th tha f maximua :o t . mm f onlm0 o y61 Neither face was preserved. To the N the turfwork tapered off to the kerb; beyond the scanty remain laif so d nondescrips turwa f t greyish soil going dow kerbe subsoilo nt th f arese o aTh .N was remarkably fre f stoneso e : ther onls ewa y one smala , l boulder half-burie subsoile th n di . There were more complicationturfwore Th . S ke th slope o st droundeo bactw n ki d steps from the kerb d botan ,h plasectiod nan n show evidenc f muco e h disturbanc around an e n ei dth sectioE e kerbsth nn facloweO .e eth ramparre steth f po t covereturs fwa thic m dm k wit5 7 ha stone, clay and earth fill with patches of clay, and burnt material on top brightening to a large patc bricf h o burn d kre t clay kerb S ove outed e san rth r pitchin rampare th f go t foundatione Th . brick red feature (for which the obvious identification is a furnace) was about 610-750 mm long. The stone, clay and earth fill extended to the middle of the trench, where it stopped against a rectangular post-hol t immediatelse e y behin e rampardth t kerbs e post-holTh . s filledwa e , not with the stone, clay and earth mixture, but with a distinct black material. A pair of closely adjacent post-holes were found, their centres distant 560-61 fro m othere 0mm th , wit same hth e filling. Both post-holes of the pair were round, about 200 mm diameter. The outer of the pair was sunk deepbelom m levee littl a w, 0 th - f paralle o eacrosl10 im m trenc d m d 0 san an l h28 mediately adjacent to the rampart kerbs; the trench was 1-22 m long and rode out above exist- ing subsoil level to the E. Apart from these features kerbe distanca th r area e f sfo betweeth d o , f 2-3sS ha eo 6m n topsoil and subsoil a yellow-brown stony fill which underlay the stone, clay and earth mixture. At 2-36 m from the kerbs were the first traces of the presumed intervallum road which will be described later. Comment Concerning the structure of the N rampart the section adds nothing to our general knowledge of the Antonine Wall rampart of which this forms a part. The W rampart section gives the steep profile of 78° for the outer face; the distinct character of the first 610 mm of rampart turfwork, together wit slumpede hth lamination centree th n si , suggests tha rampare th t t scarp may have been built up first and the main body of rampart material filled in less carefully behind it. This was the technique used in earthwork artillery fortifications of the 17th-18th centuries, though in these the scarp turfwork was thicker, and the infill was of earth not turf. In a structure built entirely of turf the method might seem unnecessarily to sacrifice cohesion; yet this is no valid criticism in the present case, where the state of preservation of the rampart shows that the structure cohered remarkably well. The results of the rampart sections cast a measure of doubt on the hypothesis that the rampart was anywhere extended to augment its wallhead, and controvert the hypothesis that the rampart was systematically extended internally and externally for this purpose. MACIVOR, THOMAS, BREEZE: ANTONINE WALL FORT OF ROUGH CASTLE | 237 definee Th certaid dan n extension 2-50 m wid ramparinterpretee W eb e founy th t ma da d with some confidenc ascensus,n a s a e which though structurally distinct fro e ramparmth s wa t probably built (since there is no occupation material between the foundation of the extension and subsoil) as part of the same construction programme. One may readily understand that Sir George Macdonald, already disposed to think of systematic rampart extensions, might regard as conclusive evidence the turfwork found in his trench. N of the ascensus, the layer of burnt woo clad dan y seale turfy db y material apparently from rampart decay afford evidenco n s o et support rampart extension. The post-holes, and the burnt layer, suggest some minor structure in the angle between rampart and ascensus, similar to the burnt-down structure in the NW angle fore ofth t described belo 238)wp ( . The existence of three road surfaces on the berm outside the W rampart section, the upper- most covere turfy db y material again apparently from rampart decay, afford moro n s e evidence supporo t hypothesise th t hardln ca maintainee t yI b . d thaturfe th t y material represente th s materia purposefuf o l l extension, particularl materias ya f seeminglo l same yth e compositios ni found sandwiched betwee surfacese nth unfortunatels i t I . y whay easie functioe sa th to t r f o n roae th d surface nots meano n whawa s ,y nee y y thasa b st wasan i ts d o cleani t y (fat i : wh rr les persistinsa g need founs themr wa ) dfo surfacee Th . givo d s e three periods f somo e kinf do operations on the site, separated by periods of time marked by the material from the weathering ramparte th f roaa o t d;bu surface twice repaired after interval t evidenc timf so no s ei threr efo e period f occupatioso n interrupte withdrawaly db . N of the N rampart there were no features of any kind, not even fallen turfwork. S of the ramparN existence th t likela f eo y furnace cutting int turfwore rampare oth reath e f th o rf ko t abovm foundatioe m onleth 0 y24 n kerbs preclude extension sa n here too. Again post-holee th , s indicate slight buildings agains ramparte th e rea f th to r . Thoug e lathth e Antonine pottery associated with the post-holes (135) does not necessarily date them, a late date would agree with the general impression of the infilling of unused space during the later part of the occupation, caused by an increased density of occupation. Internal Intervallum the roads and Background 190e Th 4 pla Rougf no h Castle (Buchanan mark1905) 1 g dottey fi ,s b d line anticipatee sth d principalis,a positiovi e th f no thoug t hshowi structurao sn ldoe d remaint naman s no t i t ei f so a 'street' s a e anticipateTh . praetoriada vi positio e th s unmarked i f no . Fragment f 'streeto s ' fore sideS th t f d correspondinso an E , N anticipatee showe e th th ar o gt n no d positioe th f no intervallum road. The text shows that Buchanan accepted the space between this 'street' and the rampart to have been filled by the several inward extensions of the rampart. To the S of the site gateW e ,o th 'Buildinf 3o 'g N (th e commanding officer's house) approache closelo e s th o yt rampar leavo t spacs o 'streeta ea t n r correspondina efo n i ' g relationshi ramparte th e o pt th f I . commandine limith plottinW f e o t th f go g officer's hous rights ei t appear i s froe a ,b e mth o st much decaye overgrowd dan n remain sitee t opeth ,sye n whano t happene intervallume th o dt road at this point is still a puzzle. Via Principalis Praetoriaa Vi d an 1959-61 (fi) g4 Parts of the boundaries of the via principalis were exposed during the investigation of the praetentura and during the superficial examination of the structural condition of the headquarters building made in 1961. It measures 7-32 m across from the facade of the headquarters building furthee th o t r margi draia f no n whic levelsideo hN boundee metallinf Tw .so th n o t di g could Q | PROCEEDING 8 23 THF SO E SOCIETY, 1978-80 be distinguished: the lower more carefully laid and including the drain; the upper lacking defined margins and spreading across the drain, the position of which was only checked in places. In the width of the via principalis NW of the headquarters building was a stone-lined tank 0-84 m by 1-32 m and 1-52 m deep. It received drainage from the N side of the via principalis praetoria.a vi e tane sidpresumablth s E f Th ke o wa th d an y covered with timber. The W boundary of the via praetoria was defined by the drain mentioned above. As origin- ally constructed centrs it f i ed linan , e corresponde centre gatth N o de t th (acceptinlin f eo g for the latter centre line the evidence of the 1903 report) it would be less than 9-10 m wide; though in fact it appears to have been much wider, as metalling which seemed to be integral with it was foun distanca t da 9-7f eo beyon 6drainm W praetoriacentr e a e dvi Th .th e y th lin ma f eo rather have corresponde centre th o e headquarterdet th lin f eo s building, wit extremeln ha y generou spacf smalo o s e n eforpoini la a s us - t t commente again do considerinn i timbee gth r buildings in the praetentura. In the via praetoria, like the via principalis, the upper metalling extended beyon lowere d th cam d poorla an , o et y defined edge.

Intervallum 1958-61 intervallume Th roaexposes greateW e dwa e th r sidth N d fo f rlengte s o par it th f n o th o praetentura, sidepreservatioe W lengtuppee s Th .o par d it th tw f an n o ts hr o it surface f no s swa remarkably good. These wer smalf eo l cobbles finished with rammed gravel lowese Th . t surface was less substantial tha e othersn th middl e Th uppe.d ean r surfaces were bounde kerbsy db , uppee th rt throug surfaccu draina s y hwa eb . principalisa Unlik praetoria,a vi vi e ed th e an th margin of the upper intervallum road surface lay within that of the middle surface. The width of the intervallum road was difficult to determine exactly at those points where its probabl limiinvestigatedes N twa ramparN kerbse ,o becausn th t d t A .sectio ha s t ei wa t ni about 2-44 m wide, leaving a space of about 1-83 m between the rampart kerbs and the road. Insid ascensuse eth withiramparW onln e nca th y t ti hav e been about 1-2 wid 2durinm d ean e gth later occupatioe yearth f similarls o wa t ni y reduce sligha y db t ascensus.buildine th f o gN e remain e latteTh th f ro s building were examine n 1958i d bee ha n t . I constructef do posts holdin stakee gwattlth a f so e frame, apparently extended froramparte bace th m th f ko , facn i ty havma ed beean n little more tha nligha t screenbeed ha n t I burn. t thica dow d knan deposi f burno t principate wooth clas d dan ywa l evidenc f conflagratioeo n found durine gth whole excavation crudA . e stone setting, which overlai burninge dth , produced several finds.

Comment roae Th d surfaces gav mose eth t unequivocal evidenc differenf eo t period worf so k within areae th s examined, thoug t necessarilhno y distinct 'period occupation'f so . Although the fire in the little building could have been caused by a careless Nervian auxiliary majoe th f ro military readils an a y b ys y eventa secon e th f so d century occupatioe th , havy nema closed finally at the time when the little structure at the NW corner of the intervallum was burnt, for the debris was never cleared up. Similar burnt remains of buildings in the praetentura were removed in 1903, for references to burnt material are made in several places in the report.

Praetentura (figs 3-4) Background remarkable spitn th I f eo e achievemen t Ardoca t 1896-n hi 7 (Christison 1898, 443-9e )th subsequent Romano-Scottish enterprise thif so s Society mad furtheo en r contributio knowo nt - MACIVOR, THOMAS, BREEZE: ANTONINE WALL FOR ROUGF TO H 9 CASTL23 | E ledge of timber buildings save at the . At Rough Castle, where in any case too much seem havo st e been attempte littlo to en d i tim witd littlo ean to h e little supervision evidenco n , e of timber structure notedplas e th swa n n I o. f the praetentura there appear fragment f 'street'o s ; L-shapen a 'draina d an ' d fragmen f pavingo t ; 'Of stone buildings', wrote Mungo Buchanan, 'there wer probabls ei t nonei erectiond y an ,ean s above groun d- occupyin blane gth k spacn eo the plan - very temporary, perhaps of timber, and subject to occasional re-arrangement. What- havevey ma er been locate thosn di e evidently much used position onln mattese ca y b f cono r - jecture evidenco ther n s definit,a y s ean wa f eo e arrangement, althoug whole hth e areexplores awa d down to the original surface' (Buchanan 1905, 481). Sir George Macdonald conjecture possible dth e plannin barrackf blange o th t kfi spaceo st s retenturae plane th th n praetenturai n e , o wels th a s a l (Macdonald 1933e th , 274-6)w no e se ; Discussion below (pp 278-83). West Praetentura Excavation 1957-61 Three squares of 2-44 m were opened in the westernmost side of the praetentura in 1957, to see if there were any traces of timber barrack buildings. The results were promising enough to encourage further exploration; larger areas were investigated in 1958 and 1959; a 7-32 m wide zone dow westere praetorianth a vi ne strippes sidth wa f eo 1959 n dsupplementari a d ;an y area in 1961. In the course of the work it was found that the operations of 1903 had been as thorough as Buchanan implies. Apart from fragments of paving and road surface (not all shown on the 1904 plan) isolated an , d stones som f whiceo h wer packine eth f post-holesgo whole th , e area withi e intervallumnth road beedha n cleared right e subsoildowsomn i th d o ent an ,place s possibly below it. It seemed to have been cleared by preliminary extensive trenching followed by the total clearance of broad strips, the strips being back-filled as work went on across the site. mighe On t expect that this wholesale clearance would have considerably simplifie identifie dth - foundatioe catioth f no n trenche r post-pito s f timbeo s r buildings t therbu ; e wer visiblo n e e foundation trenches, and the filling of both post-pits and post-holes was surprisingly difficult to distinguish from undisturbed subsoil. The post-holes within the barrack area were much less clear than those already described immediately insid rampare eth t foundation. Progreso s s swa slow and painful that the areas which could be stripped, cleaned and kept clean fell below the generous scale necessary for a really successful horizontal section. The areas examined in the 4 years will be described together, as they are shown in the conflated plan (fi . Wherbeed g2) ha n t ei extende faro dexcavates e ,th d areboundes awa N o dt and W by the intervallum road, to E by the via praetoria, to S by the via principalis. The roads have been described resurfacinge th : e intervallumth f o s e otherth road s dan wil recallede b l , together wit tongue hth metallinf eo g protruding from abou centre praetoriath ta vi e e pointh f o t betwee principalis, a intervallume vi nth fragmente e th th d d an f metallin an o s g apparently con- tinuing the northern line of that tongue. The roadways make up the possible limits of a pair of rectangular buildings with their long access E-W, takin metalline tongue th gth d ean g fragments as the remains of an intermediate road. Ther alss odetacheea wa d are pitchinf ao g measuring abou2-1y b E-W8 m S t 2-8N- , 5m within the angle formed by the via principalis and via praetoria. The pitching had on its surface a patch of burnt clay. There was another small area of pitching S of the intervallum, and yet another about half-way betwee intervallume nth possible th d an e intermediate road. Apart from thes settingw efe areasa f stone o sd an two, n si threed an s s (mos f whico t h emerge likels da y post-holes), there wer featureo en aree th a n i sabov generae eth lsubsoile leveth f o lwhic o t , h beed ha n most i cleare f o t 1903n di . | PROCEEDING 0 24 SOCIETYE TH F SO , 1978-80 The level of the subsoil below adjacent areas of metalling and pitching varied from place to place: 460 mm below the tongue protruding from the W side of the via praetoria, and the same below the angle of via principally and via praetoria; 380-300 mm below the uppermost level of the inter-vallum road, 150 mm below the middle level, and a negligible depth below the lowest mosd an t lightly constructed level. preveno T t water fro roade mth s draining int buildingse oth flooe ,th r building y levelan f so s would be above the level of the surrounding roads. So it is clear that the operations of 1903 must have destroyed a considerable depth of occupation - at least 460 mm in some parts for the lates toccupatio e yearth f o s n- withi encircline nth g framewor roadsf o koperatione Th . s could have destroyed structural evidenc s wellea , particularly evidenc e latesth r t fo eyear f o s occupation. Withi aree n th numbe a f certainro , probabl doubtfud ean l post-hole lessed san r stake- holes was discovered together with channels and shallow trenches. None of the latter seemed regula r considerablo r e enoug havo ht e bee nfoundatioa n trenc substantiaa r hfo l elemenn i t a timber building. Some of the post-holes were inexplicable in terms of structures. The rest suggested the plans of three timber structures: two, one replacing the other, in the northern areae halth f , o fone , similalatee southertheseth f e o rth o t r n i , n half t musI . stressee b t d that these structures, to be described, represent a probable interpretation of the evidence, which lacked that uniform clarity and sharpness of definition necessary to give absolutely certain conclusions. In the most substantial (and, it will be argued, later) building in the N the walls were not exactly parallel, giving mean dimensione th f o d 26-5f so en E-2m W W 5-7y e b N-S 9th m t A . building werepossiblo tw e partitions wite on ,h surviving stake-hole wattldaua a r d sbfo e an wall, giving compartments 4-12 m and 3-81 m wide respectively. Near the NE corner of this building between two of the uprights was a group of broken flags at the same level as the middle- period intervallum road surface; the group may have formed a threshold. Evidence for the main fram thif o e s apparent building seemed trustworthy enough post-holee :th s which were missing were in areas disturbed by pits and by recent tree growth. Paralle longitudinae th o t l l centre-lin wella s f thieo - wa s , buildingit f littla o d eN an , defined lin f post-holeeo s whicjusW t d hgabls shor it an ende t firsf a E eo t s twallso dt wa t I . believed tha line th te would prov lino et kforo s wit msecona d han similaS d d an r lineN o st conventional timber structure. In the course of the excavation no such similar lines emerged. Instead observes wa t i , d tha line post-holef th teo regularls swa shalloa y yb encloseS w d an dN rectilinear channel returnin roundine roundey Th g b . observes W ge wa d th angle an t da E o st two angles whic anglee channeE S hTh uncus s . coul,it wa lt NE examinede a t db d an W S : where a large natural boulder remained embedded in the subsoil; and it rode out altogether over much of the length of its N and W sides. The sides of the channel were not absolutely straight, and it was not uniform in depth. channee Th l migh associatee tb d wit line post-holef hth eo s which bisect drip-trencha s a t si , though a drip-trench around a very unconventional structure. For the only other possible related structural elements which were observed wer serieea smalf so l hole staker sfo s rather than posts (they wer t surroundeeno centrae pits)y th db f o , l equidistanS lin post-holesf d eo an N o d t an , about 610 mm from the centre-line of the channel. If post-holes and stake-holes are indeed associated, they form a series of tent-like structures 3-05 m wide with an overall length of 25-00 m. The sequence of the two structures is suggested by the supposed threshold at the NE more anglth f eeo substantial building, which overlie channee th s r drainago l e trenc t thiha s point. N U M A

P^^tf«O-JCg>^u^L^-u^LLjl JJ \ U ^) ie>^^yi^;«!VJKSp«g*| » :",?S4 a f •-^--^-^^'•m^'JBB^1'^™'^fp|^i

imu!niiMilllllliinnti>C /««.'"lii.iiiiii»ii<-4r-P'"""""""«/// |S|»^»^

^:^^x>Smw IDMilMs

^l®^I^Mi^pflP^I)l?

V I A I PALI C N S

10 20 Ft FIG 4 Rough Castle: via praetoria and area to W Crest of dump behind section ANTONINE WALL RAMPART

NORTH RAMPAR F O FORT ) T Kerbs only

reveaIe d

West Rampart Section

INTERVAL LUM

West Rampart Section

Ascensus

I NTER-BARRACK ROAD

FIG 3 Rough Castle: NW corner and W rampart section MACIVOR, THOMAS, BREEZE: ANTONINE WALL FOR ROUGF TO H 1 CASTL24 | E To the S, post-holes giving the apparent east end of a building 6-25 m wide, with a S-facing verandah-like extension increasing the width of the building to 7-32 m were found. That part site ofth e which would hav unexcavateds e givewa d Ws nit en . East Praetentura Excavation 1961 196n I 1 zona e 7-3 wid strippe2s m endewa W possibl f se o tak o dth t n ei e timber buildings E of the via praetoria. It was positioned to relate to a viapraetoria centred on the apparent centre of the N gate, which is considerably W of the centre line of the headquarters building. A number of post-holes were found suggesting a timber building with its long axis N-S, parallel to the via praetoria. This promising beginning proved to be illusory. In spite of the most meticulous examination convincino n , g building emerge ende th t .da Clearly timber structure somf so e kind had been erected in the area, as evidenced by the post-holes, and perhaps too by the badly mauled metalling which must belong to a very broad via praetoria, which it was difficult to believe had been so much cut up by the operation of 1903. A further puzzling anomaly was the existence of a drain bounding the intervallum road, and penetrating far beyond the point where the intervallum road would have abutted on the via praetoria. The inconclusive investigation of the post-holes within this stripped area occupied so much time, thalittlo to t e remaine extension a r aree dfo th a f neastwardo existence checo t ,th r kfo e of buildings which could still be fitted in beyond the area examined. And yet, had sufficient time remaine extensione th r dfo t wouli , d have been undertaken with misgiving excavatioe Th . t na this point was in danger of repeating a mistake of 1903, the pursuit of investigation beyond the point at which interpretation had begun to fail. The excavation began with modest but certain gains of knowledge in the rampart section; the operation in the Wpraetentura led to conclusions which, though probable, wer t certain considerabla eno w no ; e arebeed aha n stripped without advancemeny an r understandinou f to sitee th .f go I decided that the rest of the E praetentura, like the SW part of the W praetentura, should be left to a future second opinion. Comment The operations in the praetentura provide the unexpected element of an initial semi- permanent occupatio tent-like th fore n th i t f neo structures noted above. Though unexpected, suc occupation ha readile b y ynma enough accepted t point.I s merel delaya o yt necessarilt no , y a long one, in the erection of permanent barracks, and associates quite comfortably with the evidenc initiae th f eo l slight surfacin intervallumberme W th f e go ,th alreadd an y noticed. paie f morTh o r e solid building stakee whicb y maro nt firshe ma kth t substantial period of barrack constructio expectee nar sizn sitingd i ean , particularl ther s ystila y postue b lema - lated buildings similar in width and up to 15-24 m long in the E praetentura. The general spacing, however, does odditieemphasise fortth e f thath n o i ,f so e t thougeon h fro beginnine mth e gth e headquarterth s forwa t a 500-stron f o s g cohort ,e cohort' mosth f o t s personnel must have been accommodated elsewhere. Most surprising was the absence of evidence to indicate a second substantial period of barracks, Antonine II to complement Antonine I. The evidence as it stands could be used to sustain an argument that, at Rough Castle, the occupation - which from the pottery seems to have the usual termini of Antonine sites elsewhere - continued without any violent break. Nevertheless suc argumenn ha t woul littla e db e reckless depte Th .f materia ho l stripped from the barrack area in 1903 could have removed all traces of a later set of timber structures, if these were constructed on fairly shallow sleeper beams. The limitations of the evidence - and perhaps 24 | 2PROCEEDING S OF THE SOCIETY, 1978-80 too the severe limits of the investigation - mean that in this case the existence of a definite break dividing Antonine I from Antonine II is not proven.

