Excavation Romae th t sa n for Crawfordf o t , Lanarkshire by Gordon Maxwell

INTRODUCTION The existence of a Roman fort in the neighbourhood of Crawford, although suspected by General Roy,i was not proved until 1938 when excavation carried out by Dr J K St Joseph2 on a site lying about 400 yds N of the village on the right bank of the Clyde put the matter beyon doubtl d al for e tTh . (NG 954214S RN ) occupie smoderatela y strong positio narroa n no w plateau of hard glacial gravel protected on the south by the Clyde, and on the E and W by the Camps Wate Berried an r s Burn respectively; site acces th ewoul N froo t se mth d have been impeded in Roman times by marshy ground (fig 1). Strategically, however, it was of great importance. At this point the Roman roads from Annandale and Nithsdale met, the latter probably crossing the Clyde to the SW of the site; the route then left the valley of the Clyde, avoidin e gorge-likgth e defile between Crawfor Abingtond dan climbed an , d northward over Raggengile th l Pas rejoio st f Coldchapel Clydo e nth S e e for th e jussitin e th o t t t f Th .musg o t also have been influence presencmuce s da th areconsiderable a y hf b th a o en i e native populationJ e nee th observo dt y b s a e regular intervals betwee e garrisonnth s guardin e Romagth n road notee networkb y d thama importance t I th .t stils site elth wa recognisef o e medievan di l times when Crawford Castle, originally a seat of the Lindsays, but later ceded to the Douglas family, Romae th f o nS site.e th 4wao t Doubtles s builyd 0 5 t s from this time onwar usefors e dth dwa t quarrconstructioa e s th a r yfo associates castlrepair e it n o th d f ean o r d buildings evidence .Th e for prehistoric use of the site is discussed below (pp 187—8). In the spring of 1961, with the assistance of Dr Anne S Robertson, the author supervised triaa l corneforexcavatioe W traininpara s th a S tf f o te o r th gn n i cours techniquee th n ei s of archaeological investigation. Participant thin si s course were drawn from senior pupils attend- inFife gth e County Schools Easter Cam t Abingtonpa resulte firse th Th .t f season'so s work wer promisino es g that excavation continue sam nexe e yearo th th tw tr en sdo limitefo d scale, although now with the assistance of the senior pupils and staff of Lanarkshire Education Authority schools. In 1964 the author joined the staff of the Royal Commission on the Ancient and Historical Monuments of Scotland and was authorised to continue the excavation of the site as part of the preparatory field-work for the forthcoming Inventory of Lanarkshire. The primary object of investigatioe th defencess recoveo it t fors d e ascertaiplae d th an ntwa rf th an ,no n whe site neth was in use. However, the fort at Crawford is of such an unusual size, smaller than most forts thaassociatee ar t d with quingenary cohorts, t mucye hd biggean 5 r tha fortletna s , thawa t i t also considered desirable that at least a little should be learned about its internal layout. Accord- ingly excavatio continues nwa d unti autume th l f 1966no , usually wit assistance hth f felloo e w officers of the Commission, but in the final season with a team of diggers drawn largely from the 148 I PROCEEDINGS OF THE SOCIETY, 1971-2

FIG 1 Location maps MAXWELL: EXCAVATIONS AT THE ROMAN FORT OF CRAWFORD, LANARKSHIRE | 149 Extra-mural Class in Archaeology at Edinburgh University. The present report is thus the result weeks7 1 f o ' fiele worth d n ki wit hsmala l labour force spread ove years x indebtem si ra I . d Commissionere th o t permissior sfo publiso nt materiae hth l derived fro e lasmth t three years' wor advancn ki e abbreviateth f o e d descriptio e wholth f eo n excavatio e e giveth b n o ni nt Inventory.

THE EXCAVATION Although the defences of the fort are much denuded, it seemed, on superficial examination easgroundn e a th ye f b matte o o t , defino rt e their limits positioe .Th fouf no r anglefore th t f so appeared to be clearly visible and the artificial scarping of the edge of the plateau was very conspicuous; however, it was soon discovered that these were only relative guides to the position of the defences. The fort had in fact undergone at least two major alterations, and the visible quadrilateral platform s describea , d above e sideE th line d f th s,o e an represente N e th n do intervallu cresme th lase stree t th tlin W f periodd o te appearean S , e whilth havo dn t eo e been altered by recent agricultural operations. It was further discovered that the depth of the defence works was as much as 190 ft in places, a remarkable extent for such a small fort. Consequently impossibls iwa t obtaio et ncompleta sidesE d , an whereS sectio defencee e eth th n beend o sha n interrupte r obliteratedo modera y db n road medievae th , outbuildinge l castlth d ean f Castleso - mains farm available ,th hal whilf W o fe eth groun encumberes dwa d wit hkitchea n garden severad an l trees t wasI . , moreover, difficul folloo t t we th wit t hproba bu coursy e eth an f eo outermost ditches s theia , r profile beed ha sn obscure deliberaty db e re-fillin r re-cuttingo n gi Roman times. Nevertheless it proved possible to establish that there were three major phases developmene th n i fort'e th f so t defence system firse th t n phasI . fore eth t measure t frof 0 md38 t transverself 5 22 y b yW oveo t tura r E r earto f h rampart abou thicknesstn i 1t 8f . Immediately beyon ditcheso dtw thiinnee y th la s, r being abou thirt A widef . e oute8 dft t th ,0 ditch 1 r , measuring widthn aboui separatet s f 8 twa , frontdN e froth , mn o thes distanca t f y 2 eb 3 f eo wher mor, whern ra W r lest ee i o e sth parallen innee o t th f mucs ro 5 t a l 8 ditches y s hb a d an , it followed a divergent course. secone th n I d phase ther practicalls ewa alteratioo yn thicknese th rampare n ni th f so r o t the area enclosed, but the ditch system was subjected to a fairly drastic overhaul. The inner ditch was re-cut to approximately the same width and depth as before, but the outer was com- pletel awayg ydu , being replace ditca y db h measurin widthn i t f 0 :g2 thi s seem havo st e been of normal V-section except in the southern half of the W front where a mid-rib was left standing by the Roman diggers. The outermost ditch was re-cut and although its width seems to have varied generan i ,widenes wa t li abou. o dt ft 4 1 t In the third phase the defences were largely re-cast. Although the length of the fort remained the same the width was increased by filling both the inner ditches on the N and S fronts and building a new rampart on top of them. The external dimensions of the fort were now 300 ft from N to S by 385 ft transversely. The outermost second-period ditch was re-used, although beyon g distanca t du ditc a s w t d t i requirabouf hne wa et o re-cut f e no a b 0 d d 4 to e t di an ,t i frontN e th , We however n ditcth o ;w n ne ho e divergeth , d fro oldee mmucs th a rs untihwa t i l as 85 ft to the N of it. Near the NW angle the space between the ditches was apparently filled with obstacles. The detailed evidence for this interpretation of the fort's history follows immediately. 150 ( PROCEEDINGS OF THE SOCIETY, 1971-2

Northern Defences , pi5 s d 14a-15a(figan 4 s ) reasone th r sFo stated abov onle eth y s thi sid whicswa n eo t provehi d possibl obtaio et n a continuous section through the defences, and even here the task, already formidable by reason of its sheer magnitude - the total length was over 170 ft - was made more difficult by persistent weatherd ba . MAXWELL: EXCAVATION THT SA E ROMAN FOR CRAWFORDF TO , LANARKSHIR1 15 | E sectioe th e intervalluf o nth d Firsten S m Period. e e firsstreeth th t t f A o periot s dwa preserved for a length of 10 ft. It consisted of a 6 to 9 in layer of natural gravel overlying two bands of grey and yellow clay - an unusual but not unparalleled arrangement in the construction of a Roman road.<> Of the composition of the rampart which stood to the N of this intervallum road, very saie b onls littl da reae n yth eca r parsurviveds ha t thad an ,t onlheigha o yt . f abouo tin 0 1 t materiae Th lfina uses e dbrowwa n earth which subsequently slippe dcass backtwa downr o , , to cover the N half of the road. However the front of the rampart, which was removed at the beginning of the second period of occupation, was probably built of clay, and the total width will have been in the neighbourhood of 18 ft. About 24 ft to the N of the outer edge of the intervallum ditcstreea s hwa t whose counter-scarp face exhibite unusuan da l step indication a , n that the ditch might have been re-cut. Although the stratification of filling material in the ditch t immediatelno d di y confirm this hypothesis e evidencth , e recovered from S sectione th n o s mattee frontth W t d rpu s beyonan d doubt (see infra, 155)p thin I . s cas ditce mosd eth hha t probably been re-cut, but in the second period of its use, no doubt as a result of routine cleaning operations, the soft clayey filling of the first ditch through which the scarp face of the second was cutbeed profil,ha d n ol remove e eth revealedd dan ; originall beed ha n t y i V-shape measured dan d abou firse t deepf t th t14 t wid 0n f perioforI e .d th secona ean t f do d ditch f similao , r shape tract o remainei n f t o sectioe bu th dimensionsd , n di N nan e becausth o about y t la e,a f 6 t much larger ditch was dug in approximately the same position at the beginning of the second period. The early outer ditch survived only at the gateways of the succeeding fort where the courslargee th f ero ditc broke s entrancen a hwa r distancna t fo beyonf t 9 A . 2 largee f dth eo r ditch lay the outermost ditch of the first period, originally measuring about 7 ft by 3 ft; in the intervening distanc land-surfacd ol e eth coveres ewa d wit hlayea cleaf o r n yellow gravel which varied from 12 to 18 in in thickness, and it is reasonable to assume that the gravel was obtained in the course of cutting the ditch system of the earliest fort. When it was silted up to a considerable depth thire ,th d ditc alsd hoha been re-cut thin i st cas re-cuttin,bu e eth g appear havo st e damaged the earlier profile only slightly. Second Period. secone Th dreconstructioe phasth w esa rampare th f no approximateln i t y the same position as the first, but we can say very little about its composition as the re-casting of the northern defences in the third period entailed the levelling of the Period II structures. It seems probable, nevertheless, from the large amounts of clay and peaty turf found in innee th file f r o secone lth ditc th f ho d forte periol sideth al ,f n so thad o t these materials were used extensivel uppee th n yri ramparte partth f so . Abou lowemore e b tth n er ca certainpar e w t . The outer edge of the earlier rampart was cut away to allow the insertion of a roughly kerbed stone foundatio t broadf 5 n ; trace f claso y intersticee founth n i d f thiso s foundation suggest tha t supportei t dclaa y chee r revetmentko positioe rampare bace th Th . th f ko f no t coult dno precisele b y located widta t f abou t bu hseemeo ,f indicatede 1 b 2 t o dt intervallue Th . m street was resurfaced with a compact layer of cobble and gravel measuring 6 in in depth and extended . BeneatN e th hfurthea o t thi t f s3 rlaye seales wa r ddenariusa f Nero o 65-8 datinD A , o t g but whether it was dropped in the Flavian period or at the beginning of the Antonine, it is impossibl telo et l (se e183)p . rampare th Immediatel f inneo e th tN re firs e ditcth th t f o yperiot ho re-cus dwa t leaving a distinctive step on the counter-scarp face, and beyond this in turn there was a ditch, some 16 ft wide and 7 ft deep, with a slightly sagging V-profile. The gravel derived from this ditch spreas fairla wa n di y uniform layed r extendinan N e th go t fror abou t outee f fo m th 2 p 2 tli r immediately overlying another level compose f threo d e separate elements, peaty turf, dense 152 I PROCEEDINGS OF THE SOCIETY, 1971-2