CATALOGU FINDF EO S 1903-1961 In the following section US indicates that the artefact is unstratified, NMA that the object is in the National Museu Antiquitief mo s (the catalogue number usually follows 324)R F tha g shere W ,e , th t d wa WilliamssF examine D thasher e A r subjectes th tD dHM wa ,y db heavo dt y mineral analysiy sb Williamsr D asterisn a d ,an k tha objece tth illustrateds i t . 'Main ditch' Antoninrefere th o st e Wall ditch. Stone 2 Fragmen d an 1 t with narrow rope moulding fragmend an , t with curvilinear carving, par largef o t r 368. & 7 ,US 36 R F object A NM . 3 Fragment of squared block, 30-40 mm thick, maximum surviving dimensions, 95 by 95 mm. One surface is corrugated, with nine rough grooves, c 3-4 mm wide and 1 mm apart. From the east gate area.

12

FI G5 Roug h Castle: whetstone) 2 : 1 s( . 4 FragmenUS , mm squaref 0 o t8 y b d0 bloc20 y stonef kb o 0 33 , 5 Two fragments of grey stone slab with mica, with circular hole drilled through it, and with white edgedeposie sigo on weaf .N nn o to r around hole. Fro maie mth n ditch. ,US 6-7 Fragment stono tw ef sdiscso wite on ,h hole bored throug , NMAhit . ,US 8 Fragmen stonf o t e lamp? NMA (AndersoS ,U n 1905,. 4954) g fi , *9 Whetstone, approximately rectangula 357R . sectionF n ri , US A NM , 10-11 Two whetstones, section as 9, NMA FR 362 & 363, US. *12 Whetstone, approximately ova364. R sectionF n li , US A NM , 13 Fragment of whetstone, oval in section, from the lilia spoil heaps, US. *14 Flat upper stone of quern, approximately circular, with central hole; also, two holes on one worn par quernwhicf f to o e hon , perforates othen querno d r sidan , quernf eo hole on ,e connectinp gto and edge, US (fig 6). Also, 'a broken quern of vesicular lava' (Anderson 1905, 442-499). Baked Clay fragmene On 15 f bakeo t d clay fro lilae mth spoil heaps, US alse se ;o 290-3. Tile fragment5 tiles10 f thesx 1o 6 markee bo 1 ;e4 f ar o s d wit acute-anglen ha d lattice pattern wit3 1 , ha very compact grid, two with a very open grid and three with straight lines. Also, 16 rounded and 8 flat fragments, all plain. All from the bathhouse spoil heaps. MACIVOR, THOMAS, BREEZE: ANTONINE WALL FOR ROUGF TO H 3 CASTL24 | E

SECTION 2

SECTION 1 FIG 6 Rough Castle: quernstone (1 : 2) Samian Ware BHartleR Bd yMan Dickinson, Departmen Archaeologyf o t , Universit Leedf yo s The following abbreviations are used: D. (Figure-type in) J Dechelette, Les Vases ceramiques ornes de la Gaule romaine. Paris, 1904. . O (Figure-typ F Oswald ) ein , Index f Figure-typeso Terra-sigillatan o ('samian ware''). Liverpool, 1936-7. Rogers G B Rogers, Poterie sigillee de la Gaule centrale, Gallia Suppl, XXVIII (1974). 24 | 4PROCEEDING S OF THE SOCIETY, 1978-80 17 Central Gaulish scrap. Antonine. From road barrackbetweeS d praetentura,W an n nsi N . US 18 Form 18/31 r 31RRo , more probabl lattere yth , Central Gaulish latr o ed Antonine.Mi . 19 For , Centram33 l Gaulish. Antonine. Fro barracparN m W f o t Wn ki praetentura,. US 20 Form Curie 11 etc, Central Gaulish. Antonine. From W part of N barrack in W praetentura, US. 21 A small scrap of form 37, Central Gaulish, probably with freestyle decoration and with the partial impression of a leaf used by Albucius ii of Lezoux (Stanfield & Simpson 1958, fig 35,2). c AD 150-80. Fro mintervallumN road/W intervallum roa burnn di t material. 22 Form 31, slightly burnt, Central Gaulish. Antonine. From W part of N barrack in W'praetentura, US. 23 Form 33, Central Gaulish. Antonine. From area between N rampart and intervallum road, US. smallA 24 , eroded fragmen f for , o Centrat m37 l Gaulish Cupie Th . d (simila . 458A)O o t r , naked figure (Z>. 382abnormalld an ) y large festoobowln o e snar from Balmuild style th Cettuf eo n y i f so s Martres-de-VeyrLe e (Miller 1922, pi XXXIV 44)& .2 4 Bowl, thiy b s s potter are f courseo , , common in Antonine Scotland. For a discussion of his date, see Hartley 1972, 34. c AD 135-60. From W intervallum road, US. 25 Fragments from two dishes of form 31, both Central Gaulish and Antonine. From W part of N barrack in W praetentura, US. 26 Form 31, slightly burnt, Central Gaulish. Earl mid-Antonineo yt , Fro barracparN m W f o tW n ki praetentura,. US 27 Form 30 or 37 and an unidentified scrap, Central Gaulish. Antonine. 28 For , Centram31 l Gaulish. Earl mid-Antonineo yt . 29 Form 31, Central Gaulish. Mid or late-Antonine. 27, 28 & 29 from W part of N barrack in W praetentura, US. 30 Dish (?) fragment, Central Gaulish. Antonine. Fro mintervallumN road, phas roae3 d drain. 31 Three joining fragment forf perhaps, o m31 s slightly burnt, Central Gaulish. Earl mid-Antoniner yo . Fro intervallummW road, seale burny db t material. 32 Form 31, Central Gaulish. Early or mid-Antonine. 33 A tiny fragment of form 37, Central Gaulish, with a Perseus (Z>. 146) used at Lezoux by several potter perioe th n si Adc D 140-80. Fro intervallummW road, phas roae3 d drain seale burny db t material. 34 For , Centram31 l Gaulish. Antonine. Fro barracmparN E praetenturaf o tW n ki . US 35 For , Centram31 l Gaulish. Earl r mid-Antonineyo . From road barrack S betwee d W an n si nN praetentura, US. 36 A small, eroded fragment of form 37, Central Gaulish. A panelled bowl, perhaps with the leopard (/). 799) use severay db l Lezoux potter Antonine th n si e period. Fro mbarracS praetentura,W n ki US. (?)m 3ri , 7 slightl 7 3 Forr o y0 m erode3 n burnt a d dan , scrap. Both Central Gaulis Antonined han . Fro mbarracS praetentura,W n ki . US 38 For , Centram31 l Gaulish, perhaps froMartres-de-Veyres mLe . Earl r mid-Antonineyo . FromE barracparN praetentura,f o tW n ki . US 39 rim 7 3 For,r o Centra m0 3 l Gaulish. Antonine. 40 For , Centram37 l Gaulish, wit ovoln ha o (Rogers B52doubld an ) e festoon use t Lezouda y xb Criciro v, Divixtus i and, almost certainly, Secundus v. The sphinx (D. 497) has apparently not bee nthesf recordeo y e an potters r dfo Ac , D 145-75. Fro corneE mN fortf . ro US , 41 For , Centram31 l Gaulish. Early-Antonine. Fro cornemW S forf ro t wall. US , 42 rim 7 3 For,r o Centra m0 3 l Gaulish. Antonine. 43 Form 31, Central Gaulish, Early to mid-Antonine. 44 Form 27, burnt, East Gaulish. The stamp is almost certainly literate, but has not been identified. Ther severae ear l possible reading , nonit f seo suggestin gknowa n potter, thoug fabrie hth c points either to manufacture in the Argonne or at La Madeleine. However, Argonne ware has not been recorded fro Antonine mth e Wall, wherea Madeleina sL t )a eg war(e s eha . This tipe sth balanc favoulattern e i th fore f mo Th r. shows that this piec earlies ei r tha4 4 n & A 3 D4 160, 42 . fro bath-house mth e spoil heap. 45 Joining fragments giving approximately half of a dish of form 18/31-31, slightly burnt, probably fro Martres-de-Veyres mLe footrine Th . almoss gi t unworn. Early-Antonine. MACIVOR, THOMAS, BREEZE: ANTONINE WALL FOR ROUGF TO H 5 CASTL24 | E 46 Form 37, Central Gaulish, with a winding scroll and grooved for a rivet. The ovolo (Rogers B143) and bird (0. 2239B?) were both used at Lezoux by Cinnamus ii. c AD 150-80. 47 Form Curi , Centrae21 l Gaulish ungrittee Th . d samian mortari f thiao s type were undoubtedly introduced considerably t unknow earlieno e rar Scotlandthan d i gritte e an n th 5 4 d d form. an 3 s4 Antonine. 48 Form 31(3), Central Gaulish. Antonine. 49 Two joining fragments from a small bowl of form 37, with the single-bordered ovolo (Rogers B28, very blurred), lion (O. 1425) and trilobed motif (Rogers G164) used at Lezoux by Quintilianus i. The other figure-types are a warrior (O. 219A, repeated) and goat (D. 885), neither, apparently, recorded for Quintilianus or any of his associates. The border of large, squarish beads is not commo Quintilianus,r nfo t appeari t bu typologically-lats somn shi o f eo e work, curiously enough nearly always in association with the lion and trilobed motif of this piece, c AD 135-155. 50 A large fragment of form 37, Central Gaulish, with a winding scroll with double medallions in the lower parts. All the details were used at Lezoux on stamped bowls of Cinnamus ii. They are: ovolo (Rogers B145), leaves (ibid. H13, J89), trifid motifs (ibid. G66), bird (O. 2298) and erotic group (Oswald pi XC, B). Both the leaves are on a Cinnamus bowl at Amiens and the group is on bowls hi s from Mumrill Sensd san Ac , D 150-80. 45-50 Fro Antonine mth e Wall ditch. US , 51 A collection of material from the lilia spoil heaps includes nothing that is necessarily earlier than ACentral Dal 150s i t .I l Gaulis included han s form sdecorate o 18/31tw e 31(3)d th R an f d O .bowl s of for , botm37 h with scroll decoration certainls i e on , Cinnamu y yb ovole Th . osii (Rogers B223) bir d . 1038an d(D ) were used extensivel wors hi f Stanfiel kn (c yi Simpso& d n 1958, pi 162, 60). Ac D 150-80othee th lea e n r o fTh (probabl. y Roger smalse H21th d l circl)an e were also usey db him ,mai s bothi nn hi activit earlierd yan , whe workee nh Cerialie th n di s ii-Paullu grouv si t pa Lezoux. c AD 140-70. 52 Forms 31 and 33(2), all very eroded. Central Gaulish. Antonine. US. 53 For basem3 3 , apparently never stamped, Central Gaulish. Antonine. US . 54 A Central Gaulish scrap. Antonine. US.

The material in the National Museum of Antiquities of Scotland fragmento Tw 55 s fro mbowa forf lo wit 7 m 3 ovoln ha o (Rogers B144) use t Lezouda membery xb s of the Cerialis ii-Cinnamus ii group. The panels contain: (a) An eagle (D. 981) and seated Apollo (D. singl52)A ) .(b e medallion with cogged annulet f Stanfiels(c Simpso& d n 1958, pi 156) 7 , ,3 earle useth yn di Cerialis-Cinnamus style (ibid,Ac D. 140-708) 286-7R .F . US , 56 Form 37, Central Gaulish, with ovolo (Rogers B52) used at Lezoux by Criciro v and Divixtus i. However, since Divixtus rarely f everi , , use dwava y line belo s ovolowhi piece th , almoss ei t certainly by Criciro. c AD 135-70. FR 288, US. 57 For witm7 3 h ovolo (Rogers B223) use t Lezouda Cinnamuy xb scheme Th . f decoratiosii e o s ni not clear, c AD 150-80. FR 289, US. 58 For Centra, m37 l Gaulish panelle.A d bowl, with ovolo (Rogers B12), caryati . 1257A)d(O , erotic group (Oswald pi XC, B) and bird (0.2295A) in a double festoon. Most of the details are common boto t h Cricir Divixtud an sinct o v latte e bu bire, e th si festoonr th rd o e apparentl us e t th , no d ydi piec agais ei n mor Criciroy eb likel e decoratioe b Th o .yt precisels ni y parallele bowa n do l from Lezoux with an almost entirely obliterated cursive signature, presumably Criciro's, since Divixtus's bowls never have a signature below the decoration, c AD 135-60. FR 292, US. 59 For , Soutm37 h Gaulish, with zonal decoration vere yTh . light fabri pood can r standar workf do - manship suggest that it comes from Montans. The fan-shaped plant used here in a basal wreath appear bowa n t Carlislso a l e stampe Feliciy db o chevrone iii .zonTh e th en si abov wreate eth h are also paralleled on the Carlisle bowl, but there they face the opposite direction. The next zone singls ha e festoons smala f o .l e Felicigrouon s Montanf po wa s potter workeo secone swh th n di d century. Their work is relatively common in Antonine Scotland (Hartley 1972, 42-5). c AD 110-45. FR 293, US. 60 Form 37, Central Gaulish, with scroll decoration. The leaves (Rogers Jl, 153) were both used at Lezoux by Cinnamus ii, and appear on bowls from Segontium and Lezoux (?) respectively. The bird, apparentl commones hi f o reverse e yth on r f figure-typeeo . 1038)t knowm (D sno hi s r i , nfo 24 | 6PROCEEDING S OF THE SOCIETY, 1978-80 and doe t appeasno Dechelettn ri r Oswaldeo t presumablI . y comes fro mdouble-endea d figure- stamp, which some potters used for pairs of figures (birds, hares, etc). One end of this particular figure-stamp was probably broken very early in its life, while the other end continued in use. An early dattherefory ema suggestede eb Ac , D 150-65 294. R F . ,US 61 Form 37, Central Gaulish, with scroll decoration. The ovolo (Rogers B144), leaf (ibid, H21) and bird (D. 1019) were all used at Lezoux by members of the Cerialis ii-Cinnamus ii group, c AD 140-70. . 295R F US , 62 Form 37, Central Gaulish, with blurred panel decoration. The double medallion and kneeling stag (O. 1204A) are on a signed bowl from Barnsley Park villa by one of the Lezoux Paterni (Paternus iv), who signed moulds in the nominative. His style is quite distinct from that of Paternus v and his motifs suggest an earlier date and some connection with potters such as Sacer i. The caduceus (Rogers U103) also appear bowlstylen s so hi n s,i often heres a , multipln ,i e impressions othee Th . r motif probable sar y stylised tree Rogerf s(c , etc) sN2 A c , D 130-50 297. R F . US , 63 For , Centram37 l Gaulish, with four repeated panels: (a) A vine-scroll (Rogers M31) and lozenge (ibid, U36). (6 singl)A e festoon (?), wit hmasa . 683)k(D . Athlet) (c 403). e(D . (d) Dolphin baskea n so t (Rogers Q58). All the motifs, including the circles in (d), appear on bowls by the Cerialis ii-Cinnamus ii group. c AD 140-70. FR 298, US. 64 For , Centram37 l Gaulish, with panelled decoration distinctive th n i , e styl s f Cettueo Le f o s Martres-de-Veyre ovole Th . o (Rogers B263 smalld )an , double medallions with astragala n o e iar bowl from Corbridge (Stanfield & Simpson 1958, pi 143, 44) and the Apollo (D. 52) is on a bowl from Old Penrith (ibid 38). The scarf-dancer is not in Dechelette or Oswald. For the dating of this potter Hartlee se , y 1972 Ac D. 135-6034 , 445R . F . ,US 65 Form 37, with scroll decoration. The ovolo (Rogers B144), leaf (ibid, H22) and bird (D. 1019?) were all used at Lezoux by members of the Cerialis ii-Cinnamus ii group, c AD 140-70. FR 451, US. 6 6 For wit, m37 h windina g scroll ovole .Th o (Rogers B144 bir d . 1019an d)(D ) were use t Lezouda x by members of the Cerialis ii-Cinnamus ii group, c AD 140-170. FR 486, US. 67 This was described by Macdonald (1933, 258) and assigned to Libertus ii. Three large fragments fro mbowa forf Centra, lo m37 l Gaulish, wit freestylha e scene ovole .Th o (Rogers B114 uses )dwa by several Lezoux potter Hadrianie th n si Antonind can Libertuy eb periodst rareld no st an ybu , wit hwava y lin edouble belowth d eheres an a ,ridg , It . e belo decoratione wth , both suggest that the bowl is by Butrio. Nearly all the details are paralleled on his stamped bowls. The Cupid with torches (D. 265) is on a burnt bowl from the London Second Fire deposit (Stanfield & Simpson 1958 seatee th , 674, pd di60 )an Venu . 187) seated s(D an , d woma . 555n(D ) together wit hwarrioa r othero (Dtw n s. o fro613 e m)ar London (Oswald 1930 ; Stanfiel2 , , piV Simpsod& n 1958,, p57 i 655). The bird (D. 1041) is on a bowl at Cirencester. The probable Apollo and chariot (D. 59?) Herculed an 471. s(D ) have also been recorde himr philosophee dfo Th . figure th . 523n d eo r (D )an the left edge of the bowl (possibly D. 99) are known for Libertus ii but not, apparently, for Butrio. The small Hercules (D. 456) has not been recorded on stamped or signed bowls. The leaf (of the type Rogers G unstampen la 37-47 n o s i ) d bowl from Scol f Butrio'eo wite hon s other ovolos. These figure-typemoss hi t t commono e sar perhape n ar one d san s mainly associated wit wors hhi k using this ovolo. Thi Montane s th bow d lan s piece (59 amone )ar earliee gth r vessels from Rough Castle . 488R F . US , 68 Form 37, Central Gaulish. The festoon with widely-spaced border of beads (Rogers F37) is unusual and was apparently first used at Les Martres-de-Veyre, but also appears on an unsigned bowl at York with stylistic connections with Drusu . Botii s h Drusu Attianud an s i use i srosettee th d - tongued ovol damageg o right-han e eg wit th e hn th d o d side 125-4 D datA .A therefors e5c i e likely. Unregistered, . 1920US , 5 , 69 Form 37, Central Gaulish, with a large, leafy scroll. The ring-tongued ovolo suggests the Paternus v group, but the bird (D. 1038) was most commonly used by Cinnamus ii and does not appear elsewhere in Paternus's work or that of his associates. It is also noticeable that the bird in Paternus's scroll nearls si y alway peacocksa . Rogers assign lea e Attianu o srubbint th a f ) t (hibu sH7 f , gso ii bowle th f question so i e on n (Stanfiel Simpsod& n different1958e show) b o 26 t , , alst e psi i87 H . o attribute Rouge sth h Castle piec Cinnamuso et ovoloe vien th i t thihardlf , wn t o I . bu ,s so ca e yb MACIVOR, THOMAS, BREEZE: ANTONINE WALL FORT OF ROUGH CASTLE | 247 is more likely to be by one of Paternus's predecessors or associates, such as Laxtucissa. c AD 155-90. Unregistered, A, US. 70 Unregistered B Form 37, Central Gaulish, with a straight line beneath the ovolo, in the manner of Secondu Lezouxf o sv witd an ,hlargea , double medallion Ac , D 145-75. Unregistered. US , B , 71 Unregistered Form 37, Central Gaulish, with a basal wreath of four-petalled rosettes (Rogers C21) panea d lan wit hhara . 950A) e(D . Both appea stampen ro d bowl Quintilianuf so rosettee th , si s at Lezoux, the hare at London (Bethnal Green; Stanfield & Simpson 1958, pi 68, 4). c AD 125-45. Unregistered, US. 72 Form 37, with ovolo (Rogers B208) and conventional tree (ibid, Q42), used at Lezoux by Docilis i. c AD 140-65. Unregistered C, US. 73 Form 18/31, Central Gaulish. Presumably early Antonine, if the form is identified correctly. Recorde Macdonaln di d 1933, 258. The Potters' Stamps Superscript numbers stamp-listthe in 1. Stamp atteste pottere th t d a question n yi . t atteste potter2e No . th t da y in. question t othebu , r stamp pottee th f sro known there. All, with the exception of 74, are in the NMAS. 74 Cinnamus ii on form 37, Lezoux. Recorded by Macdonald (1931, no 171) as CINNAMI on a large label, this must surely be the usual large stamp CINNAMI retrograde (die 5B). The bowl seems not to have survived, c AD 150-80. US. 75 Do(v)eccu i s llfor n ao m 15/31 DOVIICCVS Lezoux. Form 15/3uncommon a s 1i n form made occasionall t Lezouya x (Oswal Pricd& e 1920, pi XLIII, stam43)t e recordeno Th y . s pi an n di certainly dated contexts, though two examples in the Chesters Museum are presumably from 's Wall. Do(v)eccus's decorated ware occurs commonl lato t ed Antoninmi n yi e contexts, possibly includin Wroxetee gth r Gutter (Atkinson 1942 scarceln ,, G9) pca 36 it I .earliee yb r than AD 160 and probably not earlier than AD 165. None has been recorded from Scotland. However, his plain samian does not necessarily start quite so late (cf the work of the associated potter Casurius ii, where there is reason to date some of his plain ware earlier than his decorated ware). Nevertheless, this single stam f Do(v)eccupo s fro latese mth f tScotlano piecee on ss d i recorde d fro Antonine mth e Wall Ac , D 160-90 304. R F . US , 76 Peculiari forn o mA 2 18/3 si 1 1ECVLIARI SLezouxF . Thi presumabls i s stame yth p reas da OFCVN Josepy Ib h Anderson (1905 ,bee s 492)ha n t I .recorde d froMartres-de-Veyres mLe t bu , not necessarily from the kilns. His fabrics suggest manufacture only at Lezoux. This stamp occurs t othea r Antonin materiae e th Wal n i s li l sitefrod second-centure sman th y fire t Verulamiua s m (c AD 155-60) and Worcester. It was used on a wide variety of forms, including 27 (occasionally), Ac D. 150-7580 d 7 9299an . R .F US , 77 Suobnu forn o ma 2 s 18/31 (SV)OBNI' Martres-de-Veyres MLe . Suobnus belonge latee th ro dt grou potterf Martresps o Le t sa .example n Therte e ear thif so s stam Antoninn pi e Scotlands Hi . associated forms, mainl , suggesy 18/327 d dat1a tan eAc D 130-55. Thi presumabls si y Andersen's stamp (1905, 492) . .NTM and Macdonald's (1931, no 131) (SECVN) DINI-M. Unregistered, US. 78 Tasgillus ii 9b on form 33 TASGILLIM. Tasgillus worked at both Les Martres-de-Veyre and Lezoux, but the fabric associated with this particular stamp suggest that it was used only at Lezoux. The die was made by surmoulage from an impression of die 9b. The original version appears at Camelon. His general record includes Birdoswald, Corbridge and Malton. A date c AD 130-45 seems likely for his activity at Lezoux. FR 301, US. Coarse Pottery (figs 7-10) pottere Th y examined include t onldno y that fro e excavationmth f 1957-61o s t alsbu ,o that recovered during later work on the site, principally in the main ditch and on the spoil heaps in the area pite olllia.r fth so Potter NMASe th n yi , from earlier excavations alss owa , examined t onlbu ,y typet sno adequately represente more th en di recent assemblage have been include thin di s report. The types present are amphorae, mortaria, flagons, narrow-necked jars, small flasks, handled beakers, poppy-head beakers, rough-cast beakers, cooking pots, bowl dishesd an smortarie Th . e aar discussed separatel Hartleys Mr fabrice y yb Th . s includ eligha t brown-gre amphoraee th r yfo , orange, j PROCEEDING 8 24 SOCIETYE TH F SO , 1978-80 red, white and fine grey among the flagons, orange Severn Valley ware, represented by one sherd, and black burnished 1 and 2. A very few cooking pots, bowls and dishes are in hard grey, white or coarse grey-brown fabrics. BB2 outnumbers BB1 in the ratio of 3 :1, both in the assemblage from the more recent excavations and in that in the NMAS. The Colchester are a- Thame e th f so estuard probabls an 2 ywa source BB yth e f mosth eo f o t poppy-head beakers. (Williams below p 256; Gillam 1973, 56). Most of the BB1 vessels came from Dorset, from the Wareham - Poole Harbour area, while some were made at Rossington Bridge, Yorkshire. A small amount of the BB1 sherds are from a third, as yet unidentified, source. (Williams below p 257). Severe Th n Valle produces yha shere don d (Webster 1977, 166) pottere Th . consistentls yi y Antoninen i date. The majority of the pottery was found in two areas - the western part of the N barracks in the western praetentura, westere th d nan intervallum road. followine Th g abbreviation usede sar : 1 BB Black-burnished categor (Gillay1 m 1960, 126-27). BB2 Black-burnished categor (Gillay2 m 1960, 126-27). Bar Hill 1975 Robertson, Scott & Keppie 1975. 1922 Miller 1922. Birrens 1975 Robertson 1975. Gillam Gillam 1970. Mumrills 1929 Macdonald & Curie 1929. Mumrills 1960-61 Gillam 1960-61. Mortarium-like Bowl *79 Rim sherd of mortarium-like bowl, with flat topped projecting rim, top reeded with three grooves; in hard off-white fabric with fine inclusions. NMAS. US ,