•'mm

£ 5

grey clay, and a confused mass of dark soil, greyish clay blocks and charcoal. Although these elements were quite distinc cleas wa r t tthai t thebeed yha n deposite same th et da time frod man , their natur t seemei s probable tha l deriveal t d fro earlien ma r turclad yan f rampart demolished at the end of Period I and cast forward over its own ditch system. The band of mixed clay and dirty gravel which extende thire froth d N df oute e o mditc th p rli h represent upcase sth t obtained afte re-cuttina r f thago t ditch choke s whewa t ni d with silted material. MAXWELL: EXCAVATIONS AT THE ROMAN FORT OF CRAWFORD, LANARKSHIRE | 153 The upcas e pilinoutee th bana th mad p f gn li ro n i thirte k o eth d ditc hslightla y more formidable obstacl s seconit n ei d phase firsts thait a n ,n i d witan hdepta n i f abou6 ho t f 3 t more widtOn e f . 1 ho 1poinft t that migh made b t re-cuttine thats ei th outermose n i ,th f go t ditch profile th ,bee d eha n slightl ylongeo n altered s trura wa t eI . V-shape innee th , r face being considerably longe rbotto e tha th outee nd th m an r more rounded appearance Th . rathes ewa r more 'Punic' than previously. Third Period. A drastic overhaul of the N defences now took place. The rampart of the earlier phase was levelled, and the presence of a layer of carbonised debris and large angular blocks of stone immediately above the southern half suggested that the extra space thus obtained wa erectioe s th use r dfo f buildingn o s (see infra, 176)p innee Th . r ditc completels hwa y filled with peaty turf and a mixture of clay, small stones and earth, and then given a levelling layer of yellow gravel, probabl frog uppee ymdu th r scar largpe facth ef e o ditch , whic alss howa carefully filled, but in a rather distinctive way. To begin with, a small platform of stones was built in the bottoe ditce hal th abov N d th f fo hmf an o e thicoursen ste turf so f blocks were block e laidth f I . s originally measured 6 in in thickness, the standard measurement indicated by Vegetius, the top of the pile of turves would then have been on a level with the lips of the ditch. A thin layer of spreaw graveno ds ove wa l turvee corduroa th rd san y bas f birceo h branches (pi 14) carefully laid alon reae gth rpile.e edg th t thi f A eo s poin rearware th t ditcd e filleds halth hf wa fo , firstly wit hlittla mose eth r gravelt parfo t t bu ,7wit h irregularly chopped block gref so y clay untie th l level of the top of the turf stack had been reached. On this carefully constructed base the rampart of the third phase was now erected, measuring about 20 ft in width, the front 7 ft or so being resmade cla f earthstrikind th a to tur f ys d t uniquean i o t fan .no I gf i e exampl engineerine th f eo g skilforesighd Romae an l th f o tn army materiae Th . l use r fillin dfo reae gth ditce r th hal hf o f had naturally come from the demolished rampart of the second phase, over whose levelled remain intervalluw ne a s m street constructedw , no about widef s 8 1 o sherdwa t, a Tw .f o s cooking-pot in black-burnished fabric and the base of a samian bowl (f. 37; see infra, p 190) were bottominge founth n streete i d th f o ; both assignee vessel b middle e n th th ca s o f dt eo second century AD. Between the S edge of the third intervallum street and the N edge of the second there was a gap of about 9 ft, across which was spread the scatter of angular boulders mentioned above. It should be noticed that there was little or no silt in the bottom of either innee oth f r ditches when they wer painstakinglo es ye Roma filleth y b dn engineers othen I . r word t seemsi s unlikely thasecone th tthir d dan d phase t Crawforsa d were separate lona y gdb interva timef o l . The third ditch of the earlier periods now became the inner ditch of the last, separated from its rampart by a berm 25 ft wide; in all likelihood there was as little silt in this ditch as re-cuttino n d an innee igo nth woultw r d have been required t wasI . , however, necessaro yt maintain the depth of the defences by digging another ditch, a major obstacle, measuring 15 ft in width by 5 ft in depth. In the platform about 35 ft broad, which was left between the two ditches, there were traces of a system of shallow pits and a gulley (pi 15a), varying between 9 in and 1 ft in depth, which ran parallel to the outer ditch and may therefore be associated with time th en availablI . impossiblit s wa t ei exploro et e this feature seemt i full t unreasonablt ybu sno e to interpret it as part of a series of obstacles designed to break up an attack on the fort from this quarter. Such features have been note t Rougda h Antonine Castlth n eo e Wall mord an e recently at Gatehouse of Fleet and Glenlochar in Kirkcudbright.89 It was noticed that the outer- most ditch was set at a slight angle to the other three ditches, which were all parallel. Probing established thasame th t e divergent cours followes t a ewa , r abouE dfo e th t anotheo t t f 0 5 r which point it stopped. Subsequent excavation here showed that there was a gap, c 25 ft in L 15 | 4PROCEEDING S OF THE SOCIETY, 1971-2 width a heavily-metalle r fo , d roadway, whic d apparentlhha y issuee dth gat N fro f e o mth fort heading in a north-westerly direction. The course of the road beyond this point was not traced little b t ther en bu , doubeca t tha t continuei t ford e dth an t curvin f o d gen rounW e dth finally led to a ford or bridge across the Clyde not far from where the modern road-bridge outermose th p ga e tsid th E ditc seef standss e o havho th nwa t n eO . bee t forwarnse furtheda r 10 ft, although probing revealed that its alignment ran roughly parallel to that followed to the gape th W.f o Periof o d I bot dII en ditchee he th th t s A were allowe filo dt l with natural silt s site th e;wa disturbed by no subsequent occupation in the Roman period, although a scatter of small frag- ments of green-glazed pottery found above the fully-silted outermost ditch testifies to activity medievae th n i l period. The Western Defences (figs 3, 4 and 6) On this side, although it was impossible, for the reasons stated above, to cut one continuous section throug defencese hth completa , e profilreconstructesystee e b th n f mo eca d froo mtw complementary trenches picture .Th e which emerges afte studa r thesf yo e section significantls si y differen structuras it n i t l detail from that drawn above, althoug historicae hth l interpretatios ni exactl samee yth . First Period. On the W, as on the E front, the ramparts of the three periods occupied approximatel same yth e position, althoug t appearhi havo st e been necessar eacn yo h occasion to reconstruct the rampart's outer face. Section B (fig 6) shows clearly the resulting complexity of structure. The rampart was basically tripartite. The rearwar10 d 10 ft, which consisted of dark peaty soilsubdivides wa , d introughlo otw y equal parts 'keyed' into each other t werei s e a , th , forward half being lighter and more obviously layered, the other heavier, wetter and much more homogeneous; bot immediatele Romaa h li f o p nto humun yo s level varying6 froo t m9 thicknessin thini E rampare se parth th f n lowese o tO .th t f abuto t interd sen agains- W e th t vallum road, which is about 16 ft wide and has a maximum thickness of one foot; the tail of rampare th overlais uppere twa th laterd y nb an , , roade level rampare lowesth e f th Th .s o d t an t road level were thus clearly contemporar belonged yan earliese th o dt t perio f occupationdo , for which a date in the late first century AD is suggested by the fragments of a ring-necked flagon and a mortarium of that period which were found immediately on top of the lowest intervallum road (see infra, p 193). The eastern limit of the intervallum road is marked by a fairly shallow slot, 1 ft 6 in wide and about 1 ft 3 in deep, in which stood the W wall of Flavian Building VI; beyond this to the E extended the disturbed clay-and-gravel floor level of that building, itself overlain by the yellow clay floor of the Antonine Building IV, the position of whose W side-wall was indicate severay db l dislocated building-stones. impossibls i t I certaie b o et n abou precise tth e widt thif ho s rampart sinc entire eth e forward section was removed in the course of reconstruction at the beginning of the second period; the matte s furthewa r r confuse discovere th y b d y thae sectioth t d bee t acrosnha ncu timbera s tower, measuring approximately 11 ft square. The corner of the pit in which one of the tower's four uprights had stood is shown in the section, overlain by the band of clay used in the front face of the second-period rampart. However, in the course of further investigation of the tower structure, which is discussed below (see p 163), a discontinuous layer of grey and white clay was discovered immediately abov e post-hole fillin t th beneathth e f bu o g t e secondarepi th y face, suggesting that the original rampart had extended at least as far as that point with an outer face or revetment of clay to ensure the stability of the looser peaty .earth core. The width of the whole rampart was probably not more than 18 ft, as on the N front. At a distance of MAXWELL: EXCAVATIONS AT THE ROMAN FORT OF CRAWFORD, LANARKSHIRE | 155