Flagons *80 Rim, neck, handle and body sherds of one-handled flagon with ring-neck, in fine grey fabric, smoott burnishedno t hbu . Fro maie mth n Mumrill f ditchc . US , s 1929 100.5g ,fi darn ,i ware d kre ; Mumrills 1960-61, fig 14.72, in red, self-coloured fabric, but interior different; Balmuildy 1922, pi XLIII, 3, in white or red fabric, interior different; Bar Hill 1975, fig 51.4, in pinkish-buff clay, with grey core. baso eTw 8 fragment1 finn i s e grey fabric, smoott burnishedno t hbu , possibl y. Fro par80 e mf th o t main ditch, US. *82 Rim fragment of flagon, in pale orange fabric. From W part of N barrack in W praetentura, US. 3 Rim*8 , neck, par handlf to pard shouldef ean to one-handlef ro d flagon fabricd re n i , . NMA 923R SF , US. *84 Rim and neck of flagon with ring-neck, in fairly soft white fabric. NMAS FR 925, US. *85 Rim and neck of flagon with grooved ring-neck, in orange fabric with cream slip, NMAS FR 926, US. *86 Rim of flagon with collared neck, in orange fabric with very slight trace of cream slip. NMAS FR 928, US. cf Balmuildy 1922, pi XLIII, 8. Also, in NMAS, rim of flagon in grey ware, US. Jars fragmentm ri o f narroTw o s 7 w*8 necked jar lighn ,i t brown-grey fabricsherdsm ri . Also o W . ,tw , similar in form and fabric. W. All from W intervallum road, on subsoil presumably antedating phase 1 road but possibly disturbed. sherm *8largf Ri do 8 r wit eja h turned-down flange blacn i , k fabric, lightly burnished. Fro parmW t obarracfN Wn ki praetentura, Birrenf c . US s 1975 66.2g fi , red, ,3 , grey core, Antonin 66.4g fi ; e,I grey, darker surface, burnished, Antonine I; Balmuildy 1922, pi XLIV, 1, mostly in fine red fabric t manbu 'blacn yi k fumed ware' r Hil;Ba l 1975 51.23g fi , vern i , y hard grey clay. *89 Rim sherd of Severn Valley ware jar with turned-down rim, in soft orange fabric. NMAS FR 874, Webstef c . US r 1977, 166.17. paln i sherr em ja pinkish-buf*9 f Ri do 0 f fabric with grey core. NMA 927R . SF ,US MACIVOR, THOMAS, BREEZE: ANTONINE WALL FOR ROUGF TO H 9 CASTL24 | E Unguent Flasks *91 Fragment of small flask or unguent pot in grey paste, with grey interior and exterior, on top of orange. Fro barrac parN m praetentura,W f o tW n ki . US 92 Body sher smalf do l flas unguenr ko t pot, with horizontal ribbin exteriorn go orange-ren i , d fabric with cream sli exteriorn po . NMAS Gillaf c . , AmDUS ,36 90-160, Gilla , AmD37 120-70. 93 Base of small flask in orange-red fabric. NMAS, US. cf Gillam 38, AD 150-220. Also, body sherd of similar flask, in orange-red fabric. NMAS, US.

Handled Beakers sherm *9 handlef Ri 4do d beaker with short, upright rim, witd an h junction between nec shoulded kan r sharply defined BB1n i ; , probably from Dorset Gillaf c . , AW .m D65 140-300. BB1n i sherds , m ri , probablo mm ,0 9 similaTw 6 5, diameter d 94 yan o rt fro0 5 msc Dorsetl Al . .W fro barracparN m W praetentura,f o tW n ki . US Poppy-head Beakers rimo *9 Tw bas,6 2 1 d 3e an bod y sherds, probabl samf o l yeal poppy-head beaker gren i , y fabric, with highly burnished exterior, silvery-gre d daryan k metallic grey-black; decorated with panelf o s barbotine dots. Two sherds from W part of N barrack in Vf praetentura, US; 14 sherds from W intervallum road, of which 8 from phase 3 road drain sealed by burnt material, 3 from layer sealed by burnt material and 3 from crude stone setting above burnt material; 1 US. cf Gillam 71, AD 150-200 though bigger. Probably produced near Colchester (Gillam 1973, 56). *97 Rim fragment of small beaker in unburnished grey fabric, similar in form to 96. From the lilia spoil heaps, US. Also, in NMAS, body sherd with panel of barbotine dots. US. Rough-cast Beakers *98 Three joinin sherdm gri f rough-casso t beake creamy-bufn i r f paste with greyish-brown surfaces inside and out; exterior burnished above zone of rough-cast decoration, cf Gillam 74, AD 120-60. 99 Another rim fragment, probably of rough-cast beaker, in white paste with black surfaces, similar fro 9 barrac9 par N m praetentura,W f & o t W for 8 n i 9 n k i . 98mo t ;. mm dia 0 US m10 *100 Two joining base and body sherds of rough-cast beaker in off-white paste with light red slip inside and out. From W part of N barrack in W praetentura, US, and from N intervallum road, phase 3 road drain Gillaf c , , AmD74 120-60. *101 Rim fragment, burnt, probably of rough-cast beaker, with dark-red core and black surfaces. From road between N and S barracks in W praetentura, US. cf Gillam 75, AD 130-80. bod 3 bas *10e d yeOn 2an sherds joiningl al , rough-casf o , t beake whiten ri paste, with orange-red slip. One from area between N rampart and intervallum road, US, and 3 from W part of N barrack in praetentura,W possibld an n i y contemporary with drip-trench surrounding ephemeral building erecte t beginninda occupationf go Birrenf c , s 1975 76.37g fi ,clay d ,re , black slip. US , 103 Base of rough-cast vessel with white paste and reddish-brown slip. Also, body sherd in same fabric, probably par samf o t e vessel. Both fro barracparN m W praetentura,f o tW n ki . US 104 Ten body sherds with rough-cast decoration, in white paste with dark-grey black slip, burnished above rough-cast zone. Seven from W part of N barrack in W praetentura, US, 2 from N intervallum road, phase 3 road drain, and 1 from W intervallum road, sealed by burnt material. 105 Five body sherds with rough-cast decoration in light brown-buff paste, with grey slip. From W intervallum road patcn i , f burnho t debri t anglsa e between rampar ascensus,d an t overlyinl gal road levels. 106 Body sherd with beginning of rim, in pinkish-grey paste, with dark grey slip, and another body sherd with beginnin rimf go pinkish-gren ,i y paste, wit interiod hre exteriord ran , lightly burnished above zone of rough-cast decoration. Both from W part of N barrack in W praetentura, US. 10NMASn I 7 bod1 bas ,4 d ye an sherd s with rough-cast decoration, PR 319, 320 934, . & 321 6 ,US 32 , Cooking Pots *108 Four rim sherds of one or more cooking pots with everted rims, in light brown paste with light brown interior dard an ,k grey-black exterior; much abraded . AlsoW . vera , y smal sherdm ri l , simila for n fabricri d man , dia frol intervallummAl m W . 110mm road patcn i , burnf ho t debrit sa 250 | PROCEEDINGS OF THE SOCIETY, 1978-80 angle between rampar ascensus,d an t overlyin l roagal d levels. Anothe sherdm rri , simila forn ri m in gritty-grey fabric with traces of grey-black burnish on top of rim. Diam c 110 mm. From the lilia spoil heaps, US. *109 Rim sherd of cooking pot with beaded everted rim, in hard grey ware. From N intervallum road, seale cobbliny db phasf go roade3 . W . *110 Rim sherd of cooking pot with slightly everted rim, and with 2 slight ribs beneath rim, in off-white fabric. From N intervallum road, associated with phase 3 road drain. *111 Rim sherd of cooking pot with short upright bead-rim. Fairly broad acute-angled lattice on the unburnished girth; the burnish on the shoulder and exterior of rim extends over the lip down to the inside junction of neck and shoulder. In BB1, probably from Dorset. From E part of N barrack in W praetentura, Mumrillf c . W . s US 1960-1. 3 o ,n sher*11m f cookinRi 2do t witgpo h shor t witd shallouprigho an h tw m wri t horizontal grooven o s burnished shoulder. Decorate unburnishen do d surface with vertical strokes BB1n I . . NMAS. US . fragmenm 11Ri 3 cookinf o t g pot, simila 111o witt rt bu ,h more pronounced bead rim BB1n .I ; diam 100 mm. From the lilia spoil heaps, US. 114 Rim fragment of cooking pot with fairly high, almost straight rim. Decoration as on 111, In BB1, probably from Dorset. Fro liliae mth spoil Gillaf heapsc . mUS , 119, AD 120-60. sherdm ri smalf o *11so Tw 5l cookin t witgpo h small rim, slightly curved witd an ,h fairly broad acute- angled lattice decoration on the unburnished girth. In BB1. From E part of N barrack in W praetentura,. W . US 116 Rim fragment of cooking pot similar in form and fabric to 115. Diam 125 mm. From the lilia spoil heaps, US. *117 Rim sherd of cooking pot with small, slightly curved rim, in BB1. From W intervallum road, US. W. cf Gillam 120, AD 120-60 but without the wavy line on the neck. 118 Rim of cooking pot similar in form and fabric to 117. From the lilia spoil heaps, US. *119 Rim sherd of cooking pot similar to 117 but with slight beading of rim, in BB1. From W intervallum road, phase 3 road drain, sealed by burnt material. W HMA. cf Gillam 120, AD 120-60, but without the wavy line on the neck. *120 Rim fragment of cooking pot with tall, almost straight bead-rim, and with the junction between nec shoulded kan r less visibly defined tha othen i cookin1 r typeBB f so g pots BB1n I . , probably from Dorset. From the lilia spoil heaps, US. W. cf Gillam 122, AD 120-60. *121 Rim sherd of cooking pot with thick bead-rim, with the junction between neck and shoulder sharply defined BB1n I . . Fro aree mth a betwee ramparnN intervallumd an t f c . W road . US , Gillam 124, AD 120-60. fragmen*12m Ri 2 f cookino t t witgpo h bead-rim, pointing outwards slightly. Decorated with wavy lin neckn eo BB1n I . . Fro maie mGillaf th c . n ditchW m . 125US , , AD 120-80. 123 Rim fragment of cooking pot similar to 122, in BB1, with wavy line on neck. From the lilia spoil heaps, US. W. sher*12m cookinf Ri 5do t witgpo h tall bead rim, inclined outwards slightly BB1n I . , much abraded. W. cf Gillam 127, AD 130-70. sherm 12 cookinf Ri 6do t similagpo 125 o r. Muct , diamm h0 abradedm14 notr o .B B ,f i difficul y sa o t t W. 125 and 126 from W part of N barrack in W praetentura, US. fragmenm 12Ri 7 cookinf o t t witgpo h upright rim, incline t slightlydou witd an ,hdistinc a t bulgo et necke th BB1n I . . Fro liliae mth spoi Gillaf c . l heapsmW . 129US . , AD 140-80 t withoubu , e th t wavy line on the neck. *128 Rim fragment of cooking pot with fairly short and sharply everted rim, In BB2. W. cf Gillam 137 AD 150-250. Also, another rim sherd of cooking pot, similar to 128, in BB2. Both from the lilia spoil heaps. *129 Small rim fragment of cooking pot with short and sharply everted rim. In BB2. W. cf Gillam 137, AD 150-250. From the lilia spoil heaps. *130 Rim sherd of cooking pot with sharply everted rim, in BB2. From W part of N barrack in W praetentura, Gillaf c . W m . 137US , AD 150-250. 131 Two rim fragments of cooking pots similar in form and fabric, diameters 125 and 150 mm. From the lilia spoil heaps. US , *132 Thre sherdm eri s comprising about three-quarter circumferencf so cookinf eo t wit bengm po hri t t fro shouldee ou mth distinca n i r t quarter round (cavetto) curve. Tight, slightly irregular lattice MACIVOR, THOMAS, BREEZE: ANTONINE WALL FORT OF ROUGH CASTLE | 251

FIG 7 Rough Castle: pottery (1 : 4) | PROCEEDING 2 25 S OF THE SOCIETY, 1978-80 decoration on unburnished girth. The burnish on the shoulder and exterior of rim extends over the lip to form a narrow band on the inside. In BB2. Two sherds from W part of N barracks in W praetentura, US, third from W intervallum road, US, W. cf Gillam 139/143, AD 150-250, 180-280. 133 4 rim sherds of cooking pots similar to 132, in BB2, diams 115 mm (W), 126 m ( x 2, W HMA), and 150 mm; 3 from lilia spoil heaps, US, 1 from W part of N barracks in W praetentura, US. sher*13m f Ri smal4do l cooking pot BB2n ,i , with tight acute-angled lattice - decoratioun e th n no burnished girth shoulderd . an Burnis 132n m o ri . s sa n Fro ho maie mth . n ditchW . US , *135 Eleven joining rim, shoulde bodd ran y sherd cookinf so t similagpo formn ri , fabri decoratiod can n to 132. Slightly irregular in shape, part of shoulder is indented. From area between N rampart and intervallum road, in post-hole relating to structure cut into rampart backing, and in gulley by S kerb ramparf so t relatin thio gt s structure frod intervallumm,an N road burnn ,i t material overlying phase 3 road drain, and from W part of N barrack in W praetentura, US. W HMA. 13 joinin3 6 g base sherds, whic belony hma abovgo t e vessel. Fro barracparmN W f to Wkn i praetentura, US. *137 Rim sherd of cooking pot similar to 132 in BB2. No decoration survives on unburnished zone. Another rim sherd, diameter 132 mm, probably part of same vessel. W. Both from W intervallum road, in patch of burnt debris over all levels. 138 5 rim sherds of similar cooking pots, in BB2, diams 122 mm (W), 132 mm, 120 mm (W), 150 mm (W) and 140 mm (W). 1 from W part of N barrack in W praetentura, US, 3 from area between N rampar intervallumd an t road f whico 2 ,yellon hi w clastoned yan f ramparso t backing1 d an , sealed below burnt material fro1 ; intervallummW road, phas roa3 e d drain seale burny db t material. sherdm ri f cookino s*13o Tw 9 t similagpo formn i r , fabri decoratiod can 132o nt . Fro mintervallumN road, associated with phase 3 road kerbs. W HMA. 140 Four rim sherds of similar cooking pots, in BB2, diams 150 mm (W), 167 mm, 170 mm (x2). Of these, 1 from N intervallum road, 2 associated with phase 3 road kerbs, 1 in burnt material overlying phas roae3 d drain, fourt. hUS 141 Three rim sherds of cooking pots similar in form and fabric to 132 diams }22 mm, 132 mm and 86 mm. From W part of N barrack in W praetentura, US. W. sherm 14 Ri f cookin2do t simila. Fro gpo forecourt Q m mm dia for H 2 n fabrii r 1 d 13 m12 an o ct , US. W. sherdm ri cookinf o so 14 Tw 3 g pots simila for n fabriri d man 132o ct praetoria,. a Fro. vi W e m . th US sherm 14 f cookinRi 4do t similagpo for n i fabrird man 132o . cFrot ,mmm parN dia E 0 f o mt15 barrac praetentura,W n k. i W . US 145 Rim sherd of small cooking pot similar in form and fabric to 132. From S barrack in W praetentura, US. W. 146 Thre sherdm eri f cookinso d an gm potm 8 s 15 simila , diamfor n 2 mm i rfabri d 1 m13 an 14 so c t . Fro12 intervallum6mmm W road make-un i , f phaspo roade2 phasn i , roae3 d drain sealey db burnt material, and in patch of burnt debris over all road levels. W. 147 Nine rim sherds of cooking pots similar in form and fabric to 132, diams 125 mm (W), 146 mm (W), . (Fro, 13x, m 154) 0liliam e 0m mm mm th 0 spoi12 , l10 heaps5mm . US , 148 Three rim fragments of cooking pots similar in form, fabric and decoration to 132, diams 127 mm (x2) and 130 mm. From the main ditch, US. W. sherm 14 cookinf Ri 9do t similagpo for n fabriri d man 132 o c. Frot are,e mdiamm th 1 am betwee12 n ramparN intervallumd an t . W road . US , 150 Rim sherd of cooking pot similar in form and fabric to 132, diam 132 mm. From N intervallum road, phase 3 road drain. W. 151 Six rim sherds, one joining body sherd and 17 other body sherds, all burnt, of cooking pot similar in form and fabric to 132, diam 152 mm. From road between N and S barracks in W praetentura, on gravel surface of road. W. Also, rim sherd of similar cooking pot, in BB2, diam 120 mm, from same road, US. 152 Rim and shoulder fragment of cooking pot similar in form and fabric to 132, diam 132 mm. From the E gate area, US. W HMA. 153 Rim sherd of cooking pot similar in form and fabric to 132, from area immediately outside N rampart. W . US , *154 Rim fragment of cooking pot similar in form to 132, but in grey fabric, smooth but not burnished. MACIVOR, THOMAS, BREEZE: ANTONINE WALL FOR ROUGF TO H 3 CASTL25 | E 155 Rim fragment of cooking pot similar in form to 154, in fairly hard light brown-buff fabric, diam 156 mm. Both 154 and 155 from the lilia spoil heaps, US. *156 Three rim sherds of cooking pot with short bead-rim, in BB2. From S barrack in W praetentura, HMAW . Gillaf US c . Manm& n 1970 2.14g fi , . *157 Rim sherd of cooking pot in BB2, similar to 156. From W part of N barrack in W praetentura, US. W. 158 Two rim sherds of cooking pot in BB2, similar to 157. From W intervallum road, in patch of burnt debris ove . roal ral W . d levelsUS d an , *159 Three rim sherds of cooking pot with inturned rim, in orange-brown paste with grey core, and with greyish-orange-brown lightly burnished surfaces. Fro liliae mth spoil heaps. US ,