I D,

O

outee th f rintervallue o jusedgth W f t eo t f ove 1 m2 r sectioe streeth n i tn illustrate6 g fi n di ditca ther s hV-shapes depthewa n i measurin t wa f t 3 widtI n mor.d i d t an f ehan g8 than half filled with homogeneous heav t peaywe t above which ther beed eha n lai ddensa e gravel- and-clay layer abouotheo thick n froni tw W r9 t t e pointA .t th wher n so e sections werg edu the evidence was different (section C, fig 6; see fig 3). In both cases at this point a ditch measuring deptn i beet d f hwidtha n 4 i n t d fillef han 0 1 d fro o witt m9 h grey clasubsequentld yan y re-dug 156 | PROCEEDINGS OF THE SOCIETY, 1971-2 counter-scare th i nthan y i suct notca twa cu h a r ste s ho primarpe p wa th fac f o e y ditcht A . some time later this secondary ditch was filled with the same kind of homogeneous black fill noticed in the ditch of section B. It is therefore extremely probable that the innermost ditch in sectio belongenB secone th o dt d perio f occupatiodo re-cuttino n d nvisibls an gwa e becausn ei this area the earlier ditch had been interrupted at an entrance (see infra, p 163). Elsewhere on the western front, however ditce th , h pursue same dth e course relativ earle th yo e t d rampar ha t i s a t on the N. Of the early outer ditch which lay about 6 ft further W there was no trace in sections B and C because it had been dug away in entirety by the large outer ditch of the second period, and fron t surviveW i e t th onl lefs n do ty wa opposit wherp ga ea portae eth decumana seconde th f o - period fort. Here its measurements were 10 ft wide by 3 ft 6 in deep, and it had filled with natural silt to the depth of a little more than 1 ft 3 in when the first re-occupation took place. The examination of the western ditch-system could be continued only as far as the outer large th lief p o second-perio d ditcsequence th section t hi bu , en B coul followee db d furthen i r section C (fig 6). Lying immediately above Roman humus level and extending for a distance f abouo t 1 t beyon6largf e th f eoutee o dsecondar th p li r y hara ditc s dh wa yello w gravel layer, som thickn i e9 , whic doubtless hwa s derive earldo frotw y upcase e minnermosth th f o t t ditches. This featur notices northere eth wa n do n front thickewhers wa extended t ei an r d 1t further3f . It is interesting to note, however, that the Roman humus layer, which is of fairly uniform 3 to 4 thicknesn i s beneat gravele hth barels i , thicn i y2 k indicatione f immediatelito th : W e th so yt were that the gravel had been shaved off at this point for a distance of a little more than 10 ft, before re-appearing, less homogeneou loosed san r than before t stilbu ,l unmistakeabl same yth e yellow gravel. It had most obviously been removed from its former position and cast forward, spilling ove innee th r r lianothef po r innermosditce th f ho whic W abou y tt f e hditcla 0 th 6 tf ho first period ditce V-shapes .Th hwa profilen di , measurin depthn i n i 9 ,g t widtabouf n i 2 t f y hb t7 and a glance at the section sufficed to show that it had been deliberately filled with shoots of cladirtd yan y gravel, throw fro n side i e wit d mthicon e ,an h th k laye yellof ro w gravel, mentioned above, thrown in from the other; beneath this there was a thin layer of silt, at the most about 2 in deep. Clearly the ditch had been filled while the fort was still occupied, or not long after abandonments it ditches a thi,n i t s typscu f subsoieo l would quickly have become choked with silted materia the d t beeha ly no n subjec regulao t t r cleanin attentiond gan . At a distance of 1 ft 6 in to the W, there lay yet another ditch 7 ft wide and 2 ft 9 in deep. earln a t y A excavatioe stagth f uncertaies o wa t ni n whether botthesf o r onlhe o e yon belonge d firse th t o periodt t wili rememberee r b l fo , same fronth N en e d i trelativ th tha n o t e position, there was only one ditch, re-cut at the beginning of the second period. Subsequent investigations, however, showed tha easternmoste th t , whic beed hha n deliberately filled5 , belonge firse th t o dt westernmose perioth d d an seconde th o t ; althoug t thiha s particular o poin tw ditche e e tth th f so periods pursued a separate course it was apparent that, in general, they were wholly or partially concurrent wils seee A . b loutermose nth fro9 g mfi tfirs e ditcth t f perioho d runs parallee th o t l rampart on the N but to the S of the NW angle it diverges until, at a point where it meets the steep bank bordering the flood plain of the River Clyde, it lies at a distance of more than 110 ft beyon e outedth first-perioe r th fac f o e d rampart e areTh .a thus enclosed appeare t firsda t sight oddl ytemptins irregularwa t i interpreo gd t an , definins a t ti g eithe annexn rearla n a r yeo temporary encampment t sooi t nbu ; became clear tha ramparo n t r palisado t eved eha r been associated with the ditch nor was there any sign of occupation immediately within the area enclosed by it. It must therefore be considered as an additional line of defence belonging to the early fort, similar in its irregularity to the defences of the fortlet at Haltwhistle Burn,11 on the Stanegate, and to those at Greensforge, Staffordshire.12 Wood-

Peaty turf' 'Clay

Perio I dramparII t

Blocks of clay, soil and debris

Pit, projected Clay' Clagraved yan l

North Channel

5 Ft. 10 20 30 40 , , P , , , ....., . 6,., 0 1 I i •" i i M 10 20 Fto 5 Northern defences: section A East

West

Grey clay Dry peaty turf and soil Moist peaty turf post-hole pit Period I intervallum street Perk d anI ramparI dII t Period II and III intervallum street