Bowls/Dishes fragmen*16m Ri 0 bowf o t l witexternao htw l groove sdecorated beloan m wri d wit irregulan ha r wavy previousle linth n eo y burnished exterior wall BB2n .I . Fro barrac parmN W f praetentura, to W n ki Gillaf c . ManW m & . US n 1970. ,19 161 Rim fragment of bowl, similar in form to 160, and 2 body sherds, one with 2 grooves, the other wit hwava y line BB2 n fro.I o m.Tw lilia spoil heaps fro 1 , barrac parN m W f praetentura, to W kn i US. W. *162 Eight rim, walbasd lan e sherd shallof so w dish simila 160o rt , with plain rim externae ,on l groove below rim decorated ,an d wit irregulan ha r wavy line BB2n I . . Fro intervallummW road, sealed by burnt material, and from W part of N barrack in W praetentura, US. fragmen*16m Ri 3 plaif o t n rimmed dish witsimila0 h16 singlo rt e groove below rim BB2n I . . FromW part of N barrack in W praetentura, US. 164 Rim fragment of bowl with flat rim, and with slight bulge in upper part of exterior wall. In BB1. Diam 200 mm. From the lilia spoil heaps, US. W. cf Gillam 221, AD 140-80. *165 Rim fragment of straight-sided dish with short, downturned rim, and with acute-angled lattice decoration on the previously burnished exterior wall. In BB2. From the lilia spoil heaps, US. W. cf Gillam 222, AD 150-210; Gillam & Mann 1970, fig 2.22. fragmen*16m Ri 6 f bowo t l with downturned rim; fabri d decoratioan c s 165na f Gillac . m 222D ,A 150-210. similan 16 Te 7sherds m rri BB2 n whicf i , o 8 , h from lilia fro1 spoi , barracmS W) l 4 heaps( n k i , US , W praetentura, associated with drip-trench fro 1 barrac parN m d W f o tan , k W ,praetentura,in W US. W. fragmen*16m Ri 8 f straight-sideo t d bowl with short triangular rim; form, fabri decoratiod an c s na 165. From W part of N barrack in W praetentura, US. W. *169 Rim fragment of straight-sided bowl with short triangular rim; form, fabric and decoration as 165. From the via praetoria, US. W. 0 Fou fragment17 m rri s simila formn ri , fabri decoratiod can 169 no (W)t m .(W)m ,Diamm 6 m ,22 5 s21 145 mm (W) and 150 mm. From the W part of N barrack in W praetentura, 3 US, and fourth associated with burnt material fallen onto intervallum road. basd an e *17sherm Ri 1f dis do h with short triangular rim, simila formn i r , fabri decoratiod can o nt 165. Also, anothe fragmentm rri , possibl samf yo e dish . BotW . h fro intervallummW road. ,US fragmenm 17Ri 2 bowlf o t , simila formn ri , fabri decoratiod can 165no t . *173 Two joining rim fragments of bowl similar in form and fabric to 165. 174 Rim fragment similar in form and fabric to 165. 175 Rim fragment of bowl similar in form and fabric to 165. fro5 intervallumm17 172,173W d an 4 17 , road patcn ;i 172,17 4 burnf ho 17 d t 3debrian t anglsa e between rampar ascensus,d an t overlyin roal gal d levels 17d 5an , from phas roae3 d drain, sealed by burnt material. W. *176 Rim fragment of bowl with rounded downturned rim, similar in form, fabric and decoration to 165. Fro barracparN m praetentura,W f o tW n ki . W . US joinino * 17 Tw fragment7m gri bowf so l with slight basal chamfer, simila formn ri , fabri decoratiod can n to 165. From W part of N barrack in W praetentura, US. W. *178 Smal fragmenm lri f bowo t l simila for n fabrii rd man 165o ct . Fro barracparN m W f o tW n ki praetentura, US. W. sherm 17 bowf Ri 9do l simila formn ri , fabri decoratiod can 165o nt . Fro forecourtQ m. H W . US , R 254 | PROCEEDINGS OF THE SOCIETY, 1978-80 s-r V 186

I 162 \ V 163 A 188

\ 189 T V \

^ 193

194 \

\ 178

FI G8 Roug h Castle: potter) 4 : 1 y( MACIVOR, THOMAS, BREEZE: ANTONINE WALL FOR ROUGF TO H 5 CASTL25 | E 180 Rim fragment of bowl with rounded downturned rim, similar in form, fabric and decoration to 165, diam 222 mm. From the via praetoria, US. W. *181 Two joining rim fragments of bowl with downturned rim, similar in form, fabric and decoration to 165. sherdsm ri o 18, similaTw 2 181o rt BB2n .I Fro. .mai e Diammm th 0 n. m 20 ditchW . US , *183 Nine rim, walbasd an le sherd f disso h with short triangula basad an l m chamferrri ; similan i r decoration and fabric to 165. From E part of N barrack in Wpraetentura, US. W HMA. cf Gillam & Mann 1970, 22. *184 One rim and 1 joining rim and base sherd of bowl with slight basal chamfer and short, rounded downbent rim. Rather irregular in shape, and covered with a reddish-brown deposit, inside and out. In BB2. From area between N rampart and intervallum road, US. W. cf Gillam 222, AD ISO- 210. 185 Two rim sherds of dish with short triangular rim and slight basal chamfer, similar in form, fabric and decoration to 183, diam 207 mm. From the via praetoria, US, and from S barrack in W praetentura,. W . US *186 Thirteen rim, walbasd an le sherd disf so h with rounded downturned rim, simila formn ri , fabric decoratiod an 183o nt . Fro intervallummW road, seale burny db t material frod paran m, W f o t N barrack in W praetentura, US. W HMA. joinino *18 Tw sherd7m gri f bowo s l with downturned, hooked-over rim, with acute-angled lattice decoration with wide strokes previousle th n o , y burnished surface BB2n I . . Fro parN m W f o t barrack in W praetentura, US. W. cf Gillam & Mann 1970, 21. *188 Thre sherdm eri bowf so l with flat, slightly inward sloping rim, basal chamfe irregulad an r r acute- angled lattice decoratio previousle th n no y burnished surface BB1n I . , probably from Dorset. From S barracks in W praetentura, associated with paving interpreted as barrack threshold. W. cf Gillam 306, AD 120-60. *189 Rim sherd of dish on bowl with flat, slightly inward sloping rim. Decorated with widely spaced acute-angled lattice pattern on previously unburnished exterior wall. In BB1. From N intervallum road, in make-up of road surface. W HMA. cf Gillam 306-307, AD 120-60. 190 Seven rim sherds similar in form, fabric and decoration to 189. Two from S barrack in W prae- tentura, associatee on d with paving interprete barracs da k threshold, othe fillinn ri largf go e round pit (W HMA). Five from litta spoil heaps (1W). Rim and base sherd of dish similar in form and fabri 189o ct , with irregular zig-zag decoratio undersidn no . NMA mm f basedia eo m 5 S mRi 16 . 324. R F ,US fragmentsm ri o *19Tw 1 , probabl f samyo e bowl, with very short flat-topped rim; simila forn i rd man fabri 189o ct . Decorated with elongated acute-angled lattice pattern, with thick strokese th n o , previously burnished exterior surface. From W part of N barrack in W praetentura, US. W. sher*19m f bowRi 2do l with flat-topped rim, sloping outwards slightly. Decorated with acute-angled lattice pattern on the previously burnished exterior wall. In BB1, probably from Dorset. From N intervallum road, phas roae3 d drainGillaf c . mW . 307, AD 120-60. *193 Rim sherd of bowl with short flat-topped rim, sloping outwards. Decorated with slightly irregular acute-angled lattice pattern BB1n I . , probably from Dorset. Fro mintervallumN road, phas3 e road drainGillaf c . W m. 307, AD 120-60. *194 Rim sherd of bowl with short downturned triangular rim, decorated with oblique strokes on the previously burnished surfaces. In BB2. From S barrack in W praetentura, US. W. cf Gillam 310, AD 150-210. *195 Rim sherd of bowl similar in form, fabric and decoration to 194. Also, small rim fragment similar . Both mm dia 5 4 fro barrac m19 parN 14 m f W o t o t Wn ki praetentura,. W . US 196 Rim sherd of bowl with short triangular rim, similar in form, fabric and decoration to 194. From the lilia spoil heaps, US. W. *197 Rim fragment of bowl with short rounded downturned rim, similar in form, fabric and decoration 194o t HMAW . . sherdAlsom ri f simila5 ,o s r bowls. Fou f abovo r e sherds fro parN m W f o t barrack in W praetentura, US, 1 from W intervallum road, US, and 1 from lilia spoil heaps, US. fragmenm 19Ri 8 f straight-sideo t d dish with short triangular rim, decorated wit ha shallo w wavy previousle th lin n eo y burnished exterior BB1n I . . Fro .lilia e Diammm th 0 mspoi12 l heaps. US , W. cf Gillam 311, AD 150-210. *199 Rim fragment of bead-rimmed dish with external groove below the rim, and with widely spaced 256 | PROCEEDINGS OF THE SOCJETY, 1978-80 acute-angled lattice decoration on the previously burnished exterior. In BB1. From the tilia spoil heaps, US. W. cf Gillam 316/318, AD 125-60, 160-200. *200 Rim fragment of dish with very slight groove or indentation below the rim. In BB1, probably from Dorset. From the Ma spoil heaps, US. W. cf Williams 1977b, fig 1.10. *201 Ten rim, wall and base sherds of dish with slightly beaded rim, basal chamfer, and acute-angled lattice decoratio previousle th n no y burnished exterior wall. Looped zig-zag undersidn so basef eo . Slightly irregular in shape, bead-rim in particular varies in thickness. In BB1. From E part of N barrac Wn ki praetentura, HMAW . Gillaf US c . m 318, AD 160-200. fragmen*20m Ri 2 straight-sidef o t plaid dan n rimmed dish, decorated wit hshalloa w wave y th lin n eo previously burnished exterior BB2n I . . Fro mbarracS praetentura,W n ki fillin f largo g e round pit. W. cf Gillam 328, AD 150-200. 203 Five rim sherds similar in form, fabric and decoration to 202. Two from W intervallum road, one of which from patch of burnt debris at angle between rampart and ascensus, overlying all road othee levelsth rd froan , mlayea r above-mentionesealee th y db d burnt material sherdo Tw . s from W part of N barrack in W praetentura, US, and one from the Ma spoil heaps. Another rim sherd is simila t (W) 202 no o Bf rt i . r difficult B Also y ,bu sa oo t frot liliae mth spoil heaps. US , sher*20m straight-sidef Ri 4do plain-rimmed dan d dish BB1n I . , probably from Dorset. Fro maie mth n ditch, US. W. cf Gillam 329, AD 190-340. Also, three rim sherds of dishes similar in form and fabric to 204, from the Ma spoil heaps, US. W. *205 Rim fragment of bowl with beginning of handle, in hard off-white fabric, lightly burnished, with many fine inclusions. From W part of N barracks in W praetentura, US. *206 Body sherd of bowl with beginning of handle, in hard off-white fabric, with many fine inclusions; lightly burnished. *207 Smal fragmenm ri l f bowo t harn i l d off-white fabric with many fine inclusions o slightlTw . y irregular grooves below rim, on exterior. fro7 intervallumm20 N 20d 6an road, phas roae3 d drain. The Black-burnished Ware D F Williams, Department of Archaeology, University of Southampton tere mTh 'black-burnished' war uses ei d sens e herth Webstef n eo i r (1969Gillad an ) m,5 (1970, preface), describo t eranga cookinf eo associated gan d vessel distinctiva n si e fabric. The following abbreviations are used: G Gillam (1957). GM Gillam & Mann (1970) and the second century types of black-burnished ware illustrated in figs 1 and 2. A macroscopic examinatio mads larga nwa f eo e representative sampl f black-burnisheeo d ware from Rough Castle additionn .I sherd2 1 , s were selecte heavr dfo y mineral analysis (Peacock 1967) Tabld an , e show1 resulte sth termn si non-opaquf so e minerals. In the sample, BB2 vessels easily outnumber those in BB1 fabric, which suggests an Antonine II date (unles undividen a n si d Antonine deposi. t?) All the BB2 sherds are well burnished and have received a slip, the surface colour is generally either blac pinkish-greyr ko sherdo ,tw thougr so have hon e clearly been expose oxidizinn a o dt g firing. The fabric of these BB2 samples is reasonably homogeneous, which suggests a single source for all the sherds. Thi sheava seemy borne b b y t o minerast eou l analysi eighn so t sherds (Tabl 1-8s no )e 1 whic h produced an assemblage characterised by a high tenor of zircon combined with almost equal amounts of tourmaline and garnet, and a moderate amount of rutile and kyanite, and which agrees well with an analysi larga f so evessel 2 grouBB sf p o show havno t e been mad t Colchesteea r (Williams 1977a, Group XII). addition I Rougno t h Castle, earlform2 yBB s have been identified from Newstead (ibid), Mumrills (Williams 1977b) and (Williams forthcoming), and it seems likely that Colchester was supplying nortfoune 2 th t thihBB n da i e s mostt allth time f no ,o f i ,. sherd1 BB se whicTh h sample makth p enumbeu a e n appeai e differenf b ro o rt t fabrics, suggesting several centre productionf so . However fro 1 Wareham-Poole mth BB , e Harbour are Dorsef ao t seems to be in the majority. The fabric of these latter sherds is black, or very dark grey throughout, and contains MACIVOR, THOMAS, BREEZE: ANTONINE WALL FORT OF ROUGH CASTLE 257 TABLE 1

No Type s£££||3£l* 1 183 Bowl/dish G.222/G/M.22 72-6 10-1 2-4 4-3 — 1-0 9-6 — — 502 2 186 Dish, G/M.22 74-7 8-6 3-9 3-5 0-8 0-9 6-8 — 0-8 438 7 Dish19 3 , G.310 71-8 9-5 2-6 5-7 8 46 0-6 — 1-2 — 8-6 4 156 C-P, G/M.14 76-1 6-3 3-2 4-0 0-9 1-3 8-2 — — 720 5 152 C-P, G.139/143 70-8 11-2 2-8 3-8 0-3 0-8 9-5 — 0-8 572 6 135 C-P, G.139/143 77-6 9-5 2-9 2-4 — 2-1 5-5 — — 480 3 C-P13 7 , G.139/143 75-3 9-3 2-3 3-6 0-6 1-1 8 — 67 7-0-44 8 139 C-P, G.139/143 73-6 8-7 3-9 5-7 0-6 0-9 6-6 — — 347 9 189 Dish, G.306/307 42-9 49-7 1-8 2-6 0-9 — — 0-4 1-7 349 1 20 Dish 0 1 , G.318 51-3 44-6 1-7 — 1-0-43 2 — 47 0-0-52 11 190 Dish, G.306/307 78-8 10-7 — — 0-8 — 9-7 — — 614 12 119 C-P, variation G.120 61-0 8-6 6-9 3-4 2-9 2-7 12-5 0-6 1-4 527 a considerable amount of quartz sand temper. Two samples were analysed as being likely Dorset fabrics (Tabl 10)botd & 9 an ,h es produce1no , assemblagn da e wit hhiga h teno tourmalinf ro garneto n d ean , identica thos vesselo lt 1 eBB s associated wit Dorsee hth t centre (Williams 1977a, Grou. pI) Also represente t Rougda h Castl smala e ear l numbe sherdf ro s which appea producte b e o rt th f so serie kilnf so t Rossingtosa n Bridge, know havo nt e bee nsmallea makinn o 1 r scalgBB e than Dorset from abou middle secone th t th f eo d century (Farrar 1973 William; ,95 s 1977a) predominane .Th t colour appear dare b ko st grey cooking-pote th , s havin gsligha t 'lustre themo t ' ; less sand tempe uses ri d than is normal instancer ,fo Dorsen i ,sherd e th f t o sBB1 e thoughOn . t likel havo yt e been mad t Rossingtoea n Bridge (Table 1, no 11) produced an assemblage close to those obtained from 'wasters' from the kilns at Rossington Bridge, wit hfairla y high amoun garnef o t t present (Williams 1977a, Group II). Small quantitie likelf so y Rossington Bridg hav1 eBB been recognise writee th Antoniny n di rb e context t Balmuildysa Hillr Ba , , Castledyke Bearsdend san possibilite Th . y exists that thesware1 eBB s from Rossington Bridge owed their conception to a potter, or potters, from Dorset moving closer to the important northern military market, since there are similarities of form and decoration between the products of both centres (Farrar 1973, 95; Williams 1977a). In this connection, such a proposed move from Dorse Yorkshiro t wely elma have taken plac direca s ea t change resulth f frontiersf o te o , from Hadrian's Wall to the Antonine Wall, and the resulting competition from BB2 pottery. If this is so, then comparativele th y small amoun Rossingtof o t Antonine th n n Bridgo 1 e WalehavBB y elma been partly due to the early abandonment of the latter (Hartley 1972), before the BB1 from the better situated Rossington Bridg t properle abls sitge o ewa t y established. Another, smaller, centre, as yet unknown, was also supplying BB1 pottery to Rough Castle. In previous analyses a small group of vessels from Birdoswald, Mumrills and Bearsden has been isolated from the two main BB1 centres mentioned above (Williams 1977a, Group III; Williams forthcoming). A heavy mineral analysi cooking-pota n so , G120 t lackin,bu wave gth y line decoratio nece th kn no (Tablo 12)n , ,1 e identica sampleo t l s from Mumrill Bearsdend an s , produce dsimilaa r assemblage, suggesting the same origin. Mortaria Katharin HartleeF y Fabrics Fabric I Mancetter-Hartshill potteries Warwickshiren i 10D 0A -37 c 0 + fine-grainedA , whit creao et m fabric usually with very little tempering. Mortaria madee herth n ei first half of the 2nd century are fairly soft but increasingly in the second half of the century they were 258 | PROCEEDINGS OF THE SOCIETY, 1978-80 fired to a harder, smoother texture. The normal trituration grit after AD 150-60 was a dark greyish brown and/or red-brown material; before this datespecialld ean y befor t eoftei A0 Dn 14 included quartz. Fabric 2 Colchester Ac D 140-200 (Hull 1963) This fabric may be various shades of cream, but is often yellowish; it usually contains a fair amount of flint and quartz tempering. The trituration is usually flint with a little quartz. The mortaria in this fabric were probably never fire hars da s thosda e mad t Mancetteea r 150-6aftenorman D i A t r 0bu l circumstances they survive very well e majoritTh . f thoso y Scotlandn ei , however, have deteriorated beyon groune d th belie f whed o d t fan continu nou disintegrato et e into tiny fragments. Thi probabls si y due to the high calcium content combined with burial in an acid and damp soil. This is the main reason for the high number of unidentified stamps from herringbone dies. Fabric 3 Workshop ofBellicus t Corbridgea Ac D 155-85 slightlA y abrasiv t fairlebu y fine-textured fabric with very fine temperin f quartgo othed zan r material coloun i t varie I .t lo r sa fro m cream wit hgreyisa h cor bufo et f wit orange-pinn ha k cord ean sometimes seems to have a buff slip; red-brown, pale brown, quartz and black trituration grit. It normally survives well but no 264 has deteriorated badly. Stamped mortaria from Rough Castle *20 flangA 8 e fragmen brownish-bufn i t f fabric which probabl slia pd thougyha traceo hsurviven w sno . The fragmentary stamp is from one of at least five dies used by Austinus (see Robertson 1975,239, stama r fo p, fro10 mo similaidenticat a n , no fi t g84 rbu l die) completo .N e impressio thif no s stamp AVStN is known but other examples permit the reading .,. . , for Austinus manu. Austinus was the MANT only mortarium potter who undoubtedly used manu in his stamps though it was a common enough practice among Gaulish samian potters. Fifteen stamps of his are now recorded from sites in Scotland: Balmuild stamps3 y( , perhap vessels)o stw Hillr ;Ba ; Birrens; Camelon (2); Carzield (2); Cramond; Durisdeer; Milton; Mumrills; Newstead Roug(2)d ,an h Castle. Ninetee recordee nar d from NW England: Ambleside; Birdoswald; Cardurnock (2); Carlisle (6); Chesters Museum; Corbridge (3); Lancaster; Low Borrow Bridge; Maryport; Stanwix; and the Tullie House Museum, Carlisle (provenanc eEngland) unknowW N n i likelt e .b nbu o yt There are, however, incomplete stamps from two other dies probably belonging to Austinus, which link him with the pottery workshops at Wilderspool (Hartley & Webster 1973, fig 8, N & P). Stamps from these dies are on mortaria which can be attributed on fabric and form to Wilderspool and none have been found N of Hardknott. The likelihood is that his activity began at Wilderspool, where he was one of the more successful potters and that he later moved to the Eden valley. distributioe Th stampf no s fro five mth e certain dies clearly indicates that thei overlappee rus e dth Antonine occupatio Scotlanf n o occupatio n a d dan Hadrian'f no s Wall, probabl vien i f wfirse e o yth on t probabls hi e activit t Wilderspoolya hige hTh . proportion Scotlann i % 4 4 strikings di f o . Activite th n yi Eden valley c AD 130-60 would fit all the factors and if, as suggested his initial work was at Wilderspoo! perhapy ma e h s have started ther Aec D 115-20. From road between N and S barracks in W praetentura, US. *209 Two large joining fragments from a mortarium in fabric 3, which has been rivetted. (See also nos 262-4 which are from the same workshop.) The very fragmentary stamp is from the single die used by Bellicus. Stamps of his have been noted in Scotland from Mumrills; Newstead; and Rough Castle; and in England from Benwell; Chesters Museum; Corbridge (37); and Great Chesters. The distribution of his stamps clearly show that he was working at Corbridge where mortari knowe aar havo nt e been seconmade th n ei d century. face tTh tha stamps hi t s occu botn rAntonino e hth Hadrianid ean c frontiers again shows thas hi t activity overlapped a change of frontier, the small number in Scotland perhaps indicating that the greater part of his activity took place when Scotland was not occupied. Several of his mortaria were found in the Corbridge destruction deposit, now dated c AD 180 (Richmond & Gillam 1950,190-94), and some in the deposit immediately earlier (information J P Gillam). Moreover, Bellicus used rim forms which are most unlikel earliee b yo t r tha decade nth e AD 150-60 mortaris ; hi characteristi o s e aar c thaoftes i t ti n possible MACIVOR, THOMAS, BREEZE: ANTONINE WALL FOR ROUGF TO H 9 CASTL25 | E to attribute unstamped fragments to his workshop. Moreover, this workshop undoubtedly made un- stamped mortaria like no 262; this was an unusual practice and the most likely explanation is that he was activtime th et e a whe n potters were ceasin stamo gt p their mortaria datAc A . D f e o 155-8 5 wouls hi t dfi activity very well. NMAS unregistered. slightlA *21 . 0y mm Diaburn0 25 m c t mortariu mcrean i m fabric tempered with fine gritincome Th . - plet ediee stamCricf th sworke o o f o fro os pwh i e t mSoutdon a h Carlton, near Lincoln (Webster 1944, 129-43) fabrie Th .f thi co s mortarium differs knows fro hi rese mf th o t n product beinn si g slightly darker and coarser grained but this is not thought to be significant (Hartley & Richards 1965, 33, no 84). Stamps of Crico have been noted in Scotland from Balmuildy; Bothwellhaugh; ; Rough Castle, and Newstead (2); and in England from Birdoswald; Chesterholm; Claxby, Lines; Corbridge (4); Lincoln; South Carlton (5); and the Yorkshire Museum (1 + 1 ?). Two of the Corbridge mortaria are from the destruction deposit. His rim-profiles fit best with an early to mid-Antonin Birdoswale eth datd ean d stamp could well belon data o gt e after AD 158. Activity withi perioe nth d AD 140-80 seem sonle indicatedth ys Soutwa e hH . Carlton potte a hav o rt d eha trade in Scotland; judging from his stamps the army in Scotland took more than a third of his output. In NMAS FR 339, US. (Macdonald 1934, 456, IRICO.) *21 . Fou1 mm r Diafragment0 29 m c severad an s l tiny pieces, probabl l fro ysame al mth e worn morta- rium in fabric 1 (fabric unusually soft). The two incomplete stamps are from the single die of Doballus. Stamps of his are now known from Duston, Northants (2); Rough Castle; Totternhoe Villa, Beds Verulamiumd an ; . Althoug Rouge hth h Castl Verulamiud ean m mortaria certainly appear to be in the fabric made in the Mancetter-Hartshill potteries in Warwickshire (Fabric 1), othee th Northamptonshira r thren weli y e eb lma e fabric distributione th ; , slight thougs i , is t hi unusual for a Warwickshire potter. It is worth holding the possibility of migration in mind but more examples of his work are needed to clarify his career. His rim-profiles would fit with manu- facture within the period AD 140-80. From N intervallum road, on phase 1 road surface. *212 A slightly burnt flange fragment in creamy buff fabric with pinkish core and pinkish buff slip, with fine blackish, red-brown and quartz tempering in the clay. The two retrograde stamps, impressed close together, read FEC and are counterstamps of a potter whose namestamps read EMI, presumabl Emianur yfo Emiusr so . Twenty-seven stamped fragments from between 21 and 27 mortaria are known for this potter, all from Scotland: Ardoch (2); Balmuildy (3); Bothwellhaugh; Camelon (5); Castledykes (2); Inveresk (2); Newstead Kilpatric(9)d ;Ol k (3)Rougd ;an mortarie h th Castle l al fro e s aar m A . Newstea sited dan s N of assumeNewsteae b y ma d t di tha workshos hi t Scotlandn i s pwa , probabl Newsteae th n yi d region. The distribution of his work and the rim-profiles support an Antonine date c AD 140-60/65. His rims can be clearly paralleled in the work of DIS/LDB, Docci(us) ?, and to a lesser extent that of Austinus (see no 208), who probably worked in the Hadrianic-Antonine period in the Eden valley near Carlisle. It seems quite likely that he had, perhaps in early life, worked with one of them. (For a more detailed discussion of EMI's work see Hartley 1976, 82-3). In NMAS unregistered, 1920; US. (Macdonald 1934, 456, FEC retrograde.) *213 Diam 330 mm. A burnt mortarium in fabric 1 (the trituration grit includes some quartz). The poorly impressed retrograde stamp (ICOTAScf t leasa f o tfro s sevei )e mon n dies use Icotasguy db o swh worke e Mancetter-Hartshilth n di l potterie Warwickshiren i s s stampHi . s have been founn i d Scotland from Ardoch; Balmuildy; Castlecary; Newstea Rougd dan h CastleEnglann i d d an ; dan Wales from Aldborough; Ambleside; Ancaster; Cardurnock (2); Chesters Museum; Cirencester; Corbridge (4); High Cross (3); Horncastle, Lines; Leicester (9); Lincoln (2); Little Chester (2); Mancetter (13); Manchester; Margidunum; Maxey; North Collingham; Rocester (3); Rossington Bridge; Shenstone, Staffs; Stanground South, Hunts; Stanton Low, Bucks; Templeborough; Tallington, Lines; Tripontium; Usk; Wall; Wilderspool (3); Winterton; Wroxeter; and York likelfro3 e b ( 1m& o yt York). Icotasgu fairla s ysha typical distributio Mancettea r nfo r potter workin Antonine th n gi e period. mortariMans hi f yo a show pre-Antonine characteristic trituratioe formth m d ri e an s th nn si gri t used datAa c Df d e o 130-6 an 0 would wors fithi k well NMAn .I 337R S. (AndersoF ;US n 1905,492, ICOTVS; Macdonald 1934, 456, ICOTASI retrograde.) 260 | PROCEEDINGS OF THE SOCIETY, 1978-80 *21 tinA 4y flange fragment fro mmortariua fabrimn i c 1. More complete example fragmentare th f so y stamp read LOCCI-PRO, retrograd Locciur efo s Pro(culus?). Mortari havs hi ef ao bee n noten di Scotland from Balmuildy (4)Hillr ;Ba ; Birrens; Mumrills; Newstead Kilpatrickd ;Ol Rougd ;an h Castle; and in England from Alcester; Aldborough; Ambleside; Benwell; Binchester; Kenchester; Leicester (2); Little Chester; Mancetter (5); Orton Longueville, near Peterborough; Ribchester; Sawtry, Hunts; Shenstone, Staffs; South Shields; Thistleton, Rutland; Wall; and Watercrook.