I East

Mixed clay and occupation debris

^Period III rampart

blocks West

Ft. 10 20 30 40 50 60 M 10 FIG 6 Western defences: section B top, section C middle and bottom MAXWELL: EXCAVATIONS AT THE ROMAN FORT OF CRAWFORD, LANARKSHIRE | 157 It only remains to note that at the end of the first period both the inner and outer ditch seem to have been deliberately filled. This apparent slighting of the defences may well be associated with the level of disturbed material discovered immediately above the earliest inter- vallum road in section B. On top of the W half of the road there was a mound of stones and yellow gravel slightly overlyin rampart e bace th g th f ko , whil reae eth r buriehals fwa d beneath layea soff ro t dark gre yfooe claon t ydeepo t fro n i m,4 whic clearls hwa y deposite same th t eda moune timth s egravelf a do firse th t, f periopresumablo d occupationf en do e th t ya reasoe Th . n r sucfo h destructio difficuls ni determineo t t pile gravef th eo t bu , l suggests that possibly timbers were being dug out of their post-holes and the clay can only have come from a disturbance in ramparte bode th th f yo . Second Period. alteratione Th westere th n so n secone fronth n i td period were similao t r those on the N, as section B clearly reveals. Above the disturbed material on the lowest inter- vallum, layea ther s browf ro e wa n gravesoid an l l cappe bottomine th y db upped gan r cobbling e seconoth f d intervallum street. This leves coverewa l d wit hthia t persisten nbu t scattef o r charcoal and it seems likely that it derives from an oven situated nearby but not actually located in this section. On the surface of the charcoal spread there was discovered a small fragment of the rim of a dish in black-burnished fabric belonging to Gillam's category 1, datable to some time between AD 140 and 170, with a bias towards the earlier limit.13 The road, whose width was almos , overlaft originae 7 1 reate th yf th ro l rampart re-buildine th n .I secone th f go d period, however, the front of the early rampart had been removed. Although the junction between the materiaw ne difficul s d wa l an ramparte locatbodo e d t th t th ol f yo n ei , since both consisten di middle th e sectio f similarlno y laminated peaty earth earliere th ,rathes wa ra moistef o d an r darker colour alss owa possiblt ;i where se Romae o eth t n humus leve beed intt ha ln cu o during the reconstruction work. The new material contained a fragment of a samian cup (f. 27; see infra, p 191). rampare Th re-faces wa tprobabla o dt e thicknes f abou so , wit ft 6 ht yellowish-grey clay, which served as a revetment for the less stable earthen core; a spread of greyish gravel from 4 to 6 in thick was laid just inside the lip of the inner ditch to receive this revetment. The width of whole th e rampart will have been abou. ft 7 1 t verA y short distance outsid outee eth rrampare edginnermose th th f e o y second e tla th f o t - period ditches; this has already been described and it only remains to point out that the ditch beed ha n deliberately filled with dark peaty turf befor sily tan e whatsoeve collected ha r s it n di bottom. Two feet beyond lay another, much larger, ditch measuring 20 ft between the lips and depth n middl e i coma ditce th s n th i a maximu r spinn f hbi 6 ;o wa e o t f e 4 aboun i f m t o f 2 t height. It could be seen that the ditch contained a deliberate filling of blocks of turf and clay thrown int unsiltee oth filline th d; g ditcin containe6 ht f dept a fillin 2 o t f hdo t e gi parth f o t rim of a second-century mortarium (see infra, p 193). In section C the sequence was broadly similar, although there were two interesting differ- ences in detail. Firstly, the innermost ditch could be seen to have been the result of re-cutting, and secondly the profile of the large outer ditch had been more or less V-shaped - more regularly than might appea t firsa r tuppe e sightth r innee rt awa fo ,th hal cu f ditce ro ys fth fac f hwa o e in the re-modelling of the defences at the end of the second period. The original width of the outer ditc aboudepts s it hwa filline d ftht6 th ;an 1 t 7gf comprised block claf blacd so yan k peaty naturao turfn therd s an , ewa l silt certaiA . n quantit f occupatioyo n debri mingles swa d with the turf and clay fill, the most interesting items being a shoe, in a tolerably good state of repair, and the head of a wooden mallet (see infra, p 188). The moist conditions had also preserved thif o sp in-filto e lth mucn o h humbler items, suc tins ha y twig wood san d chippings, representing, 15 | 8PROCEEDING S OF THE SOCIETY, 1971-2 doubtless waste th , e material from building operation beginnine th t thire sa th df g o period f O . more direct importance, however discoverye th clae turd s bode th y th an fwa f , yo n filli f , o , wall- and base-fragments of a cooking-pot in black-burnished ware. There was an interval of approximately 47 ft to the W between the outer lip of the large ditch and the inner edge of the outermost. Across most of this there was spread a confused mass of clay, gravel and occupation debris (i.e. charcoal and burnt daub) which probably derived from the defences and interior of the early fort and was deposited at the end of the first period beginnine th r nexte o datablth o f N .g o e materia recoveres wa l d from this level t fragmentbu , s orusticatea f mortariud an r dja m stamp (infra, 192-3)p p , foun loosn di e soil immediatelo yt , probablW e th y deriv eindicatd froan mt i s Flaviaeit n origin. Abov densa t ei e laye f yelloo r w gravel, representin up-case gth t fro largese mth t second-period ditch beed ,ha n sprea maxia o dt - distanca t beyonf r 6 fo 2 outee muf d th eo an mr lipn i dept .9 Approximatelf ho f o W t f y2 the outermost ditch of the Flavian period, which was now almost completely choked with a layer of silting above the deliberate infill, there was dug the last ditch of the second period. At this point it measured about 7 ft wide and 2 ft 9 in deep, the slope of the counterscarp being twice as steep as the scarp; the filling consisted of natural silt. When the ditch was sectioned in trenches situated to the NE and SW it was found to be slightly broader and deeper, having been re-cut through the fill of the earlier ditch. Its course was in general the same as that of e outermosth t Flavian ditch sectioy (fi b ) excepare e g4 t , th wher acu nC n i te secondar eth y ditch continue e alignmendth flattenee th f o t cornedW r anothe fo rt beyon f 0 4 rearliee dth r turning-point before itself . inclininS e th o gt Third Period. significane Th t lase differencth t n i W e d betweean N defencee ne th th n o s period of occupation is that the rampart of the western front occupied the same general position as its two predecessors, whereas on the N the whole system was moved forward almost 35 ft. t woulI d appear tha intervallue tth m lase roath t f periodo largels previoue dwa roae th yth f do s period re-used and, where necessary, repaired or patched, but the upper levels, being so close presene th o t t field-surface, have been partially disturbe ploughiny db evidence th d gan e cannot consideree b d definite. Fortunatel picture yth mucs ewa h cleare othen ri r respects. The second-period rampart was given a new outer facing of white clay measuring about 5 ft in thickness, to receive which the earlier clay front was cut away and a single line of stones inserte outen a s da r kerb (see sectio 6)g ;fi , elsewhernB banea f boulderdo t leas sa t wid f 5 t e was regulalaia s da r foundatio outee th r r nfacfo e (see finawidt e e sectioth Th lf h. ramparo nC) t littla innee s eTh wa under. ft larg 0 2 re outer ditches werpackew eno d with dark black peaty tur blockr o f claf so y (the sourc thif eo s material presumably facinseconde d beinth ol f e go g th - period rampart), and across the old innermost ditch a layer of gravel was cast, thus providing hara d berm rampart w extendinne e fronsection n th i I . r 1 f t 1gf o fo t thinC sg gravedu s wa l innee th largf e ro th fac t cass ef eo ou ditc twa som d forwart i h an f eo d acros ditce sth h itself, sealin clae turd gth y an f fill .need o Thern s , ewa however tako t , e elaborate precautions with the fill of this large ditch since it did not, as on the northern front, have to support the weight ramparte oth f facn I t .som a t e subsequen uses rubbisa t tims wa da t ei h dump confusea , d mas f clao s y being tipped int s inneoit layea r half d burneo r an f d clay, charcoa metad an l l working slag spread quite thickly over virtuall whole yth e ditch-hollowwidte th f ho . innee Th r ditches having usef falleformee o t ,th nou r outermost ditc becamw hno innee eth r lin f defenceeo , but, there being apparently perio o littln r f abandonmeno edo t betweee nth secon third dan d phasefort'e th f sso history alterationo n , s were require eithen di courss it r e or profile. Nevertheless, it was now necessary to dig another ditch which lay just over 40 ft W of the inner ditch and followed a course parallel to it. In section C the ditch's profile was slacker MAXWELL: EXCAVATIONS AT THE ROMAN FORT OF CRAWFORD, LANARKSHIRE | 159 e northerth tha n no n fronslightld an t y shallower, being just ovet dee f almosd 4 t rpf an 4 1 t broad. Both ditches were allowed to fill with silt over a long period at the end of their useful lives, exactl . happenes ya N e th n do s obvioui t I s thae evidencth t e gained from section througg du s e westerhth n defences agrees very well with wha have w t e already learned abou northere th t n front. There were three main periods of occupation, the first belonging to the late first century AD, the other two to the middl secone th f eo d century; little time appear havo st e elapsed betwee occupationslaso e nth ttw . The major difference lies in the fact that the rampart of the third period was not advanced on . N e th n o s wa t i s a theW The Southern Defences) 7 d (figan 4 s reasone th r sFo give t completa possibl t no n cu s abov o wa et e t ei sectio n througe hth defences on this side. Nevertheless, the picture that emerged was substantially the same as that already drawn in the preceding pages. First Period. Once agai e superimpositionth f lateno r structures mad t difficulei givo t t e precise limits to the defences of the earliest occupation, although their approximate position coult leasa e tdb suggested intervallue Th . m stree representes wa t layea y f hardb o r d yellow gravel measuring 16 ft in width and 9 in in maximum thickness. Its northern limit was marked b deen ysteep-sidea i 0 p1 (set f ; thifoundatio 1 a ewid n 7) sectio s i sd i g 6 slotefi an t f , n1 D , n trenc timbea r hfo r buildin firse th t f gperiodo trence beeTh .d hha n half filled with loose gravel, probably the original packing material which held secure the individual uprights of the timber framework, and then sealed with a plug of hard yellow clay. However, although the S side of the trench presented a smooth vertical face the opposite side was rather irregular, and it is not unreasonable to assume that this irregularity was caused by demolition parties digging out the timbers at the end of the occupation (see infra, p 169). Immediately to the N of the trench and just abov Romae eth n humus masa leve y f partlso la l y burned clay mingled with fragmentf o s charcoa burned an l d daub dismantlee Th . d buildin intervallue th d gan m street were coverey db a thick laye f looso r e brown earth characten i , r very similae materiath o t r l constitutine gth bulk of the first-period rampart on the northern front. It is, indeed, very likely that this layer, too, derived in large part from the earliest rampart, as it continued for a further 7 ft S of the intervallum street, at which point it overlay a band of greyish clay. Beyond this the material became more peat mord yan e obviously layered, while claagais ywa n presen e bode th th f yn o i t rampart and on top of the old ground-surface. It seems extremely probable that this body of material represent first-perioe sth d rampart attenuate decay db r deliberatyo e demolition before ensuinre-use th n ei g period s widtIt . h will hav es constructio it bee d nan about f 0 n2 t similar thao t t noted elsewher sitee th , n witeo h clay revetments enclosin earthen ga n core. The inner ditch, too, told the same story as before, for it had been packed solid with a light grey clay, doubtless deriving fro outee mth r revetmen ramparte th f o re-cutd t s it an , n I . first form it measured approximately 10 ft in width by 3 ft 6 in in depth. In a section dug through earle th , ydefencee E inneth e th r so ditct sombeet d f hha 0 ne9 packed with dirty grey gravel, in which a piece of samian bowl (form 29) of first-century date was found (see infra, p 189). Of the middle and outermost ditches there was no trace because the former had been obliterated diggine blarge yth th e f secong e o ditc th f ho d lattere periodth d , an whic, h would have lain f thio fa sd r sectionbeyonen S probabld e ha ,dth y been destroye constructioe th y db e th f no modern road and the medieval castle. Second Period. e openinTh f thigo sdecayee repaie perioth th f w o rr demolishe do sa d early rampar rampard newe ol bode th usee s tha r th y ,bul e th sucys n fo wa dti tth f ta wa k o h a | PROCEEDING 0 16 SOCIETYE TH F SO , 1971-2 the width of the whole now being 20 ft. A new outer face of clay was inserted on a foundation measurin t wid f madd g5 e an roundef eo d boulders packed together with gravel well-cobble.A d intervallum street, some 16 ft wide, was laid on top of the tail of earthen rampart-material abou furthet f 0 1 t tha S rprevioue nth streee sidth N s f thie a tstreeto t distinc th n bu n O . t layer of gravel with occasional cobbles rammed int s surfacoit e extende r approximateldfo t f 0 1 y stopping short at an area of disturbance 3 ft 6 in wide. On the far side of this, and at the same sharp-edgea level s wa , d beginninbouldee th d layea an r f f clagsmald o o r yan l stonee th s- barrace th f o ke floobuildingon f o r s discussed below (se e175)p t woul.I d seem tha disturbee tth d area marke end-walpositioe S dth buildinge e th th f f no o l . rampare ditcheso th Immediatel tw f innee o y beine th la ,tS on re gth fairlo yt y shallow, 12 ft wide by 3 ft deep, cut through the fill of the earlier ditch. It, too, had been deliberately filled with a confused, but solid, mass of dark peaty turf and clay. The outer ditch, of which only half appear sectioe th n si n drawing s mucwa , h large probabld an r y measure dwidtn i about f h6 1 t depthn i t f 6 , wity b hsaggina g V-profile fillins mosIt s . gwa t curiou bord san e comparison with large th tha ef northere o tditc th n ho n front, for, althoug materiale hth s used were different, the purpose appeared to have been the same. Third Period. Here, as on the northern front, the last period saw a thorough-going alteration defencee oth f s wit rampare hth t pushe dextre forwarTh a. ft spac mucs 5 da 3 es ha thu s gained was used for the extension southwards of the barrack building mentioned above, and in section D the new S wall of the building can be seen founded upon the cobbling of the previous inter- vallum e occupatioth n I . n debris associated wit t werhi e several fragment f black-burnisheo s d ware, while almost corniced-ria hal f o f m beake Caston i r r war founs ewa d buried undee th r tumbled core-material from the W wall of the same building. A terminal date for its occupation in the period AD 165-170 has been assumed (see infra, p 179). The reason for the advancement rampare , wheft th mucincreass e f 5 a n o 3 th lengt e barracy e s hb tth a th n f eonli s ho kwa y , 18ft is clear whe considere non groune th s d available previoue th site f eo Th . s rampar beed ha tn covered by the new intervallum, a boulder and gravel layer extending from the S edge of the old intervalluo havT . eft buila distanc 7 rampare 1 r mth tfo c f eo t immediatel woulS e th d o yt have meant either settin deee outes th g it p f vero e middl r p th facy li n eo e ditchd ha r elset o i , f i , extended further S, breaking its back on the spine of natural gravel left standing between the two inner ditches evente th Romae n th I ,. n engineers chos movrampare o et th e sit f eth eo t even , whollN e yth oven o large furthe buils d th r a e an , dditchit S r , having fille ditce dth h first with massive boulder minimiso st e subsidence thed ,an n wit hthica k laye peaf ro t blocks rampare Th . t consisted of an earthen core, with a well-built inner revetment of turves. Unfortunately it appears that the peat blocks of the infill were compressed by the superincumbent weight more than the engineers expected sectio e andth s a ,n drawing shows rampare th , te middle santh n ki , while the inner revetment slid forward and down. The slope of the ground at this point, which is greater than on the other three sides of the fort, would naturally have made the rampart even more unstable. Over the inner ditch which had been deliberately filled, but not covered by the rampart, intermittenn thera y ela t scatte f graveo r smald an l l stones t seemI . s probable that lefs thitswa as an open area during the third period.