This potter's activity is dated primarily by the stamps on the Antonine Wall, and at Benwell and South Shields where they must be dated earlier than AD 140 or later than AD 158. There is also evidence to show that he worked at one time with lunius Loccius, probably a mid-to late-Antonine potter. A date withi perioe nth d AD 135-65 wouldatine th t dfi g evidenc type th rim-profilf eo d ean e used. worts i t I h noting that althoug totas hhi l output seem havo st e been much smaller than thaf o t Icotasgus, his sales in Scotland were greater. It seems likely that this difference is due to a slight difference in date. It is, in fact, quite possible that Loccius Pro- did not begin his career until c AD 140. He worked at Mancetter wher activits ehi y overlapped with tha f otheo t r potters sharin name gth e Loccius, Loccius Vibius who sold mortaria in Scotland, and lunius Loccius, to whom he was surely related in some way. n NMAI S unregistered . (MacdonalUS , d 1934, 456, LOCCI.bb not& . , e1)

*215 Diam c 330 mm. A mortarium in fabric 1 with two retrograde stamps impressed close together five th e f dieo frof Minomeluse so mon s stampHi . s havbeew eno n recorde Scotlann di d from Mumrills; Newstead d Roug(2)an ; h EnglanCastlen i d an ; d from Aldborough; Ambleside; Catterick; Corbridge (7); Fisherwick, Staffs; Gloucester; Hartshill (several); Heronbridge, Ches; High Cross; Leicester (7); Little Chester; Mancetter (several); Margidunum; Rothley; Shangton, Leics; Shenstone, Staffs; Wall; Water Newton; Willington, Derbyshire; Wroxeter (7); Yorkd an ; crossine th Watlinf go g Stree Ryknield an t d Street.

kilA Minomeluf no bees sha n excavate Hartshilt da l (unpublished) presence ;th larga f eo e number of stamp Vitalif so Ac ( D 115/120-45)V sI havy ma e o beewh ,n sharin kilne gth , suggestt no s thawa t i t later than AD 145. A date of AD 130-60 might well cover his activity. In NMAS FR 338, US. (Anderson 1905, 492, VINON ; MacdonalIF d 1934, 456, MINOME retrograde.) *21 flangA 6 e fragment fro mmortariua fabrimn i . Therincomplete2 c1 e ear , retrograde stampf o s Mossius, one of whose dies was found at Hartshill where he worked although none of his kilns have been discovered s stampHi . knowe ar s n from Birren Rougd an s h Castl Scotlandn ei d an , Ancaster; Aunsby, Lines; Corbridge (2); Duston, Northants; Gargrave, Lanes; Hartshill (die); Leicester (3); Lincoln; Mancetter; Margidunum; Rocester; Tripontium; Verulamium; Wall (2); and Wroxeter in England. The rim profiles used by Mossius would best fit a mid- to late-Antonine date which might perhaps accounsmale th r l fo numbet stamps hi f o Scotlandrn si NMAn I . S FR 341, US. (Anderson 1905, 492, IOSSIA; Macdonald 1934, 456, OSSIA (SS retrograde).) mortariu*21A . 7 Diamm 0 mfabrin mi 29 wit1 c hvirtualla y complete stamp reading SAR.RIr Ffo Sairius fecit. Sarrius worke Mancetter-Hartshile th n di l potteries wher t leaskilnea o tw ts usey db him have been found. He was perhaps the most important moratarium potter of the second century and is especially notable for having opened a second workshop at Rossington Bridge, near Doncaster while continuin Warwickshirs hi n ru go t e mighones A .expecte e tb d stamps morhi f eo s than of any other potter have been found in Scotland, 12 on mortaria in his midland fabric (Ardoch; Balmuildy Hilr Ba ;l (2); Birrens (2); Camelon; Carzield; Mumrills; Newstead; Rough Castle; and Wilderness), and 16 in his northern fabric (Balmuildy (2); Bearsden (12 stamps, at least seven vessels); Birrens Camelon)d an ; . Sevent s midlanhi f yo d mortaris eighd hi aan f o t northern ones have been noted from England relative Th . e number f midlano s northerd dan n mortaria in Scotland may be a little unbalanced by the discovery in recent years of 12 Rossington Bridge mortaria at Bearsden (information David Breeze). The large number of stamps in Scotland leaves no doubt of Sarrius' Antonine date. Only one of his stamp notes si d from Hadrian's Wall, from Birdoswald (Richmon Birled& 187n o y , 2 1930,188o n , 13 g ,fi described as an illiterate stamp in Period la, AD 125-40), but a stamp from Verulamium is recorded from a deposit datedc AD 155/60 (Frere 1972,378, no 34). Furthermore, none of his stamps have been recorded MACIVOR, THOMAS, BREEZE: ANTONINE WALL FORT OF ROUGH CASTLE | 261 from Pennine fort sals y reoccupie notee ob ma t i d d thaan dAtc 0 D althougstamps 16 hi f knowe o sar 1 h2 n from Corbridge none is recorded from the destruction deposit now dated to c AD 180. It may reasonably be assumed that his activity should be dated c AD 135-65/170. In NMAS FR 336, US, and joining fragment from 1958 excavation from W part of N barrack in W praetentura, US. (Anderson 1905, 492, SAR R.IE; Macdonald 1934, 456, SARR.I.F). *218 Possibly the stamp referred to in Anderson 1905, 492, as DVRS. .IA, but the reading is quite uncertain wall-sideA . d mortariu msandn i y textured orange-brown fabric with grey cord an e cream slip: no trituration grit survives but the fine tempering in the fabric includes flint and red- brown grit with some mica. In NMAS FR 340, US. *219 Diam 290 mm. Another wall-sided mortarium burnt to a slightly purplish red with remains of a slip NMAn I . S unregistered. US , Bot f thesho e mortaria have stamps impressed alon walle gth , whic almose har t certainle th y yb same potter but exfoliation of the surface makes it impossible to identify the potter or even to read the stamps witconfidencey han , though further example helpy sma . They have previously been attributed luniuo t s Loccius (Birle Gillay& m ) probabl1948D , e , 27 th 182for yo m d n ri ,mbecaus an e th f eo knowpositiow stampe no th ns i f thant o I . t othe quitw fe rea potters , including Sarrius, occasionally followed these practices. Thes mortario tw e a could have been mad t Rossingtoa e n Bridge though manufacture in Scotland is not impossible; they are undoubtedly Antonine in date, c AD 150-80. Mortaria in Fabric 2, with potters' stamps of herringbone type (Hull 1963, fig 60) stampe Th s liste probable dar differenn o l yal t vessels. *220 Hull no 30. Form as no 221. In NMAS FR 343, US. *221 Diam c 360 mm. About a quarter of a partly burnt mortarium with a stamp surviving to each side of the spout. Hull no 30. From the Antonine Wall ditch, US. 222 Diam c 340 mm. Burnt to black throughout. Hull no 30. Form as no 223. From the Antonine Wall ditch. US , *223 Hull no 30. From W part of N barrack in W praetentura, US. 224 Hull no 37. In Museum, US. *225 Diam c 310 mm. Hull no 33. In NMAS unregistered, US. NMAn I . *22 Hul. 33 S6o unregistered ln mm Dia0 34 m c . US , *227 Fro. Hul38 Antonine mo th ln e Wall ditch. US , *228 For. 235 o NMAn n Hul38 I . ms o a ln S unregistered. US , *229 Hull no 38. Form similar to no 242. In NMAS unregistered, US. *230 Diam 300-2. Stamp . Hul38 o 0mm n ls surviv boto et h spoute sideth f so . 227Foro n n I .s m a NMAS unregistered, US. NMAn I *23 . . Stam2 1S C mm Diaunregisteredp0 K/ 30 m c . US , *232 Stamp K/C 3. In NMAS, unregistered, US. *233 Diam c 360 mm. Stamp K/C 4. In NMAS unregistered, US. 234 Stamp attribute Colchesteo dt t identifiableno t rbu . For236o n .ms a Fro mgatE e area. US ,

Nos 231-3 are in fabric 2 and have stamps from three different dies (K/C 2-4 on Table 2). No stamps from these dies have yet been found at Colchester or on any kiln-site. The fabric used is visually identical with that produce t Colchesteda therreasod o n the an ry s ei y nwh shoul t havdno e been made there but, as the rim profiles are not among the outstandingly typical Colchester forms, the possibility of manufactur Kenn ei t canno rule e maie b t th d nf outo outletColchestee e .th on r Ken s sfo wa t r potters but there is some evidence that some herringbone dies were used in a workshop or workshops in Kent, somd an e suitable clays were available. Non thesf eo e workshop knowe sar wid a hav o nt d ee ha market . The distribution in England of the stamps from the dies K/C 1-4 (Table 2), merely permit us to say that the mortaria were made either in Kent or Colchester. herringbone Th e stamp mortarin so a supplie Scotlano dt d from potterie south-eastern si n England, possibly all from Colchester, are from only 10 dies (Table 1). Nevertheless, the situation is rather more complicated than it appeared in 1963 (Hull 1963, 114-15). Although there is no reason to doubt that the bul thef ko m came from Colchester ther possibilita s ei camy w thafe ea tfro m Kent 231-s (seno ed 4an comments) t shoulI . d als addee ob d tha tformerl s Regaliswa o ywh , though havo t t e worked onlt ya 262 | PROCEEDINGS OF THE SOCIETY, 1978-80

o D.

I O

o\ eO MACIVOR, THOMAS, BREEZE: ANTONINE WALL FOR ROUGF TO H CASTLE 263 TABLE 2 'Colchester' mortarium stamps foun Scotlann di d Overall Huls lno total r^iitvu e % 30 5 34 2 8 3 3 7 3 Attrib K/C K/C K/C K/C Total stamj Colchester Ardoch 1 — — 1 1 — — — — — — 3 12 25 Balmuildy — — — — 1 — 1 — 1 — 1 4 31 12-9 Bar Hill — — — — — — 1 — — — — 1 20 5 Barburgh Mill — — — — — — — — — — — 0 1 Bearsden — — — — — — — — — — — 0 15 Birrens — — — — — — 1 — — — — 1 16 6-3 Bishopton j _ — — — — 1 — — — — 2 3 Bothwellhaugh 1 — — — — — — — — — — 1 4 Cadder 1 — 1 — — — 2 — 1 — — 1 5+ 10 60 Camelon 6 — 3 — 2 1 1 _ j — 1 15 + 1 32 50 Cappuck — — — — — — — — — — — 0 4 Carzield — — — — — — — — — — — 0 3 Castlecary 1 — — — — — 1 — 1 — — 3 6 Castledykes — — — — — — 1 — — — — 1 4 Cramond — — — — 1 1 — — — — — 2 6 Croy Hill — — — — — — — — — — — 0 1 — — — — — — — — — — — 0 2 Durisdeer — — — — — — — — — — — 0 1 Inveresk 1 — 2 — — — 1 _ j — — 5 14 35-7 — — — — — — — — — — — 0 1 Lyne — — — — — — — — — — — 0 1 Milton — — — — — — — — — — — 0 1 Mumrills 6 3 — 2 1 — 1 — 1 1 1 16 36 44.4 Newstead | __— — — — 1 — — — — 2 52 3-7 —— —— — — 2 — 2 1 1 — — 6 18 33-3 Rough Castle 4 _ — 24 1 1 — 1 1 1 15 27 55-5 Strageath i — 1 — — — — — — — — 2 2 Wilderness — — — — — — — — — — — 0 2

24 3 12 1 12 2 15 8 1 2 4 84 + 2 325 26-46 % Stamps in 38-7 50 70-6 12-5 66-6 Scotland 75 9-1 50 61-5 66-6 (kilns excluded) Exactly 50% of all known mortaria stamped with the above herringbone type dies were sold in Scotland and this per- centag markedls ei y affectenumbee th y b d r sol Colchesten di immediats it d ran stamp4 e8 vicinite th s f t leaso y(a 5 2 t from English sites). These were not the only herringbone dies used at Colchester, etc, but few others are recorded in any number away fro kilnse mth . Allowin possiblr gfo e difference extene th excavatiof n i so t differenf no t site t seemsi s likely that more than a third of all the Colchester mortaria with herringbone stamps were sold in Scotland. Notes 1. 'Colchester attrib'. Many stampattributee b n sca d with certaint cannoColchestee o yt di identifiee e b tth t rbu d due to their fragmentary state. 2. No percentages are given for totals under ten. 3. Stamps of the Colchester potter, Messor, found at Cadder and Camelon are shown as +1. 4. K/C 1-4 Manufacture at Colchester or in Kent; no stamps found to date at Colchester (see p 262). 5. Minor differences fro table m th Hul en discoveri e stampsth l w 1963,11o t ne e f yo du , stampe 5ar s being regarded as from one vessel and the intractability of certain stamps notably Hull no 32, K/C 24- 3. Colchester knoww no havs no ,i t e use dkila Ellinghat na Norfolmn i k (Frere 1977,403) herringbono ;tw e stamps of similar type to those produced at Colchester were found though none of these are from the same dies as the stamps in Scotland. However, the fabric is again fairly similar and, as a solitary stamp from the Brampton workshop in Norfolk has been recorded from Cramond, the possibility of some export from acceptede b Norfol o t Ellingha e s Th .kha m workshop was, however, linke tradition di n wit Colchestee hth r potteries whereas mortaria mad mora n ei e local tradition lik Bramptoe eth n example, would probabl distinguishablee yb . 264 | PROCEEDINGS OF THE SOCIETY, 1978-80 There can, however reasonablo n e b , e doub mortarie th f t o tha l aal t exported Scotlano t d from workshop Colchestern si , Ken Norfolr o t k were transporte coastay db l transport. Whils percentage th t e of these stamps appears high at many of the forts, the really significant ones are for those sites where extensive excavations have take totane th plac ld numbeean f mortariuo r m stamp fairls si y largee Th . high proportio thesf no e stamp t Camelonsa , Mumrill Rougd san h Castl7 2 d e an wit 6 3 h , total32 f so stamped mortaria of Antonine date, compared with the smaller proportions at sites in the central and western sectors of the Antonine Wall, and their near absence from Newstead certainly implies import and distribution fro Forte mth h rather tha roay nb d fro Tynee mth , although they were also landed there for distribution in NE England (Hull 1963, 114-16.) e maiTh n dating evidenc e herringbonth r efo e type stamps supplied mainl e r solelo yth y b y Colchester potteries, comes from their popularity on Antonine sites in Scotland but they have also been recorded from South Shields (2); Benwell destructioe th ; n deposi t Corbridga t e (Richmon Gillad& m 1950 103) o 180 daten D alsd w , , A t an Birdoswald ,no o,a 19310 d c g fi , datA . e withi perioe nth D dA 140-70/18 evidencee 0 th woul l al t dfi . One of the peculiarities of the distribution of Colchester mortaria in Scotland is that only two namestamps, both of Messor, have been found there compared with 83 of the herringbone type. It is, however, appropriate that the potter should be Messor since a mortarium from one of the Colchester kilns is stamped with both his name and a herringbone stamp, thus establishing some connection and contemporaneity in production (no stamps from these dies appear in Scotland). The evidence from rim profiles strongly suggests that most of the named potters at Colchester were working somewhat later than those usin herringbone gth e dies, though they would obviously have overlappe somo dt e extent mortarie Th . f Messoao Titusd ran , however, rarely differ from e thosth f eo potters using trademarks .140-70/18 DateD 160/70-20A D f A so d 0an 0 respectivelo ytw woule th t dfi main groups, thoug cleas i t hi r that there were links between them. herringbone Th r leaeo f typ f stameo p was doubto n , , only use s anotheda r for f trademarmo k presumably mainly by illiterate potters. As many of the Colchester examples are closely similar in design, impressiond an s were frequently only partia unlikels i t li y that they were intende distinguisdo t wore hth k of different potters thoug groupa hs a the , , ydistinguisdo h fairly readil produce yth workshoa f o t r po group of workshops. Such stamps were used elsewhere though never in such profusion but it is reasonably certain that the relatively similar stamps found on some mortaria made in NE England in the second half of the 2nd century were the result of direct imitation of the Colchester products which were selling areae welo th s .n li

Unstamped mortarium fragments in Fabric 2 235 Similar in form to no 228 which has stamp Hull no 38. From W part of N barrack in Wpraetentura, US. *23 NMAn I 6 S unregistered. US , 237 In NMAS unregistered, US. 238 Fro mbarracS praetentura,W n ki . US Nos 236-8 are fragments from three mortaria similar to form to no 234. 239 From spoil heap in NE corner of fort, US. . Abou240 mm mortariuDiae t0 halth 37 mf c o f m survives NMAn I . . 335R SF US , *241 Burnt throughout to black. From the Antonine Wall ditch, US. Nos 239-41 are similar in form of nos 220-22. *24 Almos. 2 mm Dia0 t burnhala 34 m f c fo t mortarium. Fro Antonine mth e wall ditch. US , 243 From W part of N barrack in W praetentura, US. 244 From the bath-house spoil heaps, US. 245 Heavily burnt. Fro barracparmN W praetentura,f to W n ki . US 246 Several fragments probably fro same mth e vessel. Fro barracparmN W praetentura,f to W n ki US; from W intervallum road, from crude stone setting above burnt material, and from the viapraetoria, US. 247 Fro bath-house mth e spoil heaps. US , 24 NMASn I 8 , unregistered. ,US e ar d whic9 stama an , s 22 Hulf h38 ha o po n o n lo t d an 2 l simila243-s al 24 e No o for8n ar n i r mo t generall wit5 y h22 simila stamo n o rt p. Hul33 o ln *249 Diam c 340 mm. Similar to no 226 but with a cordon on the flange. In NMAS unregistered, US. o i

H §

rn « 1

> r t-| 3

1 oa o P en

FI 0 GRoug1 h Castle: potter) 4 : 1 y( | PROCEEDING 6 26 S OF THE SOCIETY, 1978-80 *250 A mortarium with more rounded flange but with cordon and grooves. In NMAS unregistered, US. *251 Two joining fragments, from the Antonine Wall ditch, US. *25 NMAn I 2 S unregistered. US , . 25US 3 *25. US 4 *255 In NMAS unregistered, US.