The Eastern Defences (figs 3 and 7, pi 15b) possibls t onlwa cu t ypartiaI a o et l section throug defencee hth thin o s s fiele sideth d s a , in whic e outehth r inaccessibls ditchewa y sla e owin agriculturao gt l operations. Nevertheless the evidence, however curtailed, was largely in agreement with that already obtained on the North

Hard yellow clay Loose brown earth South Band f whitso e clay Period I intervallum street Mainly clay •'"' '"" "'"''lv~ Peaty soil Earthy mixture L Period I rampart; approximate site Bands of white clay Peno dramparI I t Perio dintervalluI I m street

Perio dI intervalluII m street

Period III rampart West

East

Black peaty soil

I ramparPerioII d an d I t Period I and III intervallum street

1 Ft. P 20 30 40 5P 60 .,.,,j. —— ...... —— .,.._ ! . , ——— ————————i ———————— I ———————— i ———————— r ———————————[ ———————————— i ———————————— 1 ———————— ————1 ———————————— | ——————— ———i ——————— r —i —————— i ————— i ———— 1 1 1 1 M 10 20 FIG 7 Southern and Eastern defences: section D above, section E below MAXWELL: EXCAVATIONS AT THE ROMAN FORT OF CRAWFORD, LANARKSHIRE | 161 other sides of the fort. Section E shows clearly that the sequence of structures here was the same westere th n o n s fronta . First Period. The bulk of the rampart consisted of dark peaty earth lying immediately on e Romath f o n p humuto sreae level th Wr e t (i.eth )A . thibeen d .o sha n slightly overlaiy nb lowese th tintervallue leveth f o l m street beed , front e awawhilha nth cu t permii o t tyt ea e th t insertion of a clay cheek and stone foundation. It seems quite probable that this represented at least the rear two-thirds of the Flavian rampart, whose original width, if we assume that its outer face consisted of a turf or clay cheek of the same approximate size as the later revetments, would have been about 18 ft. There were no visible traces of the inner ditches of the first period both having been remove subsequene th y db t re-cutting observe othee th n rfortd e o side th .f so The only point of difference in this section is that no tell-tale notch has been left in the face of the innermost ditch. It is likely, nevertheless, that the size of both was about the same, 10 ft wide by 3 ft deep. The earliest intervallum was represented by a heavy band of gravel and cobbles which measured 1 ft 1 in in maximum thickness and overlay Roman humus for a distance of about 12 ft from the rear edge of the rampart. To the W of this point not only the street but also the Roman humus level had been cut away, probably in the course of demolition of the first- period fortconsequentld an , innee yth rintervallue edgth f longeeo o n s mrwa visible originas it ; l widt probabls hwa subsoie yth betweebeed Farthe. bact W ha lft n e cu 0 kth 2 n slightlo d rt 1 8an y into the slope to provide a level floor for the later stone building, with the result that the first- century occupation levebeed ha ln removed. That ther beed earlien eha na r structur thin ei s position was suggested by the presence of a shallow slot, about 1 ft in width and 8 in deep, representin originalle gth y much deeper post-hole trenc timbea f ho r building. Further investiga- tion f sectioundertaket beeo no nN , wher d nE e necessar ha th t ei no t somt f preparo yt 0 e3 e a levelled area for the secondary stone structure, revealed the same trench in a better state of preservation, cut to a depth of over one foot into the old ground-surface, with a 4£ in thick clay and gravel floor extending on top of subsoil from its inner or W edge. It was also noted that the buildin whico gt t belongehi d stoppe clearls wa dd y shorcontemporaran , of t y with intere th , - vallum of the earliest period on the northern front. Second Period. The ensuing period saw the erection of a building, whose timber framework rested on footings of stone, immediately beside the intervallum street. It measured 11 ft 6 in wide, inside walls 2 ft 8 in thick; the natural gravel subsoil had been cut away to a depth of over receivo t n i footinge mor9 ee th th f eo swesterl clay r gravewallsigyan o o n therf s no s A . l ewa floor having been laid insid buildin e naturae eth th d glan gravel surface itself seemed very loose, it was assumed that a raised wooden floor had been constructed, and as excavation progressed possibls iwa t identifo et arrangemente yth s Romae madth y eb n builder r supportinsfo g sucha floor (see infra, p 176). A new intervallum street, 17 ft in width, was constructed at a much higher level than before, the previous street surface having been covered by a confused mass of dark peaty eart clay d doubo han n . e Therb t than eca t this material came fro earle mth y resulrampara s a somf to d etan deliberate slightin thaf go t structure whethet bu , demolitioy rb n firsgange th t f perioso r buildindo g partiesecone th f so d determinede cannob w no t . oute givew rampare w ne no rTh n a materiad s fac ol f clay wa eto e th , l bein t awagcu r yfo the insertion, as the section drawing shows. The matter is somewhat complicated by the fact revetmenw ne tha e s itselth t wa tf almost entirel e thirt awath ycu d n y i perio a secon r dfo d re-facing of the rampart, traces of clay being preserved only beneath and to the rear of the final kerbing (pi 15b). However thie th ,n ban yellof do w gravel, cobble flad stan stones which serveouten a s da rsecond-perioe kerth o bt d revetment survived unde largee th r r boulderf so the later revetment foundation secone widte th Th f . ho d rampar thus wa st see havo nt e been | PROCEEDING 2 16 THF SEO SOCIETY, 1971-2 ditcw ne dugh w systeA no 1. , 8 completels ft mwa y removin l earlietracee gal th f so r defences: the inner ditch was V-shaped and measured c 11 ft in width and 3 ft 9 in in depth, while the large outer ditch probably attained a width of 16 ft and a depth of 5 ft. Third Period. The final phase in the fort's history affected internal structures intervallum street, rampart, and ditches. The stone building seems to have undergone a fairly drastic over- haul f whico , re-buildine hth wall-footinge th f gmose o th s t importanwa s t feature thinnea , r , bein in againss a 8 g t n i f constructe 3 2 twall t f 2 , dfoundatiod upool e nth n courses (pi 16a); theralsy oema have been some internal alterations smalA . l fragmen Castor-wara f o t e vessel and a few sherds of black-burnished ware were found under tumbled core-material from the E wall of the building. The only alteration to the intervallum street was the addition of an extra layer of large cobble holloa grid filp o st an u l w nea reae rth r edge, wher second-perioe eth d road surfacd eha subsided. Both inner ditches had been deliberately filled up, the larger with blocks of clay and a mass of mixed clay and gravel, the smaller mainly with dark peaty turf over which had been cast a confused band of clay and lighter coloured turf. The innermost ditch had then been sealed wit hlevellina g laye yellof ro w gravel which extende ditce d underla d th ove f hinnee an o rth p rli y massivw ne e eth stone ker foundatiod ban finae th lf no re-facin eastere th o gt n ramparte Th . new foundation was more than 6 ft wide and the total width of the new rampart about 19 ft. Just enough rampart materia survived ha l d abov e stonth e e foundatio indicato t n e thae th t revetmen thin i t s phas constructeds ewa lowes t leasit a , n i tr courses blockf o , f peatso y turf, as it was on the northern front. The important points which emerge from this study of the fort's defences are as follows: firstly, the evidence on all four fronts supports the view that the fort experienced three major period f occupationso firse th , t being databl late th e o firset t centur othero e yth tw Ad o st D an barely separable periods extending from abou 170o tt A0 ;D secondly14 assemblee ,th d evidence shows that the overall area of the fort in the first and second periods remained fairly constant at approximately 2 acres, whereas in the third period, by increasing the breadth of the fort over areincreasee s th a wa , ramparte ft th 0 abouo dt 30 o tst 2-66 acres; thirdly t provei , d possible identifo t t leasya t three distinct structural phase internae th n si l building relatford e th an t f eso each to the appropriate period of the defences. The Entrances and Streets (figs 8, 9, 11 and 13) Although it had been possible to determine the position of all the gates of the second- third-period an Flaviae dth f fortso t leasnwela s o a fort,s tw a tdetailelo n , d examinatios nwa thesef o e . on Nevertheless y madan f eo , several point worthe sar commentf yo . firse Th t entranc earliese th f eo t perio recognisee b o dt thas dwa t situated nea middle rth e of the S side; its approximate location was indicated by the discovery of the junction point wher outee eth r Flavian ditch curve entrancemeee o t sidinne th e n dW f i th te o S re ditc.th n ho At an early stage in the excavation it was thought that this marked the position of one of the portae principals. Subsequent investigation, however, proved this hypothesi wronge b o r st fo , it was discovered that a timber building had been erected in the earliest phase in the middle of e fortth , straddlin e lingth e principalisa whicvi e hth would have followe f alignei d thin do s gate. Now as this building (No. I on fig 9) faces directly towards the S entrance and has been identified as the headquarters building (principia) of the Flavian fort (infra, p 165), it follows that entrance th e itsel fportae musth principalise a b tpraetoria, vi e th d woulan alonn dru frone gth t centrae oth f l timber buildin , i.egW . froo parallet lone me E th th gforte o s t th axil A .f so MAXWELL: EXCAVATIONS AT THE ROMAN FORT OF CRAWFORD, LANARKSHIRE | 163 discovery came late in the series of excavations it was impossible to test this second hypothesis adequately with the spade, but earlier trenches had shown that there was an extensive metalled surface on the S side oftheprincipia. Furthermore, if the alignment of the presumed via principalis seee ramparb W nn e tha ica roase th t producei th ts da r woulfa s da d have passed through the defences immediately north of the timber structure mentioned above (p 154). The discussion e significancoth f f thio e s structur s beeeha n deferred until now proximite vien th .I f wo f yo the tower to the S angle and the extreme shortness of this side, absolutely as well as relative to the lengtforte th t seem i ,f ho s much more reasonabl interpreo et structure th t gata s eea tower than as an interval tower. Moreover, a comparison of the inner-ditch sections on the W front