The 36 fragments listed, nos 220-255 appear to be from different mortaria and wherever the appropriate part of the vessel has survived it has been stamped. Many tiny fragments in fabric 2 from these or other mortaria are not listed on account of size, or indeterminate form, or the possibility of above th attributebelongine f eb o vessels e n EnglandE ca on S l o gdo t t .Al , probably Colchester, though nos 236 and 246 are unusual in having solely quartz-like trituration grit. Mortaria liks 221eno , 226 hav0 ,profile m 23125 e ri d , an 242s 9 whic, 24 typicae e har th f o l Colchester workshops resmore e ;th ar t e individua typ n t mosi l ebu t could certainl parallelee yb d there. welt fi l l witAl h manufactur latese eth withie t typologicallar perioe 9 nth 24 dd Aan D 3 140-70y23 s .No but this merely means that they can be attributed with reasonable certainty to a date after AD 160.

Unstamped mortarium fragments in Fabric 1 and made in the Mancetter-Hartshill potteries 256 A worn fragment with incomplete rim section. The rather exaggerated treatment underneath the bead is fairly uncommon and was probably never used before AD 140 at the earliest. It is often found in the mortaria of Loccius Pro-, but unfortunately his stamp from Rough Castle is on too tin yfragmena comparisor fo t vesselse th f no Ac , D 140-70. 257 A flange fragment, c AD 140-80. From N intervallum road, phase 3 road drain. 258 A flange fragment, c AD 140-80. From the lilia spoil heaps, US. 259 A spout fragment in keeping with an Antonine date. From the lilia spoil heaps, US. 26 flangA 0 e fragment. Antonine NMAn I . S unregistered. US ,

Other mortaria midlandsmadethe in *261 Partial burning has affected the fabric of this mortarium but it approximates to fabric 1, although it contains some quartz-like tempering and has a buff slip; no trituration grit survives. This vessel is certainly from South Carlton , e Mancetter-HartshilLineth r o s l potteries than i , t ordef o r probability. AD 140-80 NMAn .I S unregistered. US ,

Mortaria Fabricn i from, 3 workshope th ofBellicus t Corbridgea survivine Th . *26g2mm piece5 Dia35 msc join togethe whole mako t rth f thip eo eu s mortarium excepting onl spoute cleays th i t neves I .r thawa rt i tstampe conditios it d dan n suggests that i t was broken near the end of the occupation. From the main ditch, US. *263 Diam c 290 mm. About one-third of the rim survives. In NMAS unregistered, US. 264 Thi similas si fragmene forn th i r mo t t stampe Bellicuy db s (p209)o n : . From road betweenN and S barracks in W praetentura, US.

unusuale Th forms represente 262-s no typica e mortarie 4y ar d b th f lo a workshomade th n ei f po Bellicus. AD 155-85.

Mortaria perhaps made in Scotland mortariuA *26 . 5 mm Dia0 mdensen i 29 m c , basically fine-textured, buff fabric, heavily tempered with small, quart red-browd zan brownisa ns gritha t I .h buff somslid pan e quart red-browd zan n (? haematite) trituration grit survives. This mortariu munusuas i botn i l h for s fabrii d md an can clearlo s y identical wit better-preserveha d vessel from Croy Hill (NMA 469R SF ) that thern eca be no doubt that they are from the same workshop. The Croy Hill example is sufficiently complete to show that it was never stamped. There is a slight similarity to a mortarium of Invomandus (Hartley 1976, 87, fig 2, no 14), though the fabric is totally different. Further examples will help to determin e whereabouteth workshoe th f o s t therpbu s certainlei ya possibilit n i s y thawa t i t Scotland. In NMAS unregistered, US. MACIVOR, THOMAS, BREEZE: ANTONINE WALL FOR ROUGF TO H CASTLE 267

Other fragments 266 A flange fragment from a mortarium in slightly sandy orange-brown fabric with fine tempering and traces of cream slip. Probably made in the N of England, though production in Scotland is t impossibleno NMAn I . S unregistered. US , 26 flangA 7 e fragment fro mmortariua similaa mn i r typ fabrif eo t red-browcbu n wit hgrea y core. Probabl Englandf o yN made jusr th o , tn e i possibl Scotlandn yi NMAn I . S unregistered. US , discolouree b fabri*26e y Th 8 cma d through burnin t appeari t orange-browe gb bu o st n with gred yan sometimes pink in the core and with a cream surface: the fine tempering includes quartz. Origin unknown. From the Antonine Wall ditch, US.

FIG 11 Rough Castle: mortaria stamp) 2 : 1 s(

A total of about sixty different mortaria (no 259 a spout fragment is excluded), is represented by rim-fragmentse th placee Th .manufactur f so followss a e ear : Unstamped fragments, Stamps different mortaria Total Colchester 15 21 36 Mancetter-Hartshill 6 4 10 Corbridge 1 3 4 Eden valle Carlisler y(n ) 1 1 Probably South Carlton 1 1 2 Probably Scotland 1 1 2 Rossington Bridge or Scotland 2 2 Northern Englan Scotlanr do d 2 2 Unknown 1 1

27 33 60 t musI saie b t d thaproportioe th t stampef no unstampeo dt d fragments indicate thatotae r th t lfo Mancetter-Hartshile th l potterie unduls si thougw ylo offerereasoe o hb n thisr n dfo nca . Nevertheless, doubo n e therb t tha n 'herringbonee eca th t ' potter Colchesterf so , etc, supplied Rough Castle o witt p hu 60% of its mortaria, more, in fact, than all other suppliers put together. The story would be much the sam onlf ei y stamped fragments were considere perhapd dan discrepance vien sth i f wo Mancettere th yn i - Hartshill numbers, 55- fro5% m Colcheste mora s i r e reliable guide Mancetter-Hartshile Th . l potteries | PROCEEDING 8 26 S OF THE SOCIETY, 1978-80 provided at least 16-6% and Bellicus only 6%, while all other suppliers must be regarded as very minor indeed. numbere Th proportiod an s f Colchesteno r mortari t Rouga h Castle, Mumrill Camelod san n contrasted with thos t otheea r site Scotlann si d must mean that their main mortarium supplies cama evi Forthe th , though som 262-d e an (no 52 samon21 g others) would come from Newstead mighe On . t expect that the Austinus mortarium (no 208, made near Carlisle), to be the result of troops carrying a modicum of supplies with them but his mortaria have too wide a distribution in Scotland for this to be readily accepted. e mortariTh a from recent excavation t Bearsdea s n (information David Breeze) e knowar , o nt includ even ea n higher proportio f mortarino a fropotterye mon , tha t Rossingtoa t n Bridge, which presumably came N by road via Carlisle. There are no sites at which Mancetter-Hartshill mortaria stand out in this way but although they are in smaller numbers they seem to be more evenly distributed. It does, however, seem that further excavation and detailed study of the pottery already available from Scotland could well reveal more about the organisation of supplies. It seems likely that different areas received their main supplies from different sources instea f completdo e dispersal being fromaie mon n supply base at Newstead. There is every reason to believe that trade with the Mancetter-Hartshill potteries existed from the beginnin occupatioe th f go Scotlandf no , being merel ycontinuatio a Englandf o N e trade th th . f n ei no The Colchester trade was, however thingw ne least ,a ,a mortarin ti a though som probabld eha y travelled coastay b N l Flaviae traffith n t knoi c n no exace periodwo th d te datW . t whicea 'herringbonee hth ' potters started working or started sending mortaria N by coastal transport but the trade at least is likely post-dato t constructioe eth Wale th yearsf lw no eve fe f onla .ni Thiy yb s factor will also affece th t distributio mortarie th f no a fro varioue mth s potteries. It was suggested in 1963 (Hull 1963, 115) that the army might have had a military contract with Colchester but this should be regarded as a possibility rather than a certainty. It would have been simpler for the army to buy through negotiatores and it is difficult to see what differences would necessarily result in the distribution of pottery.

Amphorae 269 Fragmen handlf o t amphoraf eo orangn i , e fabric, with stamp CORCLE NMAn I . 317R . SF ,US cf Callender 1965 426o ,n , probably Spanish second mi , d century AD. 270 Fragment of handle of amphora, with stamp DOM. In NMAS FR 318, US. cf Callender 1965, no 552, South Spanish, c AD 140-80. otheo 27Tw 1r amphora stamps, COR.ALFOd an L Callendef .L .c r 1965 425o ,n , probabl Spanisf yo h origin, mid second century AD, and no 37, Spanish (Anderson 1905, 493; Macdonald 1934, 455). necd an k m 27sherdRi 2 f threso e amphorae joinino tw d g an fragment, f handleo s lighn i , t brown- greyish-buff fabric . RimUS , , basbodd ean y sherd f amphoraso lighn i , t brown-buff fabric, from spoie th l heapbate th ht sa house . Fragment handleso tw f s o pinkish-buf n i , f fabric, fro maie mth n ditch. Amphora body sherds from the via praetoria, US; S barrack in W praetentura, associated with drip trench; HQ forecourt, US; E part of N barrack in W praetentura, US; W part of N barrac praetentura,W n ki USfrod intervalluman m; W road, phas roae3 d drain, seale burny db t material.

Glass Miss D Charlesworth, Inspectorate of Ancient Monuments, DOE *273 Beake goon ri d quality colourless glass, fragments give almost complete baseprofilo n t , eheavbu y strain cracking. Moulded, slightly everted rim, ground lin relien ei f below anothed an , r neae rth inward curve for the base. The beaker is unusual because, although the glass is thick, the side has not been decorated. Similar beakers generally mould-blown, hav eclosea d diape f faceto r s decoratin side gth e betweeo tw e nth relief-cut lines (Isings form 21) and can be dated to the late 1st and early 2nd century. Examples found in Britain include fragments from Newstead, Lyne, Castlecary and Cadder (Charlesworth 1959, 41-2, fig 4) and dated examples from Cardean, Agricolan (Wilson 1969, 202, pi XIV, 1) and Caerleon, under the Trajanic rampart (Nash-William amphitheatree th n i s d (Wheele0 1929an 10 ) 2 c , , , 257-8& r18 g fi , MACIVOR, THOMAS, BREEZE: ANTONINE WALL FORT OF ROUGH CASTLE | 269 7

273

FI 2 GRoug1 h Castle: glass beaker (1:2) Wheeler 1928, 170typ e bees , Th eha pn. i mor6) 34. & 5e recently discusse connection di n wit freee hth - blown and faceted beakers at Fishbourne by Harden and Price (1971 340-1, nos 41-4). The probable centres of manufacture are Alexandria and Italy. From the drip-trench between the N 'tent-like structure' intervallumN e th d an (fi) road3 g e glas s associateTh . wa s d with stones, spread fro intervallume mth road, which fille drip-trenche dth . 274 Fragment greenisf so h glass fro base flasa m th f eo k wit open ha n pushed-in ring identificatioe .Th n shape oflasa th f s determines ea ki fragmena y db f infoldeo t flattened dan witm dri h e parth f o t handlea e b y neckd ma vesse t I .t normall bu l handlee yth , bein thickef go r glass tha bode nth f yo the vessel, survives. lst-3rd century. From the burnt debris overlying all the road surfaces in the angle betwee ramparascensus.e W th e d nth an t 275 Fragment multi-ribbef so d handl greenisf eo h glass, lst-3rd century barrack-blocS f .o Frod en m E k in praetentura, US. 276 Three fragments of window glass, green, rough on one surface so presumably moulded. From W end of N barrack-block in praetentura, US. 277 Fragment banglea f so , plano-convex section, green glass wit hpot-hooa k trai opaqun li e white. These bangle wele sar l know siten no s between Hadrian' Antonine th d san e fac n Wali t t havlbu e a very wide distribution as Stevenson indicates (Stevenson 1976). They are more commonly found in the native settlement sites than in the Roman forts but at Newstead, for example, several were recorded. This bangle is Kilbride Jones type 3F (Kilbride Jones 1938, 366-95). The metal looks like Roman bottl anothee eb glas y thid san ma r exampl re-use th Romaf f eo o n glass frag- ments. From the lilia spoil heaps, US.

Bronze *278 Small fragment of brooch of 'elbow' type, with spring and circular blue enamel inset. From W intervallum road, seale burny db t material. smalo *27Tw 9l bronze studs, corroded together. US , *280 Two fragments of mountings ? From area between N rampart and intervallum road, in yellow clay and stones of rampart backing. *281 Small fragment of bronze, with spoon-like depression on one side, and slight grooves on other side. From N barrack in W praetentura, US.

Iron *282 One long thin nail, US. irregula *28o *28d an 3Tw 4 r rings, with oval sections.

282, 283 and 284 from area immediately outside N rampart, US. *285 From N intervallum road, 1 bent nail from phase 3 road drain, 1 short nail on phase 1 road surface; 8 nails from area between N rampart and intervallum road, US; 2 nails from W intervallum road, in burnt material above all road levels nai;1 l from road barrackbetweeS d an Wn sN i praetentura, s 270 | PROCEEDINGS OF THE SOCIETY, 1978-80

278 279

282

283

285

284 FIG 13 Rough Castle: metal work (1 : 2) on gravel surface of road; 1 bent nail from N barrack in W praetentura, 1 nail head from W rampart, associated with rampart pitching nail0 1 ; s fro liliae mth spoil naiheaps1 ; l fro US ,maie mth n ditch, US nai1 ; l from area immediately outsid nails2 . 4 ramparteN d US , an , ,US 286 Dr Elizabeth Slater has kindly contributed the following note: ovaBasn a f le o iro n bowl, les sthick m tham n1 , with currenm t dimensionm 5 4 lon m d gm an 5 s6 across. It had a very rough surface and, as there was a thick deposit of corrosion product on one part of it, this was probably due to extensive corrosion. The corrosion product was identified as a mixtur irof iroed o nan n oxide X-ray sb y diffraction techniques materiae th bowe s d th f an ,lwa lo identifie iros da n from thifrod appearances an m it samplA . bees eha nResearce senth o t d han Reactor Centre, East Kilbrid neutror efo n activation analysi confiro st m this initial diagnosis.

Lead 287 Short round objec - weightt ? Fro barrac parN m f W o t k praetentura,inW . US 288 Two joining fragments of spoon-shaped object - mould? From area between N rampart and intervallum road. US , 289 Flat, spoon-shaped object with slightly raised edges - mould ? From road between N and S barracks in W praetentura, US.

Slag and baked clay r ElizabetD h Slater, Departmen Archaeologyf o t , Universit Glasgof yo w 290 Four piece bakef so d clavitrified yan d material. Three pieces fitted togethe givo rt esemi-circulaa r mass, of maximum diam 200 mm, with a thickness increasing from the edge to a maximum value MACIVOR, THOMAS, BREEZE: ANTONINE WALL FOR ROUGF TO H 1 CASTL27 | E lowere Th . , curvemm 5 o4 fd surfacf bakeo s dewa cla y whic uppee fused th hha o drt layerf so porous, silicious material. Samples were vitrifiee taketh f no d materia X-rad an l y diffractiod an n analysis showed that it contained small pieces of iron, iron oxides and silicates - the overall composition was of a fayalite slag, a mixture of iron, oxygen and silicon. This type of slag is typical of primitive copper and iron smelting but the presence of small pieces of iron in this sample indicates that it is the result of iron smelting. The shape and composition of these three samples indicate that they came from a bowl furnace, with the inner lining at the base of the furnace fusing wit smeltine hth g product t beeno nslae d tappeg.Th ha d fro furnacee mth fourte Th . h sampld eha a similar composition to the other three but differed in form. It was a mixture of slag, charcoal and pieces of iron with a few pieces of baked clay entrapped within it. It was rectangular in shape, , wit hmaximua mm 0 6 y b smalmA m . thicknesm l 0 portiomm 4 surfac5 e 3 th f smoots sf o n o ewa h alsy o ma curved havt i an d e dan com e fro basfurnacee a mth f e o . Fro draie m t th throug ncu h the phase 3 intervallum road. 291 Piec bakef eo d clay irregulaf o , r shape t witbu , hsmootha , curved outer surface. Maximum length wit m ; widt m hvariabla 4 55 mm h3 e thickness averagin outee Th . rg surfacaboumm d 0 1 teha suffere more dth e intense heating, wit hblackenea d layer extendin fron g i surfacee aboumth m m 1 t . The fabric was clay with small inclusions of mica, but no added temper. There was a groove running across the rough inner surface which, with the curved outer surface, suggested that this was part of a tube or tuyere of clay, the outer surface of which was heated during use. A definite, positive identification is difficult when only part of an object survives and, in this case, the groove was very narrow for a tuyere, less than 10 mm in diameter (if this portion is representative of the whole t follono convee d wth di object t i x d outecurve )an th f reo surface. pieceo 29 Tw 2f bake so d clay, each witflae hton surface, otherwis irregulan a f eo s r shapewa e On . almost square, of side measurement 30 mm and maximum thickness 13 mm. The other was triangular, 20 mm across, and of maximum thickness 15 mm. They were similar in fabric to 291, clay with some adde o mic n sand t adan d bu temper. The beed ycoloun weri ha n d d subjecere ran t to some degree of heating. Their form and fabric suggest that they were part of a structure, rather than an artefact. 293 Piece of baked clay, roughly rectangular in shape, with one smooth surface, otherwise irregular in form. Maximu mwit , m widt lengtm hmaximua mm 5 h3 0 h6 smoote Th m . thicknesh mm 0 2 f so upper surface, which was slightly concave, had been subject to the most heat with a blackened, burnt layer extendin fron gi surfacee aboumth m m 2 t. 292Similad an . 1 fabrin ri 29 o ct

Wood and Charcoal BarberW J , Central Excavation Unit(AMD SD ,) Althoug s possiblwa t hi identifo et y each sample surfacee th s suffered a , ha s d damage, evidencr efo woodworkine th g techniques uselackings dwa . 294 Four fragments wer f similao e r proportions with roughly rectangular cross-section, 15-2m m 0 thic 60-7d widekan m 0m . Their lengths varie sinct cleas di bu t ei r tha beetd fragmentha n3 d an s2 time excavatioth f eo t a t cu meaningfua nt thano s ti l dimension. Fragment radialle ar 4 yd san 1,2 spli f Quercuso t planks2 Abiesf d o an 4 (Oak1 ,d albaan ) (Silver Fir). Fragmen , Quercus,3 t a s i sawn plank. 295 A heavily carbonised piece of Carpinus betulus (Hornbeam). Its original form would seem to have been serratethaa f o t d disc wit hcentraa l perforation, somewhae th e b wheeltg y likco ma e a t :i offcut created by drilling a circle of small holes in order to produce a single large hole in a plank. The hole would have been about 10 mm in diameter and the plank 25 mm thick. *296 A curious paddle-like object with an elongated oval paddle face 210 mm long and 70 mm wide set at right anglecross-section i rectangulaa m o st m 0 2 long11 d m y 0n b an m rt seem. I stockm m o st 0 ,7 be complet tha n estoce i tth neatlks i y t cutbrokenno , fashiones i t .I d fro msingla e piec f Quercus.eo *297 190 mm long, 38 mm wide and 27 mm thick (maximum dimensions); now oval in section. It is fashioned fro mpieca Betulaf eo (Birch showd )an s'edges's weait n ro . Ther cleae t markear rcu s edgee on .n Probablo yspoka wheela f eo f Curic ; p, ewher& i LXIX19112 2A e e29 th & , 2 , spokes illustrated are 'probably of willow'. The oval shape of the Rough Castle spoke is presumably resule th latef to r pressure. 27 | 2PROCEEDING S OF THE SOCIETY, 1978-80

297

FI 4 GRoug1 h Castle: wood (1) :2 MACIVOR, THOMAS, BREEZE: ANTONINE WALL FORT OF ROUGH CASTLE | 273 298 This seems to be part of a stake tip, now carbonised. Only 120 mm of its length survives and it would have been a little more than 40 mm in diam originally. This is identified as Pinus probably silvestris (Scots Pine). mose 29Th 9t wooe distorteth l d al fragments f do , this survive hemisphericaa s sa l mass, carbonised on its curved surface. Its base is oval, 70 mm by 90 mm and it is 40 mm high. Its original form is not readily discernible and since it is a very knotty and distorted fragment it may never have been fashionea d object. This als Abiess oi alba. 30 singlA 0 e large piec charcoaf eo Pinus,f lo again probably silvestris. 301 Eight piece Fraxinusf so excelsior (Ash) charcoal. Some t leas a , same parte th ar t f es o tree. 302 Four large fragments of Pinus Silvestris charcoal. Two are clearly from the same tree. 303 A single small fragment of Betula charcoal. With the exception of the hornbeam, all of the wood species identified could have been found locally during the Roman period. The hornbeam is almost certainly an import, possibly from southern England wher t woulei d have been readily available. Samples from each specimen have bee ne palaeoecolog kepth t a t y laborator e Centrath f yo l Excavation Uni Scotlandr fo t . Textiles J P Wild, Department of Archaeology, University of Manchester Both textiles are replaced by the corrosion products of the iron to which they adhere, but their structure is clea placesn ri .

304 Two layers of half-basket weave, c 5 cm in all. It is a regular weave and the spin quality of the yards is good. Fibre indeterminate. 2 System (1), doubles, weak Z-spun, c 14 pairs per cm, maximum length 3-5 cm. System (2), singles, Z-spun, c 10 per cm, wide set, maximum length 2 cm. The yarn in (2) is of the same quality and weight as that in (1), but it is harder spun and may represen warpe th t . From area betwee ramparnN intervallumd an t road. US , 305 While the amalgam of corrosion products carrying the textiles do not make a direct join, it is clear that they belong together. Same textile as 304, amounting to c 2-5 cm2, US.

Half-basket weave, with pairs of yarns in one system, is relatively common in the Roman world, both for wool and linen textiles (Wild 1970, 46). Animal bones G W I Hodgson, Duncan of Jordanstone College, Dundee animae Th l remains apparentl l camyal e from mammal consisd san bonf o t e fragments, teet antlerd han . materiae th poo a l n i Al rs li stat preservatiof eo thid thoughns san i relatee b o t soi o dt l conditions rather than destruction by heat or force. Most of the fragments are associated with heavy deposits of vivianite. item e ditch N th froe l e sar m th Al , wit exceptioe hth f 314no , whic fros h i bath-hous e mth e spoil-heap. Cattle 306 3 slightly worn mature upper molars M3, 2, 1. 307 A single neural arch from a large bovine-like vertebra. 308 Four fragments probably fro same mth e right radius (distal widtmm)2 7 h= . 309 proxima e lefa Par th f t f o metatarsalo t d en l . 310 Several fragments from long bones. These fragment t butchersno e ar s ' chipping t appeasbu o t r have split off from the main bones by weathering. Horse 31 matur2 1 e premolar teeth fro uppee m th PM4 w rja . ,3 312 Shaf distaequind n a an tf o l e metapodial (distal widtmm)7 4 h= . | PROCEEDING 4 27 SOCIETYE TH F SO , 1978-80

316

FIG 15 Rough Castle: leathe) 2 : 1 r(

Red deer 313 2 heavily eroded tines from antlers.

?Pig 31 fragmenA 4 heavila f o t y eroded incisor toot h- possibl y from pig. MACIVOR, THOMAS, BREEZE: ANTONINE WALL FOR ROUGF TO H 5 CASTL27 | E Leather The sixty-three fragments of leather consist of soles, heel stiffeners, uppers, offcuts and scraps. Two shoes are nearly complete, with sole and upper still attached. The uppers mostly appear to belong calcei,o t which wer least ea madf to foup eu r part nailea s- d outer sole insolen ,a heelguar,a stiffenerr do , and a one-piece upper, stitched at the front of the vamp. Most shoes also had a middle sole consisting of several layer leathef so r betwee outee innend th ran r soles. Thongs were use layere holdo th t l dsal together. Keppie divided the calcei from Bar Hill into three types; his classification has been followed, where possible (Keppie 1975, 68-77). *315 Calceus, right foot, with insole, middle sole, heel stiffener and upper. Insole has pairs of slits forming tunnel hole r thongfo s . Middle sole consist f threo s e thicknesse f leatheo s r thonged together; bottom laye s beeha r n perforate nailsb ho .y db Semi-circula r heel stiflener, with lower margin folde tucked dan d under bottom laye middlf ro e sole. Upper consists of quarters, complete inner side of vamp, and part of outer side. Quarters decorated with two very slight semi-circular projections on either side and with a short triangular t centrea b edgp ta to ; e marked, insid outd ean , with slight cuts, possibly imitatin pattere gth n left by a hem stitch. The quarters end with a latchet on each side. Inner one complete, though broken wides it ;perforated ren approximateln a y db y triangulay b othe, d mm ren 3 1 r y holeb 8 ,c an oval hole, c 4 by 10 mm. Outer latchet broken; also has triangular hole. Both latchets decorated with stamped diamond pattern above and below triangular holes, and with rouletting along both margins; roulettin from gmm end 0 ova1 sc l hole. Rouletting also parallel wit edgp h quartersto f eo , as far as semi-circular projections, and along top edge of vamp. Inner side of vamp has second latchet, connecte boto dt h centr seame f vamth e o d , pleavinan g oblong cut-out torareaw nNo . at junction of strap and centre of vamp, but cut edges suggest triangular hole similar to those on latchets at throat. Two oval holes on latchet, c 2-5 by 9 mm and c 1-5 by 4 mm. Stamped diamond pattern beneath junctio f strapno vampd sease an To m. survives, with stitching channe f fleso l h to grain holes, stitch length 4 mm. Bottom margin of upper folded and tucked under margin of heel stiffener and middle sole. From the main ditch, US. cf Keppie Type A (1975, 68-69, 71, fig 22.1, 3, 5). *316 Calceus, right foot, outer sole with hob nails, middle sole, insole, heel stiflener and upper. Hob nails penetrate outer sole, middle sole, insol bottod ean m margin heef so l stiffene upperd an r e ;th latter have been tucke between di n middl outed ean r soles. Upper comprises quarters, small part of outer sid vampf eo , mos innef o t r sid vampf eo loosd an , e fragmen vamf o t p with thong which passes through an eyelet on inner side. Quarters very worn, no top edge survives. Inner side has one elongate r latcheo b t ankledta a t ,rounde o wittw h d pointedan , dmm ova3 1 ly holeb 3 c , projections on vamp, each with an oval eyelet, c 6 by 9 mm. Tab and projections have been lined or faced with a thin strip of leather, stitched to upper along top edge and round the perimeters of two of the eyelets. Toe of vamp is missing. Thong passes through eyelet nearest front of vamp. One end of thong knotted, other end is continuation of fragment of upper with stitching channel with grain to edge holes, stitch length c 4 mm; probably toe seam of outer side of vamp. Outer sid f vameo p very worn, onlprojectioe yon eyeled nan t survives, opposite innemiddln o e r eon side. Eyele ovals ti , 4xStitchin. 8mm g channels with grai fleso nt h holes, stitch lengt h3-3-c , 5mm round eyelet and along top edge, most probably for facing, which does not survive. From the main Keppi f ditchc . US ,e t withou Typbu , radiatine C th t 24.40gg stripfi , , 76 s41) , (197574 . , 72 , *317 Calceus fragmen f uppero t , with quarter pard f botsan o t h side f vampso . Raised pea t centrka e of quarters. Stubs of parallel ribbing on both sides of vamp. Bases of stubs are marked by small arc t int leathere scu o th t perforatt whic edgeno p bu , o hTo d s. eit betwee n centr quarterf eo d san ribbing cut in tightly scalloped formation - cut, not stitched. Very worn. US. cf Keppie Type B (1975, 70, 73, fig 23, 18, 19, 20). 318 Calceus, very small fragment of upper, with stubs of parallel ribbing. US. cf Keppie Type B (1975, 70, 73). *319 Calceus, four fragment upperf so . . Fragmena t with rounded projectio ; stitchinovad nan mm l eyelet8 gx 4 channel c , f graiso o nt flesh holes, stitch lengt mm5 h3. , round hol alond eedgean p gto , probabl attachmenr yfo facingf to , b. Thin strip with two rounded projections, each with an oval eyelet, c 4 x 8 mm, and with stitching channe graif lo fleso nt h holes, stitch lengt h; mos 3-5mm t probably facin r (a)gfo . 276 | PROCEEDINGS OF THE SOCIETY, 1978-80

FI 6 GRoug1 h Castle: leathe) 4 : 1 r( MACIVOR, THOMAS, BREEZE: ANTONINE WALL FORT OF ROUGH CASTLE | 277

. Fragmenc t witstitchino htw g channels wite on ,h edge-grain holes, stitch , othelengtmm r h4 with grai flesno t h holes, stitc wit d edge han h on e, lengt continuinmm h3 thons ga g which joins (d). Most probably front of vamp with toe seams. fragmento dTw . f thono s g knotted together joinine on , g thon f (c)go ; thong passes througn ha eyele (a)f o t . US. cf Keppie Type C (1975, 72, 74, 76). *320 Calceus, fragment of upper, with stubs of thin ribs and with stitching channel of grain to edge holes alon edgee gon . Very worn, stitch lengt t measurablhno Keppif c . eUS e Typ (1975? C e r Bo , 70, 72, 73). *321 Calceus, fragment of upper, with quarters and part of vamp of one side. One side of quarters very worn, but on the other, remains of stitched top edge formed by folding leather over and stitching it with grai fleso nt h seam, stitch lengt samn O h . e stubd 3- sidethic5an mm m sb m ,k ri par 5 f o t of 6-2 mm thick ribs; beneath ribs, three stamped rosettes. Bottom edge of fragment folded and perforated with holes from hob nails of sole. US. cf Keppie Type C? (1975, 72, 73 fig 23.5). *322 Calceus, fragmen f uppero t , with quarter partd san f bot so h side vampf so edgp f quarterTo .e o s does not survive but probably ended in a raised peak at the centre. On both sides of quarters two very slight round projections, similar to those on 1. On one side, elongated ankle tab on latchet with oval eyelet beginnind , an 5x1 , 5secondf mm g o , thinne r strapo othen b O .rta r side, stump of ankle latchet. Rouletting parallel with top edge on both sides. US. cf Keppie Type A (1975, 22.1)g fi , .69 , 68 *323 Calceus, heel stiffene fragmend an r upperf o t . Bottom edg stiffenef eo r flattene attachmenr dfo o t sole piercenailsd b an ,ho .y dVerb y worn quarter partd san botf so h side vamf so p survive, with stump f thiso n strips, simila thoso t r e foun calcein do Keppif o side f vamTypeo On . epC ends with stitching channe graif lo fleso nt h holes, stitch lengt seame hto 4-5-- . Bottom 5m m edge folded pierced an nailsy db Keppif c , e 23.25) g Typfi , (1975eC .73 , 72 , *324 Calceus, fragment of vamp of upper, with stump of ankle latchet, and with two rounded projections; one with a short oval eyelet, 5x13 mm, the other with a longer hole, 6 x 28 mm, with short tri- angular hole, , 4xbeneat5mm . Arehit a beneath both eyelets decorated with cut-out patterf no radiating strips. Fragment end stitchinn si g channe edgf lo graio et n holes, stitch length- 3-m 4m toe seam. US. cf Keppie Type C (1975, 72, 73, fig 23.26, 28). 325 Calceus, fragmen vamf o t upperf po , with stump parallef so l ribs witd ,an h bottom edge perforated by holes for hob nails. NMAS FR 398. cf Keppie Type B (1975, 70, 73, fig 23.19). *326 Carbatina Fragmen? uppef o t edge-flese r wit loopon e d hon an , h seam, stitc h, and lengtmm , 5 hc t righa t angle, shorit o t st split, with sides stitched together with edge-flesh seam. NMAR SF 399. 20.11Keppif g 14 c . fi , , 12 ,63 e , 197562 , ,59 327 Fragment of leather with pair of parallel stitching channels with grain to flesh holes stitch length seame to t lowe- 5-5- ?A m r5m edg fragmentf eo , trace holef so snailsb madho .y eProbablb y part of upper of calceus. US. 328 Calceus, very worn fragment of upper consisting of quarters, and one side of vamp, with stumps of fine ribbing and part of toe-seam with grain to edge holes, stitch length c 3 mm. US. cf Keppie Type C (1975, 72, 73). 329 Five other fragments of upper, two with bottom edge folded and pierced by nails, one with stitching channel of grain to flesh holes, stitch length c 5 mm. US. 330 Heel stiffener, approximately semi-circular, maximu , bottom mm heigh m5 4 edgc t e folder dfo attachment to sole, and pierced by nails. Grain side inwards. US. 331 Fragments of 7 similar heel stiffeners, maximum height 25 mm, 40 mm, 45 mm, 50 mm, 55 mm US. *332 Sole. Slim forepart of small right sole inscribed on underside with a monogram ME. Pointed toe. Consist layero tw leatherf f so so , held together with ironailsb nho , end somf so whicf eo h penetrate insolee th havd ,an e been hammered flat. Insol grais eha n sid Nai. eup l pattern: around edged an , up centre of forepart in irregular zig zag. Position of some missing nails marked by slight cracks and by circular impressions. Maximum surviving length 135 mm, thickness c 8 mm, excluding hob nails. US. 333 Sole. One layer of left sole with pointed toe, with holes for hob nails, and with 4 pairs of tunnel holes, one with fragment of thong. Grain side uppermost. Maximum surviving length 275 mm, maximu. m, thicknesUS widt. mm 9 hmm 9 s3 | PROCEEDING 8 27 S OF THE SOCIETY, 1978-80 *334 Sole. Two layers of left sole, held together with thongs. Upper layer is smaller, and is filling or packing, not insole. Holes indicate position of hob nails. Maximum surviving length 235 mm, maximu m, thicknes widt, excludinmm 2 hmm 9 s3 g filling. US . 335 Sole. Two layers of left sole, one of which constitutes filling or packing. Tunnel holes indicate that the two layers were held together with thongs. Holes mark position of hob nails. Maximum surviving , maximulengt. mm 1 US , thicknes h. m27 widtmm mm 4 3 h10 sc *336 Sole. One layer with holes for hob nails. Maximum length 243 mm, maximum width 93 mm, thickness c 2 mm. US. *337 Sole, fragment, probably left, consistin insolf go fillind ean g held together with thongs; triangular offcu - extrt a packing nails.b Holeho .r Maximusfo m survivin maximu, g lengtmm 1 mh27 width 100 mm, thickness 8-5 mm. US. 338 Sole, fragment, possibly child's shoe, mos foreparf o t t missing. Hole nailsr tunnesfo d an , l holes for thongs. Maximum surviving length 169 mm, thickness c 2 mm. US. 339 Fragment soles6 1 f s,o wit hnailsb holewitho x r si , hsfo tunnel hole thongsr sfo . US . 340 Offcut, with nail holes - most probably part of filling of sole. US. 341 13 similar offcuts. US. heelguarde Th fragmentd san solef o sprobable sar y par calcei,f o t although they could belono gt caligae (Keppie 1975, 78). DISCUSSION The fort at Rough Castle was demonstrated in 1903 to be secondary to the Antonine Wall rampart (Anderson 1905 rampartE , d for e 459)foundatione an th t f Th wer.so W e eth widef so r and constructe highea t da r level tha Antonine nth e Wall rampar rested tan d agains facS e tth of that rampart fore tTh . thus falls int same oth e clas Westerwoods sa , Croy Hill, Cadded an r possibly Bearsden, additionl whical e Antonine har th o st e Wall (Gillam 1975, 52), though unlike , Croy Hill and Bearsden there is a causeway across the Wall ditch outside the N gate of the fort at Rough Castle. However, it may be noted, for what it is worth, that the causeway is not symmetrically placed in front of the gate, but lies slightly to one side. It has been suggested tha so-callee th t d wagon park smale th , l defended enclosure gatE immediatel f e o th f o E yN Rough Castle, could have been a primary fortlet, later adapted for a different use. This, however, would not explain the off-centre position of the causeway, not least because there are no cause- ways outsid fortlete gateN th e f seo t Seabegth s a s Wood, Croy Hill, Glasgow Bridge, and Duntocher ( was a special case as it guarded the road through the Wall, while the position at Kinneil is not known), just as, with the exception of certain special cases, there are no causeways outside the N gates of milecastles on Hadrian's Wall. fore t Rouga tTh h Castl secone th s ei d smalles Antonine th n o t e Wall t measureI . s 0-a 6h (1-5 acres) over the ramparts, 0-4 ha (1 acre) internally. The garrison at one time, probably in the first Antonine period, was cohors VI Nerviorum (RIB 2144 & 2145), one of the smallest type of auxiliary unit in the Roman army, containing six centuries of 80 men each. Nevertheless unie mand th incapabls foran te n yi th wa t n mus me holdinf eto e havth l egal been permanently outposted. The fort was defended by a turf rampart on a stone base, 6-1 m wide to the W, S and E, 4-3 m wide to the N where the fort was attached to the rear of the Antonine Wall. An ascensus ramparte th f bere o p Th .m to beyone ramparW th e o metalleds t dgat th d W wa t ele e b yth . Two ditche beyony sla rampare southerde th th n o t n circuidefencese th f o t , three betweee nth fron non N d corner E e an tN e th e cornerbetween W th o : gatN E e d e e an on , n th th gatd W ean annexen a , containin bathhousee e gfronth th N f forte e o tattachesid th s E Th . f e o wa , th o dt fort was further protected by 10 rows of defensive pits (lilid). There is no obvious tactical reason for thes approacN e pitth s sa h rises slightly toward forte sth . MACIVOR, THOMAS, BREEZE: ANTONINE WALL FORT OF ROUGH CASTLE | 279 The batter on the outer face of the W rampart, 78 degrees, is unusually steep. Jones (1975, studs note70s hi pre-Hadriani) ha f yn o di c fort defence Britain i s n that eart opposes a h- o dt timbe r- revette d ramparts were battere angl degreesn 5 a 7 betweet f dea o d an . 5 Recenn6 t work fore t Strageatth a t t a h (Frere demonstrates 1979ha ) ,37 d tha outee ramparE th t e r th fac f teo in the Flavian period, which was constructed on a timber strapping, had a slope of 67 degrees, while the rear face, standing to a height of 1-2 m, was vertical. The turf sector of Hadrian's Wall at the front had a batter of 76 degrees above its turf base while the back had a slope of 78 degrees though this probably changed after the first few courses of turf to a more gentle profile (Simpso Richmonn& d Antonin1935e th 13) & n ,9 O . e Wall similar wide variationn si the angle of batter have been noted. The best survey is still that carried out by the Glasgow Archaeological Societ 1890e th n ysi (GAS 1899) thren .I e section turf-wore sth k stood 1-2-1-5m high. At Croy Hill E expansion (section 11) the batter on the S face was 62£ degrees, at Croy Hil expansioW l 12Ao n(n )t averagei d 56 r Hil£ degreesBa l fact sectioS a e t th slope bu ,n 1 d at an angle of 83£ degrees. At the fort on Bar Hill 10 years later one section near the NE corner revealed an angle of slope of about 75 degrees (Macdonald & Park 1906, 22). These comparanda emphasis unusualle eth y steep outee ramparanglW th f e reo facth t Rougf a t eo h Castlee th n :i sections noted only two, those at Strageath and on the Antonine Wall at Bar Hill, surpass that t Rouga h Castl steepnessr efo range .Th angl f eo slopf eo e noted above demonstrate difficulw sho t attempo t s i restoratioe t i th t rampary an f no t particula profiln i determino t d y ean tr o e t r eth original height of a rampart. fore tTh containe normae dth l complemen principaf to l buildings: headquarters, command- ing officer's house and granary, all in stone. Space was so much at a premium that the com- manding officer's house was partly built over the intervallum. The buildings located in the praetentura were of timber. The initial occupation in the W half of this area appears to have been temporary t thisucceedes bu ,stimbe wa o tw y drb buildings northerle .Th y bloc 5-7s k9wa m wid d 26-5ean lon2m g whil s neighboueit s 6-2 wid wa 5m rd presumabl ean y 26-5 long2m . These appropriate buildingth n i e ar s e positio r barracksnfo t the muce bu ,y ar h shorter than normal barrack-blocks, whic nature site th eTh h f . has eo varm lengt0 n ,yi 5 o t h from 1 m4 however, obviously restricte e building th e siz f dth o e d barrack-block an s f similao s r length have been foun t somda Scotlane W siteS n si d where similar conditions prevailed roomo tw e s .Th at the W end of the N block are of the size normally found in barrack-blocks, while the appearance o verandaa fbuildin S e th n ghi suggeststorehousa t no s s thar workshopwa eo t i t alss i t oI . unlikely thabuildinge th t s shoul comparee db barrack-like th o dt e building Llystynn Pe t a s , interpreted as a hospital (Hogg 1968, 133-4), and at Corbridge forts II, III, IVA and possibly also IA and IB, interpreted as quarters for the administrative staff (Gillam 1977, 58-9), as all these buildings were located in the central range and not the praetentura or retentura. In short, therreasoo n s ei assumo nt e that thes buildingo etw s wer t barrack-blockseno , conformino gt arrangemenn a t prophesise Macdonaly db d (1933, 174-6). The two buildings are best compared to the 'half barrack-blocks found at some sites in SW Scotland (Tabl littl. a The 3) e eyar longer t rathebu , r narrower tha blocke e nth th t a s contemporary fortle t Barburga t h Mill, Dumfriesshire (Breeze 1974, 136 noticeabld )an y longer than similar buildings, interprete barrack-blockss da smale th n l,i fort Raeburnfoot sa t (Robertson 1961, 32-9) and Crawford (Maxwell 1972, 175). The two rooms at the W end of the N block compare well in area to barrack rooms at Bar Hill on the Antonine Wall and at Barburgh Mill constructioe (TablTh . barrack-bloce 4) on f no k wit verandaha anothed han r withou parals i t - leled at Barburgh Mill and at Bearsden on the Antonine Wall, while the fact that the blocks did not face each other is again paralleled at Bearsden ( 10, 1979, 277). 280 | PROCEEDINGS OF THE SOCIETY, 1978-80 Some post-holes were located belo vie alinth e wpraetoriaf th e o undoubteo n t bu d buildings were located in that part of the E praetentura examined. At Cadder on the Antonine Wall part of the praetentura was left free of buildings and a similar situation may have existed at Bar Hill retenturae parn th i f d o t an t Bearsdea n (see Robertso plans)r fo 0 2 .n d How 1979an 8 -,1 , fig16 s ever possibls i t i , praetenturabuildingo eW tw tha e e th tth n si t Rouga h t for Castle no m th d edi buildinhalvee o on tw f t ratheso gbu r formebuildinghalveo W tw e f dsth o s which, albeit divided by an unusually wide via praetoria, spread across the whole width of the praetentura in the