Ft. ,,,,, ,P 20 M 10 FIG 8 Flavian West gate, suggested plan

showed that immediately opposit towee eth re-cuttine traco rth n ther f s eo ewa g carrien i t dou secone th d this periodswa featurr no , e presen anothen i t r section,f o whic N barely t hf la 0 y1 the tower. The evidence has suggested that there had been a gap in the innermost early ditch at the precise point where the via principalis would have crossed the line of the defences. In further notesuppore b y d thif tha ma o tsecond-centur e t t botsi a th t f ho y entrances which could be exactly located the Flavian inner ditch appeared to be unbroken, while trenches dug across the later via decumana failed to recover any trace of an earlier street under it or to either side of it. Other trenches tended to confirm the suspicion, aroused by this discovery, that the dis- position of streets, and therefore gates, in the Flavian fort was markedly different from that obtainin secone th n gi d century. Although recognisverw Flaviano ye y littlth s f whaema a e o n e portaw t principalis dextra was uncovered durin excavationse gth possibls i t ,i offeeo t partiaa r l reconstructio originae th f no l plan (fig 8). The remains comprise the pits and post-holes marking two opposed corners of the S gate-tower. The NW pit was sub-rectangular in shape, measuring 3 ft 4 in by 3 ft, and 2 ft 8 depthin ni t containei ; well-preservee dth d impression squaro tw f eso posts e outermosth , t measuring almost 12 in across, the other, which lay only 4 in away, about 9 in. The SE pit, which appearef leso se b regula littla o dt d en r i an ove t shapef n i 3 r7 t ,f measure2 y b t f d3 depthoriginalld ba t i ; y hel dsingla e square pos f abousection n o ti cleas 2 wa 1 tr t I tha. e th t larger posts represente maie dth n member towera f so , abou squarn i t 6 1t 1f e over all, whose 164 | PROCEEDINGS OF THE SOCIETY, 1971-2 lower half had been occupied by the rampart terminal on the S side of the gate; on the analogy gatE Hill,id t Inchtuthile e a ogatsmalleHo e S th f t th 5e ea th rd posintendes 1an 4 wa t o dt suppor heav e frame th th tr ye fo doors . Consideratio relativfore d sizth e an t f th e o f eno narrow- ness of the via principalis would lead us to expect a single portal of the same general width as the tower, wit hsecona d tower crownin rampare gth entrancee tth sid N terminaf o e th .n o l spreaA f burndo t dau charcoad ban l whic founs hwa d immediately above grounde leveth n o l N side of the setting of two post-holes suggests that the upper parts of the towers and possibly the wall-walk abov gate eth e were protecte wattle-and-dauy db b screens . exaco Thern e tear parallels to the reconstructed plan, but useful comparisons may be made with the S gate at Lyne, gatS t Slacke a 1 th 6 probabld 1 7an gat N t Ilkley. e a rampare widt e yth th th f s ho 1 8A t was greater than the depth of the gateway by about 7 ft, it would have been necessary to decrease the former before it impinged upon the entrance passage; it seems likely that this was done quite sharply as at the SE gate of The Lunt, Baginton,19 where the gate was also set back some 5 ft from the outer rampart-face. It remains only to draw attention to a fairly shallow square pit, situated on the berm before the S gate-tower; it measured about 1 ft 9 in in width by somewhat less in depth, and part of its W side had been cut away by the digging of the innermost Antonine ditch. Unlik othee eth r pit t containesi post-impressiono dn beed ha n t looselbu , y filled with large stone graveld san positios It . n clearly preclude interpretatioe sth n tha t formei t d parf o t the gate structure while its relationship to the Antonine ditch shows that it must pre-date that period of occupation. The most likely explanation must be that it is a derrick-hole comparable to those found at Oakwood20 and Fendoch.21 closo Thern e ear parallel typBritaie N f th layou n eo si r nfo t represente Flaviay db n Crawford. Neither the practice fort at Cawthorn, Yorkshire,22 nor the Agricolan site at Amble- sid Westmorland,n ei 23 bot f whicho h appea laco t r knormaa l retentura, affords sufficiently clear evidence of internal layout. The closest analogies are in fact provided by Hod Hill, Dorset,24 and a number of early forts in Holland which are typified by Valkenburg I.25 In both these examples the via principalis is aligned on the long axis of the fort and there is no retentura or porta decumana. Such an arrangement appears to represent a tradition which is Claudian in origin, although Crawford I and Cawthorn D show that it survived into the late first or early second century e dispositioTh . f buildingno s insid Flaviae eth n for t Crawfora t discusses i d d below (p 165). secone th n I d centur roae buildind yth dan g alignments were re-orientated througc ar n ha of 90° to the E and the internal layout became orthodox. The porta praetoria on the E is now largely covered by outbuildings of the modern farm; its position is nevertheless made certain by the discovery of the course of the via praetoria, which had a width of 20 ft at the point where it was examined. The site of the porta decumana is fixed more or less certainly by the discovery terminae o th flarge th ef o lsecond-perio d metalle e inneth sidd N r s ean ditcd it causewan ho y leading out from it over the filled innermost ditch. It was also possible to verify this with a section acros vie asth decumana which prove juse b widthtn o i dundet t positioe f porta2 e Th .1 rth f no principalis dextra s beeha n deduced fro e observemth principalisa dvi course th f eo whics hi abou t widef porta9 e 1 tth ; principalis sinistra locates trenca wa 196n i y t db 5 h cu whic t hno only revealed the terminals of the large ditch beside the Antonine I causeway, showing that its width was just over 28 ft, but also revealed two of the pits in which the outermost timbers of the Antonine II N gate had stood. These were 10 ft apart, one on either side of the entrance passage, and were of irregular shape, measuring 3 ft 6 in in depth by 2 ft 6 in in greatest width. The irregular shape of the pits was the result of the timbers having been dug out at the end of the last period of occupation. It was easy to see how each pit had been dug into on one side, MAXWELL: EXCAVATION ROMAE TH NT SA FOR CRAWFORDF TO , LANARKSHIR5 16 | E while the remaining sides were undamaged. The pits had been partly filled with the large boulders thaoncd ha t e probably serve packing-stones da werd san e then levelle witf dof h mixtura f eo dirty gravel, small block ramparf so t clarubbishd yan . This last include dfragmena cruciblea f o t , which had been used in the working of brass (see infra, p 198) and two fragments of a cooking- black-burnishen i t po d ware .traco Thern f burnins eo en wa o pite r gth so fil e f eitheo lth n i r surface roadwaye th th f o e shorn I .evidenc e th t e suggested tha gate beed th te ha n deliberately dismantled evacuatioe priolase th th to for e t r time th fo t f no . Sign f similarlo s y painstaking demolition work, presumably carried out by Roman troops, were subsequently discovered in the headquarters building. The implications of this evidence are discussed in the concluding page thif so s paper (see infra, 180)p .

The Buildings inside First-centurye th ) Fort10 d (figan 9 s Period I. The nature of the investigation of the fort at Crawford did not permit the recovery of a complete plan of the interior, even in the latest period of occupation whose remains were tolerably well preserved buildinge Th .Flaviae th f so n period were particularly difficul traco t t e because thebeed yha n interrupte r destroyedo latee th ry db Antonin t e no structure d di d an s t firsa t appea adherstandaro y t r an o et d plan facn i centra;o t tw onl e ly th building I I d an I s( ) coul9 g examinee db fi detaily n o an n di . Nevertheless d beeha n t i ,considere d worth while to attempt a partial reconstruction of the plan of the earliest fort using the tenuous pieces of evidence available onlf i , y because that plan unusualappearo s e b o st . Buildings I and II. We have already seen from our examination of the entrances that the fort faced S, the via principalis running parallel to the long axis of the fort and the porta praetoria being situated in the middle of the S side. Building I, which stood at the junction of the via principalis praetoriaa vi wit e occupied hth an dcentraa l positio forte th , n i mus recognisee b t d as the headquarters building. It was extremely small, measuring only 30 ft in width by 32 ft in depth with a portico or verandah 8 ft deep on its S side. The building was constructed of timber, and had, apparently, been demolished at the end of its life; it was nevertheless possible to recover its plan by tracing the construction trenches in which the main uprights of the timber frame d stoodha .evidenco n Ther s ewa e that thes beed ha en slotted int oa horizonta l sleepers a , occasionally occur Roman si n timber structures; impression packine strencheth e lefth n i tf go s suggested that the uprights were almost square in section and about 6 in in thickness. The building comprised a room, or possibly a suite of rooms, 12 ft deep, which ran along the full width of the building at the rear; in front of this there was an open courtyard, 12 ft square and heavily gravelled, with open porticoes or ambulatories 8 ft deep on three sides. The plan is un- doubtedly vera tha f yo t small principia wit customare hth y suite f roomth reae eo f th o rt a s buildin porticoea d gan d courtyar frontn di e basilicath ; r cross-halo l appear absente b o t s , unless it is represented by the rear portico of the courtyard. The building was obviously far too small for the purposes of routine administration of a full cohort, and is in fact the smallest principia of orthodox plan so far discovered in Roman Britain. Building II lay on the same align- ment as Building I about 15 ft to the W; its remains consisted of five parallel construction trenches measuring 30 ft in length, symmetrically arranged so that the intervals between the outeo tw r trenches were both jus tinne o ove whilt f tw rr4 e intervaleth s measured almost exactly 8 ft, the total depth of the building being 30 ft. On the S side there was a verandah 10 ft wide on a level with that of the principia. The style of construction, implying the need for raised, well-ventilated flooring is that frequently employed in timber granaries, and it would be reason- abl identifo et buildine yth granara s ga jusr yfo t this reason positios ;it n besid headquartere eth s building would also support such an interpretation. However, it must be pointed out that the 166 | PROCEEDINGS OF THE SOCIETY, 1971-2