TABLE 3 'Half-sized' barrack-blocks in Antonine I Scotland Rough Castle fort 26-5 x 5-8 ? x 6-25 Raeburnfoot fort 18-0 x 7-3 Crawford fort 19- 87-x 2 Barburgh Mill fortlet manner postulated at contemporary Crawford (Maxwell 1972, 175). The eastern sections of such buildings can have been no more than 15-24 m long. The total possible length of these two, hypothetical, divided barrack-blocks, 41-76 m, is a little on the short side, but still within the normal rang barrack-blocf eo k lengths. The two rooms at the W end of the N block together may have formed the officer's block r alternativel o e contuberniath f mene o th o r . ytw fo Ther pointere ear eithen si r directionn I . the N building the two end rooms did not project in the normal way of officers' suites and the building lacked a verandah which may suggest that the officer's quarters were elsewhere. However, at Bearsden one barrack-block did not have a verandah while another did and in neither building e officer'th d di s quarter sroom o project tw slightle e ar s Th . y different size n t i thano t bu s ,i t unusual in barrack-blocks. The available floor space compares well to that in barrack rooms at Bar Hill, Barburgh Mill, Strageath and in forts on Hadrian's Wall, and is considerably more generous than that at Bearsden, though here the barrack blocks are unusually small (Tables 4&5). combinee Th d roomo flootw , howeverre s is areth f ao , much less than that provider dfo e officeth t Barburga r hfortn i Hadriansn Milo d s an l ' Wall: Bearsden t strictlagaino s ni y comparable owing to the unusual barrack-blocks here. This might be taken to suggest that the rooms were more probably barrack rooms. However, floo reliabla rt areno s eai statistice th ; barrack-block t Bearsdesa anomaloue b y nma t the exisd wersbu d y di an te smaller than those

TABLE 4 Officers' suite contuberniad san Antoninn i eI Scotlan d Timber/ Area of Area of Stone officer's suite contubernia (in sq m) (in sq m) Rough Castle fort T T24-6 . 46-7^ 1 21-8 T 9 24-5 Bearsden fort T 25-6 14-4 21 14-4 Strageath fort T 1 27 9 20-6 Barburgh Mill fortlet T 64-7 22 MACIVOR, THOMAS, BREEZE: ANTONINE WALL FORT OF ROUGH CASTLE | 281 TABLE 5 Officers' suite contuberniad san Hadrianie fortn th i f so c perio Hadrian'n do s Wall Timber/ Are f ao Aref ao Stone officer's suite contubernia ) m q s n (i (imq ns ) 4 2 0 7 S Wallsend 9 2 0 9 S Benwell 5 2 6 7 Housestead S s S 74 26 Birdoswald S ? 26 at Rough Castle. Further, the unit at Rough Castle was a cohors quingenaria peditata, whereas the statistics quoted for Hadrian's Wall refer to larger units, possibly allowed more generous accommodation (Breeze & Dobson 1969,27-31). It would thus not be impossible for a centurion statione t Rougda h Castl allowee b o et d only floo46-m q 7rs area. Ther s anothei e r consideration. Mos e samiath t n notee warb l threy edal (anma e t i d fragment f windoo s w glass (276)) recovered from thi praetenturae s th hal f o f foune s th wa t da W end of the N building (19, 20, 22, 25-29) or on the W intervallum (21, 24, 31 and 33) rather than the E end of this building (34 & 38) or its neighbour (37). Work at Bearsden has suggested that there at least usually only the officers used samian ware (Breeze 1977), and if the same is tru t Rougea h Castl t seemei s probable tha centurioe th t n occupied one r moro , e likely both, buildingN e th average f o Th . d een widtW roomo eacf e ho th tw h t e sa ocontubernium th f n i e remaininth g putativ e barrack-blockseth f 33-8o 3m , assuming that there wer e normath e l 10 rooms, will have been 3-38 m, and the floor area of each room nearly 20 sq m, a not ungenerous allowance. However, this is all speculative and further discussion must await the excavation of the rest of the praetentura. finally ma notee yb t I d tha t seemi t s unlikely that ther spacs esimilarla i r efo y arranged barrack-block in the retentura: only 6 m separates the S wall of the commanding officer's house and the N edge of the intervallum street, insufficient for a barrack-block and a road, though a narrower stores building could have been squeezed in. However, this intervallum street may not be primary and the wider distance of 12 m between the rear wall of the commanding officer's rampare facth N f e o housth t would ean d have been wide enoug accommodato ht widm 6 e ea barrack-bloc fore gatth S t f move e So e roath :a th th f ed o o dk t an bordere N e pata th y do hb t poiny modificationso ma t t E metree w th fe o a st bein retenturae gth carrien i t d ou durin e gth fore th lif t f (Macdonaleo centrae th f do l1933 d rang presumablen s , E e176) wa e Th . y occupied by timber buildings as no stonework was found here in 1903. The available space E of the headquarters building, includin therd retentura,E an e s some, i g th m e5 1 measurey b m 0 3 s indication that the buildings were arranged running N-S. The buildings could have been store- houses and/or workshops. Conceivabl preciouo spacs s ya s 'halo ewa stw f barrack-blocks were placed here (Macdonald 1933, 176), but this seems unlikely: it would certainly be most unusual. discovere Th horsesf yo ' teet t Rougha h Castle (31 3121& unremarkabls )i officere th s ea s would have been mounted, and baggage animals may also have been present, but it does emphasise that ther alsy oema have been stables withi forte nth . cleas i t I r tha t leasa t t twomord an , e probabl centurieyx thresi e fourr eth o cohorsf f so o , VI Nerviorum were outposted. The outstations manned by these troops presumably included the adjacent fortlets at Watling Lodge and , but as each of these fortlets probably hel moro dn somn e thame e 2 soldiern3 s must have been stationed elsewher wells ea . CohorsI V Nerviorum havy ma e been divided betwee fortso ntw , possibly Rough Castl Falkirkd ean n i , the same way that it has recently been suggested that cohors IV Gallorum was divided between 28 | 2PROCEEDING S OF THE SOCIETY, 1978-80 Castlehill and Bearsden towards the W end of the Wall (Britannia, 9, 1978, 413). The pottery assemblage is clearly Antonine in character and there is no hint of an occupation at Rough Castle before the 140s though there are, as is to he expected in an Antonine fort, a number of sherds from Hadrianic-early Antonine vessels (59, 67, 71 and 78). The occupation of the fort, on ceramic evidence alone, continue othey s lonan da e Antonin r s th ga sit f eo e Wall, wit e possiblhth e exceptio Castlecarf no y (Hartle 2A)yg fi 1972 ., lato Ther t 28 numbe a ,e d e Antoninemi ar f o r e fragments and in particular five which are dated later than about 160, (75, 201, 204, 231 & 247; cf 18, 69, 209, 260-2). This suggests that the occupation of the fort continued through both the secone th firsd dan tAntonin e periods , abouo about 2 t14 t 163. Fou sherdm rri s (204), possibly all from the same vessel though found, unstratified, a little distance apart, represent the latest dated pottery fro sitevessel(se e mth .Th plaia s )i n dis anonymouf ho typa f s eo for hithertd man o considere havo dt e been introduced intsecone norte th oth f ho d towardd centuryen e sth t I . seems better, however brino t , g forwar arrivae date th d th f thif e o o l s typnorte f disth eo n hi rather than push backwards the date of the abandonment of the fort. Two phase buildinf so g have been praetentura.notee th n di firse tTh appear havo st e been o temporarfa ypresumabls naturwa d ean y succeeded afte greao n r interva n permanenta e th y lb t buildings. Three phases were noted in the intervallum street and on the W berm, while inside the rampartN d an sW there appea havo rt e been slight building lean-tf so o constructio latee th rn i days of the occupation of the fort. In only one area, the commanding officer's house, a building notably pron modificationo et 190e th 3 d excavatordi , s note 'evidenc f alteratioeo r additionno s mose th . t . complet . finished ean d workmanship sid sidy eb e wit hinferion thaa f o t r character' (Buchanan 1905, 481). t woulI foolharde db relato yt e threr theso o ee periodtw phase e th o sst recognise othen di r e Antoninth fort n o s e Wall. Whil s possiblei t ei , indeed probable, thae 190th t3 excavators remove remaine dAntoninth e th f so I occupatioeI praetenturae th n ni wholesale possibles i t i , , though unlikely, tha barrack-blocke th t s excavate 1957-6n di beed 1ha n occupied through both firssecone d th an t d Antonine periods withou breaka t . Indee case r occupatiodth e fo e th f no for Antoninn i t I resteI s solely upo ceramie nth e close th dat e cth f eth f evidenco e o s a d ean second Antonine perio Scotlann d i considere w no e onls di b yo d t about five years later than the end of the first phase it may fairly be questioned whether the date of the pottery can be so closely distinguished with certainty. This, however, is too large a problem to be discussed here. Accepting the pottery dating at face value the fort would appear to have been occupied through both Antonine periods, but the structural evidence for different phases of building cannot be relate theso dt e periods. The range of pottery from Rough Castle reflects the normal pattern for an Antonine Wall fort . samiae Mosth f o tn ware came fro Centrae mth l Gaulish centre manufacturef so e th t bu , East and South Gaulish potters are also represented. The Mancetter-Hartshill, Colchester, South Carlton, Corbridge and Carlisle kilns all supplied mortaria while two vessels (212 & 265, cf also 266 & 267) were probably made in Scotland. However, between 50 and 60% of the market was supplied by the potters working at Colchester, presumably shipping their goods by sea to Forthe th Colchestee Th . r kilns also provide dconsiderabla e quantit black-burnishee th f yo d cooking and table-ware. Black-burnished ware was also supplied by potters working in Dorset, at Rossington Bridge in South Yorkshire and at another, unidentified, centre. As Rough Castle lies towards the E of the isthmus it is to be expected that black-burnished ware 2 from the Col- chester areKentd aan , which would have bee ncoastE shippee th , out-numberp du s black- burnished ware 1, most of which would have been imported via the W coast; in fact the relation- vessee On f Sever o . l 1 : shin3 Valles pi y war bees eha n recognised (89). Amphorae fragments MACIVOR, THOMAS, BREEZE: ANTONINE WALL FORT OF ROUGH CASTLE | 283 demonstrate communication with Spain. Finally interestinn a , g fin glasa pars i d f o ts beaker (273) probably manufactured in Italy or Alexandria. Most of the pottery-and presumably other goods - which came to Rough Castle were probably imported by sea, either to the Forth or the Clyde, and then shipped overland across the isthmus. However, the pottery from Corbridge, perhapd an sEnglanf elsewhero Scotland S N d e dprobablan s th n ei ,wa y transported completely by land. artefact e basie th th f for e saiso e n th Littlb o liff t sdo n n ei fro eca 1957-6e mth 1 excava- tions. A quern is a not unusual find (14), and whetstones are to be expected (9-13). The range of metalwork (278-89) is uninformative. More interesting is the evidence for industrial activity fore parth r mbow a o n f i to l furnace (290 29f 292),c 1 & relate e aree b : burninthif y th ao d o st ma g on the inside of the N rampart. The recovery of 63 fragments of leather (315-41), including two almost complete shoesnumbea d an ,fragmentf o r textilef o s s (304-5) alss i , o noteworthye Th . leather fragments are mainly from calcei, the normal footware of the auxiliary soldiers in the Antonine period (Keppie 1975 textilee , th typa 78) d f seo an , relatively commo Romae th n i n world. Cattle boneg ,pi deed s (306-10ran , 313-14 commoe )ar n find militarn so y site reflecd san t variee th dRomae th die f o t n soldier (Davies 1971, 126-7) varietA . f timbeyo availabls wa r e locall woodworkinr yfo g and, presumably, burnin g- oak , birch, ash Scotd , silvean sr fi rpin e (294-303) - though of course not all the artefacts need have been made at Rough Castle. The presenc possibla f eo e offcut (295) suggests that carpentr practises yfortwa e th .t da Thi s object was of hornbeam, which would probably have had to be imported, possibly from S England: wooe havy th dema been importe statw alternativelr ra e o a n di offcu e havy yth ema t come from a re-used timber. It is unfortunate that most of the finds were unstratified.

ACKNOWLEDGMENTS We would than CartwrightJ r kM FergusonA Gillamr P MacleanM J ,D r r G M , ,M r M , Maxwell and Professor J Wilkes for help during the excavation; Dr K A Steer for advice; the Scottish Police College at Tulliallan Castle for providing accommodation during the excavation. We would also lik thano et followine kth g specialist r preparinsfo g BarberreportsW J r ,M : Mis CharlesworthD s I , DickinsonMis HartleyM W F Hartley R s B G K B r s r M , Mr M ,, Hodgson r ElizabetWilliamsF D , D BorthwicT r r M hM Wil .P d Slater J dan kr kindlM , y prepared the plans, sections and figs 12-14 for publication. Finally we would like to thank Mr Crude encouragemenH s hi S r nfo t throughou whole th t e operation.

ARCHIVE The finds will be deposited in the National Museum of Antiquities, the drawings in the National Monuments Record.

REFERENCES Anderson ,J 1905 'Notic potterye th f eo , bronz othed ean r articles foun Rougt da h Castle' Buchanann ,i , 1905, 489-97. Atkinson, D 1942 A report on the excavations at Wroxeter, 1923-1927. Oxford. Birley, E & Gillam, J P 1948 'Mortarium stamps from Corbridge, 1906-1938', Archaeol Aeliana, 4 ser, 26 (1948), 172-204. Breeze J 197D , 4 'The Roman fortle t Barburga t h Mill, Dumfriesshire', Britannia, (1974)5 , 130-62. 28 | 4PROCEEDING S OF THE SOCIETY, 1978-80 Breeze J 197D , 7 'The for t Bearsde a tsupple th potterd f yRomao e n an th o yt n Army' Doren i , & J , Greene, K 1977, 133-45. Breeze, D J & Dobson, B 1969 'Fort types on Hadrian's Wall', Archaeol Aeliana, 4 ser, 47 (1969), 15-32. Buchanan M 190, 5 'Repor Society'e th n o t s excavatio f Rougno h Antonine Castlth n eo e Vallum', Proc Soc Antiq Scot, 39 (1904-5), 442-99. Callender H 196 M , 5 Roman amphorae. Durham. Charlesworth, D 1959 'Roman glass in northern Britain', Archaeol Aeliana, 4 ser, 37 (1959), 33-58. Christison, D, Cunningham, J H, Anderson, J & Ross, T 1898 'Account of the excavation of the Roman statio t Ardoch'na , Proc Antiqc So Scot, (1897-8)2 3 , 399-476. Curie J , 191 RomanA 1 Frontier People,s it Post Fortd e an Th ofNewstead. Glasgow. Davies 197W R , 1 'The Roman military diet', Britannia, (1971)2 , 122-42. Detsicas, A P (ed) 1973 Current research in Romano-British pottery. London. Dore, J & Greene, K (eds) 1977 Roman pottery studies in Britain and beyond. Oxford. Farrar, RAH 1973 'The techniques and sources of Romano-British black-burnished ware', in Detsicas, A P 1973, 67-103. Frere, S S 1972 Verulamium excavations, Vol I. Oxford. Frere, S S 1977 ' in 1976,1, Sites explored', Britannia, 8 (1977), 356-425. Frere S 197 S , 9 'The Roman for t Strageath'a t Breezen i , J (ed D ,) Roman Scotland, some recent excavations, Edinburgh, 37-41. S Antoninee 189Th GA 9 Wall Report, committea y b Glasgoe th f eo w Archaeological Society, 1899, Glasgow. Gillam, J P 1957 'Types of Roman coarse pottery vessels in northern Britain', Archaeol Aeliana, 4 ser, 35 (1957), 180-251. Gillam P 196 J , 1 'The coarse pottery Steer n 'Excavationi 'A K , t Mumrillsa s Roman fort, 1958-60', Proc Soc Antiq Scot, 94 (1960-61), 113-29. Gillam P 197 ,J 0 Types of Roman coarse pottery vessels Northernn i Britain. Newcastle-upon-Tyne, ed d 3r . Gillam, J P 1973 'Sources of pottery found on northern military sites', in Detsicas, A P 1973, 53-62. Gillam P 197 J , 5 'Possible change coursconstructioe e plan th si th f n i eo Antonine th f no e Wall', Scot Archaeol Forum, 1 (1975), 51-6. Gillam P 197 J , 7 'The Roman fort t Corbridge'sa , Archaeol Aeliana, ser5 (1977),5 , 47-74. Gillam, J P & Mann, J C 1970 'The northern British frontier from to Caracalla', Archaeol Aeliana, ser4 (1970)8 ,4 , 1-44. Gordon 172A , 6 Itinerarium Septentrionale. London. . . Harden, D B & Price, J 1971 'The Glass' in Cunliffe, B, Excavations at Fishbourne, Vol II, The Finds. Oxford. Hartley R 197 B , 2 'The Roman occupation Scotlandf so evidence :th samia e th f eo n ware', Britannia, 3 (1972), 1-55. Hartley F 197 K , 6 'Were mortaria mad Roman ei n Scotland?' Glasgow Archaeol (1976)4 , J , 81-9. Hartley, K F & Richards, E E 1965 'Spectographic analysis of some Romano-British Mortaria', Bull Inst Archaeol Univ London, 5 (1965), 25-43. Hartley, K F & Webster, P V 1973 'Romano-British pottery kilns near Wilderspool', ArchaeolJ, 130 (1973), 77-103. Hogg, AHA 1968 'Pen Llystyn: A Roman fort and other remains', ArchaeolJ, 125 (1968), 101-92. Horsle yJ 1732 Britannia Romana. London. . . . Hull, M R 1963 The Roman potters'kilns of Colchester. Oxford.. Jones, M J 1975 Roman fort defences to AD 117. Oxford. Keppie F 197 J L ,5 'Leather' Robertsonn i , , Scot Keppid an t e 1975, 59-90. Kilbride Jones E 193 H , 8 'Glass armlet Britain'n si , Proc Antiqc So Scot, (1937-8)2 7 , 366-95. Macdonald, G 1931 'Names of potters on Samian ware found in Scotland', Proc Soc Antiq Scot, 65 (1930-31), 432-48. Macdonald, G 1933 'Notes on the Roman forts at Rough Castle and Westerwood with a postscript', Proc Antiqc So Scot, (1923-3)7 6 , 243-96. Macdonald G ,193 Romane Th 4 Wall Scotland.n i Oxford. Macdonald Curie & O 192 G , A , 9 'The Roman for t Mumrillsa t , near Falkirk', Proc Antiqc So Scot, 63 (1928-29), 396. Macdonald, G & Park, A 1906 The Roman forts on the Bar Hill. Glasgow. MACIVOR, THOMAS, BREEZE: ANTONINE WALL FORT OF ROUGH CASTLE | 285 Maxwell S 197 G , 2 'Excavation Romae th t sa n for Crawfordf o t , Lanarkshire', Proc Soc Antiq Scot, 104 (1971-2), 147-200. Miller N Roman192 e S , Th 2 fort t Balmuildy.a Glasgow. Nash Williams E 192 V , 9 'The Roman Legionary fortres t Caerleosa Monmouthshiren i ; reporn o t excavatione th s carrie 1926'n i t dou , Archaeol Cambrensis, (1929)4 8 , 237-307. Nimmo 177W , Historye 7Th f Stirlingshire.o Edinburgh. Oswald, F 1936-7 Index of figure-types on Terra-sigillata ('Samian ware'). Liverpool. Oswald, F & Pryce, T D 1920 An introduction to the study of Terra Sigillata. London. Peacock, DPS 1967 "The heavy mineral analysis of pottery: a preliminary report', Archaeometry, 10 (1967), 97-100. Richmond, IA & Gillam, J P 1950 'Excavations on the Roman site at Corbridge 1946-1949', Archaeol Aeliana, 4 ser, 28 (1950), 152-201. RCAMS 1963 Royal Commission on the Ancient and Historical Monuments of Scotland, Inventory of Stirlingshire. Edinburgh. Robertson, A S 1961 'Excavations at Raeburnfoot, Eskdalemuir, 1959-60', Trans Dumfriesshire Galloway Natur Hist Antiq Soc, 39 (1960-61), 24-49. Robertson S 197 A , 5 Birrens (Blatobulgium). Glasgow. Robertson, A S 1979 The Antonine Wall. Glasgow. Robertson, A S, Scott, M E & Keppie, L J F 1975 Bar Hill: a Roman fort and its finds. Oxford. Roy 179W , 3 Military antiquities Romanse oth f Northn i Britain. London. Sibbald J , 1707 Historical inquiries. Edinburgh. Simpson, F G and Richmond, I A 1935 'The Turf Wall of Hadrian, 1895-1935', J Roman Stud, 25 (1935), 1-18. Stanfield, J A & Simpson, G 1958 Central Gaulish potters. London. Steer A 196 K , 1 'Excavation t Mumrillsa s Roman fort, 1958-60', Proc Antiqc So Scot, (1954-6)8 8 , 208-21. Stevenson, R B K 1976 'Romano-British glass bangles', Glasgow ArchaeolJ, 4 (1976), 45-54. Stuart R ,185 2 Caledonia Romana. Edinburgh. Webster, G 1944 'A Roman pottery at South Carlton, Lines', Antiq J, 24 (1944), 129-43. Webster, G 1969 Romano-British coarse pottery: A student's guide. London. Webster, P V 1977 'Severn Valley ware on the Antonine Frontier' in Dore, J & Greene, K1977,163-76. WheelerWheeler& M E , ,TR V 1928 'The Roman amphitheatr t Caerleon'ea , Archaeologia, (1928)8 7 , 111-218. Wild, J P 1970 Textile manufacture in the northern Roman provinces. Cambridge. Williams F 1977 D , a "The Romano-British black burnished industry essan a : characterisation yo f no heavy mineral analysis' Peacockn ,i , DPS (ed) Pottery, Early Commerce and Economics: Characteri- sation Traded Romanan n o Laterd an Ceramics. London. Williams F 1977 ,D b 'Black burnished ware from Mumrills re-appraisa:a sourcef lo heavy sb y mineral analysis' DoreGreenen d i , an J , 1977K , , 177-87. Williams F forthcomin D , g 'Black burnished ware from Bearsden' Breezen i , , ExcavationsJ D , t a Bearsden, forthcoming. Wilson R 196 D , 9 'Roman Britai 1968,1n i , Sites explored', / Roman Stud, (1969)9 5 , 198-234. The Society is indebted to the Scottish Development Department (Ancient Monuments) for a grant towards the cost of publishing this paper Editorial note: circumstances beyon editor'e dth s control have resulte platee th thi r n dsi fo s articlef beino t gou sequence.