i

u S. 5 I

wide intervals between the three inner trenches are greater than is normally found in timber granarie f thio s s typ f constructioneo t Fendocha ; r examplefo , e largesth , tt f interva 5 s wa l average th d ean abou internaft.e 3 t Th comparabls i lt f are q abouf s a o 0 90 te withe thath f o t granar t Rougya 26 h Castl 1,05c e( ft)q closeld 0s 2an 7 y approache second-centure s th tha f o t y granar Crawforn yi d itsel 1,00c ( f ft)0q s . MAXWELL: EXCAVATIONS AT THE ROMAN FORT OF CRAWFORD, LANARKSHIRE 167 Althoug impossibls i t hi certaie b o et n abou superstructure th t f eitheeo r buildings i t i , clear that their stout framework consiste timbef do r uprights abou squaren i t6 , standing, without sole plates construction i , n trenches which average littldepta n i d n ei dh an ove 0 abou1 t rf 1 t one foot in width. The abundant traces of burned clay and charcoal found in the area of the prindpia suggested tha walls it t s were compose panelf do wattlf so daubd ean , wherea come sth - plete absence of such a deposit on the site of the granary led to the conclusion that this building was covered with weather-boardin lesa g- s expendable more materiaon e d e likelb an l o yt salvaged if this was at all possible (see below).

,

ff -!

20 30 ' 5 ' ' 10

FIG 10 The Flavian headquarters building and granary

Whatever uncertainty there may be about the use or construction of these buildings, there can be no doubt about the manner of their destruction. There was ample evidence to show that thed beeyha n carefull d systematicallan y y demolished e uppeTh . r e constructiorillinth f go n trenches had clearly been disturbed when the timber uprights were pulled out bodily, and in several possiblplaces where wa se t verticae si o eth t trence l th fac beed f t awao ehha n cu y yb the spades of the demolition party. It seems probable that the main timbers and boarding of buildin I wergI e preserve e elsewhereus r fo d , whil e uselesth e s wattle-and-daub paneld an s scantlin gprindpiae oth f were torn dow burned spote nan th severan n I . do l place disturbee sth d construction trenche lattee th f rso building were choke e seeb o nt d with block unburnf so t daub scattered durin demolitioe gth n wor finalld kan y sealed wit hthica k laye charcoaf ro baked lan d clay. 168 | PROCEEDINGS OF THE SOCIETY, 1971-2 Very few relics were recovered during the examination of these buildings and fewer still could be used to date the structures. Perhaps the most significant find was the small fragment of a samian cup (form 27) found along with a bent nail in the mouth of a drainage channel which ran parallel to the W wall of the principia and was associated with it. Other finds of late first-century date were layee foun th f gre o rn d i y charcoal-flecked soil spread acros site f sth eo Building II after its demolition (see infra, pp 191-2). Buildings III-VI. Althoug remainine hth g structures insid first-centure eth y fort havt eno been examined in such detail and consequently no complete plans have been recovered, in view unusuae oth f l naturfort'e th f s eo street-pla s beeha n t n i considere d appropriat examino et e their probable relationshi overale th o pt l layout. Buildin f Buildinwidtn o i , abous whic t f gIV W hwa 17 y t 8, f 2 handtla g II , like Building III in the NE angle of the fort, probably extended to the N as far as the intervallum street; if, as is most likely, it extended in the other direction as far as the via principalis the total length would have bee nlittla . Althougeft ove0 10 r h onlinternao ytw l divisions were discoverede th , fact that these define rooda m measurin deptn i n abou y i t acrosh gb f 6 17 t widt1e 2 f t sth f ho the building - an area corresponding closely to that occupied by the average contubernium in an auxiliary for - tsuggest s strongly that Building barracka s i V I . Ther juss te wa enoug h space for five more contubernia of similar size between the above-mentioned room and the N end of sucw buildinge hho difficuls i th divisione t i se t o t tbu , s coul fittee db d int remainine oth t f 0 g3 separates wa d an d gravellea V froy I b f mt o i d W S e streett .f th Buildin 8 n 1 dimens o ;it y la gV - sions were probably the same as those of Building IV. Building III, which lay in the N angle representeforte s oth f wa , cornerpartitioE e N th walde E d sectioa ,th onle an ln y th y b f no betwee fiftsixte d nth an h h contubernia assume end N frow e f .mI th e tha widte th t Buildinf ho g interestins i t i abou, s ft wa 7 noto , g2 t III V ewal, d W thalik an s l eit twoul V thaI f o td have lain exactly the same distance from the principia as did Building V on the other side. Such apparent congruity tends to confirm the assumptions made, for it would indicate that the builders of the fort had adopted a regular and probably symmetrical plan, as would be expected in this context Buildinf .I allowes i I gwidtII e dth h suggested remainine ,th g unexplored ground between principiae th d an t i could have accommodate fourtda h barrack-bloc f identicako l dimensions as well as a small praetorium of the same 30 ft frontage as the principia and horreum. Buildins representewa I gV variouy db s section f constructioo s halW nS f trence th n hi of the praetentura; not enough of it was cleared to demonstrate beyond doubt that only one buildin involveds gwa totae Th . l area covered measured approximatel sucd an h t f 7 2 y y 15b t 0f internal divisions as were discovered seemed to be set at an average interval of 11 ft 6 in, with resule th t that ther spacs exactler wa efo y 13 room thif so s width aree .Th a availabl buildinr efo g in the NE half of the praetentura, on the other hand, did not measure more than 140 ft in length - sufficient spac r onl roomsefo 2 1 y . Accordingl t doey i t appeano s r thae buildingth t e th n i s praetentura followed the pattern of the barracks in the retentura and their purpose cannot be understood withou re-examinatiota lattee problee th f th rn o structureso t my lieke theie n si .Th r size and number. The type of barrack-block represented by Buildings III-V is too small for the accommodation of a normal auxiliary centuria of eighty men, whose quarters might be expected Furthermore. ft exteno 5 t 17 betweeo d t an 0 ,entir e even12 th f ni e lengt eacf ho h were divided into contubernia there would be space for little more than eight rooms, and it seems more probable tha buildinge tth s comprise onl contuberniax ysi wit hlargea r room suitr ,o rooms f end.ee o on t ,a Barrack f thiso s desig quite nar e unsuitabl r centurieefo f infantryso t the exactlbu o ,y d y suit turmae th thirty-twf o o troopers,2s sixtee whicf no h unit foune ala n cavalrf sa ar o n d i fou d yan r a cohors n i quingenaria equitata. n thiO s assumptio e contubernianth fivf o e eacn i h barrack MAXWELL: EXCAVATIONS AT THE ROMAN FOR CRAWFORDF TO , LANARKSHIR | E16 9 would have hel troopersx dsi sixte th , h would have principales, o housetw e dth junior o r NCO's, while the larger space at the end would have been assigned to the decurio in charge. We may therefore suggest that the Flavian garrison at Crawford consisted of four turmae of cavalry, or at least 132 officers and men. The purpose of the buildings in the praetentura now becomes apparent r thesfo , e must reasonabl identifiee yb stablese th totae s a dTh . l area availabl r stablinefo juss gi t unde figura r , 8,00ft eq 0s whic h approaches closel internae yth l capacity of a four-turma stable at Benwell, and exceeds that of the stable at Gelligaer. If six mounts were assigned to each of the thirteen stalls which Building VI was estimated to contain, whole th e block could have provided stablin r seventy-eighgfo t horses, whil twelve-stallee eth d building which coul accommodatee db othee th praetenturae n rdi th hal f o f would have held seventy-two. The resulting total of 150 is sufficiently close to the figure of 138 horses required fule byth l suggested garrison e 12 (onr eacth 8 fo ef troopers ho principales,d an r eacfo ho tw commandant)e decurioth d an extre Th . a stal Buildin n perhapi ly ma I gsV represen stabline th t g accommodation reserve commandane th r dfo possibld an t staffs remainine yhi Th . g stalls could thus have been evenl yeaco t allotte x basihe si turma. th f sn o d Flaviae o Th n garriso Crawforf no d was clearly a vexillation of a larger unit, and as such may be regarded as further evidence of the manpower shortag enaturae whicth s l hwa Agricolae resulth f o t n annexatio Britain.N f no 29 f identifyino y wa Thero gn s wit ei degrey han f certainteo unie ywhicth o originalld t t ha t hi y belonged t seemi t bu ,s reasonabl assumo et e tha cavalre tth y complemen cohorsa f o t quingenaria equitata would have been more easily detached fro parens mit t body tha nsmala l vexillatiof no ala.n a Indeed t wouli , d appea coursre thath preparation f i teo Trajan'r nfo s Parthian campaign a large numbe f composito r e regiments were strippe f theido r mounted element orden i s o t r for mlarga e cavalry vexillatio.30 Whil singlo en e regimen Britain i t n durin Flaviae gth n period can be shown to have lacked its complement of horse, it may be more than a coincidence that the length of the main axis of Crawford I narrowly approaches that of Slack and Ilkley, both of which were garrisone part-mountey db d cohort exhibited san d entrances simila forn i r o mt suggestee th gatd W t Crawfordea . Whatever its source there can be little doubt that a garrison of this nature would have been most appropriat rigoroue th r efo s dutie f patrollinso maintainind gan g communicatione th n si rough country of Upper Clydesdale. The fate of Buildings IV and VI appeared to be similar to that of I and II. There were extensive layers of burnt wattle and daub immediately overlying the floors of these structures, sometime r moreo presence n i depta th ; 6 o st f ho f ben eo t nail thin si s debris suggests such careful demolitio s toona k principia. e placth n ei Althoug datablo hn e relics were recovered fro barracke mth r stableo s spreaa s f occupatiodo n debris immediatelf o d y en outsid S e eth Buildin containeV gI smalo dtw l fragment f samiao s n vessel feweo n d sr (foran tha ) mn29 three coins, an as of Vespasian, a sestertius of Titus under Vespasian (both AD 77/78) and another, als t f bronzeidentifiableo no t bu , fourtA . h coin f Domitiao , identifies a 86) D n ,a n(A s a d which was discovered in the demolition layer above the Flavian intervallum street to the S of Buildin , indicategVI s even more closely whe firse nth t abandones forwa t d (see infra, 184)p .

The Buildings inside Second-centurythe Fort thin I s perio layous fore it dth td facetan assume, dE dmora e orthodox form, exhibiting bot hretenturaa praetentura,d an whicn i buildinge hth s were dispose scamna,r dpe t a tha , is t right-angles to the major axis of the fort. All the buildings in the central range appeared to be mainly stone-built, whereas the others probably consisted of a timber and wattle-panelled superstructur dwarf-walln i t ese f stoneso . M 17 | 0PROCEEDING S OF THE SOCIETY, 1971-2 The Principia (figs 11 and 12). Occupying a central position and fronting on the viaprinci- palis, e headquarterth s buildin t ovef 7 gr6 measure ally s maib it ; t f n 9 walls5 d , whicd hha averagn i bee t f n3 e thicknes rested ha clay-and-cobbln do d san e foundations, were very poorly preserved, being nowhere more thacourse non heightn ei . Long stretche wale beed th ha lf nso completely e buildinremoved th e pla f th no d thesn gi an , e parts could onl recoveree yb y db tracing the foundation trenches. From the first it seemed unlikely that stone-robbing in recent times could have accounte r sucdfo h widespread destructio placew nfe evente ande s th th n n ,i i , wher remaine eth beed sha n covere deepea y db r accumulatio possibls soilf wa no t i ,sho o et w either that no robber-trench had ever existed or that, when the wall was demolished and the core-material disturbed e building flooth e e debri th f th o ,d spiller o t ha s . n Furthermoredo , presence th f beneo t nail debrin i s s fro sacellume mth cross-halld an discovere welth s a ,s a l y immediately to the S of the principia of two pits filled with angular blocks of stone, doubtless derived from the wall facings, would seem to indicate that the demolition of the building occurred at the end of Period III and probably as part of a programme of deliberate dismantling of the fort's installations by the Roman garrison. We have already seen that the N gate was also dis- mantled at this time (see supra, p 164). The internal layout of the principia in both periods conformed with the standard pattern for such buildings. The forward part consisted of a heavily-gravelled courtyard about 30 ft portica y b S osquare d 1t deep2f an serie,N A . flanke e post-holef sth o n do s belongine th o gt S colonnade, whose chock-stones were still in situ, appeared to indicate that in Period III the portico was supported by timber uprights measuring about 6 in in thickness and separated by averagn a ; intermediatin e 6 interva t f 7 f o le posts situated half-way betwee maie nth n timbers may have secured light wooden or wattle screens, whose existence is suggested by the shallow slot connecting three of the post-holes. An earlier setting of posts, probably in use during period representes wa roughlo , II tw y db y circular post-holes measuring about 1acrosn 1i lyind san g same th en o alignmen settine th s a tg just described f theso betwee y e ela On . intermediatn na e a majo d ran membe latee th rf o arrangementr closo to ,contemporareithee o et b o t r y with them; the other had been cut into by the easternmost post-hole of the period III portico and was thus clearly earlier. It would thus appear that in period II the portico was supported by four main uprightaveragn a t a t esse interva noteworths i about f I lo . ft t9 y that althoug porticoee hth s showed two distinct structural phases the courtyard itself did not seem to have been resurfaced, the original 12-in thick layer of gravel having served throughout the life of the building, as did more th e lightly gravelled porticoesfloore th f so . onle Th y other features notice excavatee th n di dcourtyare areth f ao d wer post-holeso etw , lying just principia, e ove t insidwalf th E shalloa f 4 re o ld eth an w channe wid n i littld 5 le an e more than 1 in in depth which ran parallel to the wall on the W side of the courtyard at a distance t frof o . 3 fWhilm it difficuls i certaie t ei b o t t n what purpos serveformes e th e wa y db r features, it seems most likely that the latter represented the eaves-drip from the sloping roof of the basilica or cross-hall, which lay immediately to the W. Running the full width of the building and measuring 15 ft in width, the basilica was the fort's hal f assemblo l y where courts-martia othed an l r formal ceremonies were convened under the supervision of the commandant. A setting of boulders which abuts against the W wall of the basilica near its mid-point appeared to be the remains of a rectangular foundation measuring 5 ft by 4 ft; its purpose is uncertain but it may have been the base on which a statue or altar originally stood.3* Two roughly-dressed blocks of stone were found amongst a layer of debris overlyin latese gth basilicate flooth f o r abouend-wal N t fro f e 6 mtth l (pi 16b) face .Th t that no examples of mason's work have been found in any other part of the principia, or the fort as MAXWELL: EXCAVATION THT SA E ROMAN FOR CRAWFORDF TO , LANARKSHIR | E17 1

c Io

if S

wholea , suggests that thes belono structurea tw o gt f particulaeo r importance, probable yth tribunal, daisr o whicn ,o commandane hth t woul dpositios takhi p eu n when dispensing justice.32 Two floor-levels were noted in the cross-hall during excavation, the upper a thin layer of gravel, lowee th clayf o rsherd o N . potterf so coinr yo s were found between these uppe e levelsth t r bu , surfac coveres intermittenn ewa a y db t sprea f debrido s containin gfragmena mortariua f o t m made in the Hartshill-Mancetter factories and dating to AD 140-180 (see infra, p 193). 2 17 PROCEEDING SOCIETYE TH F SO , 1971-2 The rear part of the principia consisted of the usual suite of five rooms, the central one of sacellume whicth s hr regimentawa o l chapel, measuring jusn i deptt n i 6 unde 1d t t f 1f h an 0 1 r in width. The remaining rooms were of a similar depth, but varied in width; those immediately adjoinin sacellume gth wer t widef e9 roomd , whilen e se th wer partitioe eTh . onlft yn7 walls betwee foundationroome n o nth y sla s simila thoso t r e beneat maie hth n walls t werbu , e only

FIG 12 The Antonine headquarters building

2 ft 3 in thick. As the sacellum not only housed the regimental standards and other objects of veneration but also came to be used as the fort strong-room where the garrison's pay and savings coul storede db interestins i t ,i g thadenario ttw Trajaf io n (AD 112-117 Hadriad )an n (AD 119-138) were found among the debris which overlay the upper floor-level of that room; another denarius of Trajan (AD 112-117) was found just below modern turf-level on the line of the E wall of the principia. floo e roome f eacTh th o r f ho s examine mixtura s d wa gravef eo clayd therd an l an , e appeared to have been no such re-surfacing or repairing as was noticed in the cross-hall. shallosacellum,e A flooe th th f t whico rwn pi i beed g hprobablha n du y durin finae gth l demolition, contained several fragments of window glass and a quantity of burnt daub; more MAXWELL: EXCAVATIONS AT THE ROMAN FORT OF CRAWFORD, LANARKSHIRE ] 173 glass was found lying on the floor of the rooms on either side of the sacellum, in the cross-hall porticS e th on i (se d e an infra, 189)p , whil ethia n scatte burnf o r t daub alss owa noticed over- lyin uppee gth rcross-halle flooth f o r . Assuming tha burne th t t daub derived from clay-plastered wattle panels uncertais i t i , n wha structure t par th principia e f th o t f eo this material represents, although it is more likely that it may have come from some internal partition than from upper parts of the main walls of the building.33 Two arrowheads, which were also found amongst the final debris in the sacellum, probably came from the armamentaria; a third was found in the uppermost level of Building IX (see infra, p 186). One more detail falls to be noticed. During excavation an external buttress was discovered principiae wale th th f o l n o abou corne buildingS e fron i th e t m 6 f th 1o rt 8 f ; examinatiof no foundatione th t thia s s point demonstrated quite clearl originae parth s y f thao wa t t li t build. Althoug otheo hn r buttresses were found, symmetr f playo n would demand that a ther s ewa wallW leas e ,more abouth corner.ton n froeW o n i e t m6 1th t 8 f 3 4 Thus positioned, they would have served to counter the heavy downward and outward thrust of a pent-roof covering the suite f fivo eprincipia.e roomth reae f th o r t sa t therefor I e seems unlikely tha roof-levee th t e th f o l sacellum was higher than that of the flanking rooms, as was probably the case at Fendoch35 and in stone-built headquarters buildings where the rear wall of the sacellum projects beyond the line of the main building-wall. The Granary (fig 11; pII i, 15c) horreume f Th .o S r granar o e littla th y o t eyla t ovef 0 1 r the principia and about 17 ft W of the via principalis towards which it faced. It measured 70 ft t ovef b2 ry2 wall heavils t leas swa a t thic f d y3 t k an buttresse l sideal n sd o , excep thuE e s th t giving wheeled vehicles less restricted accesgranare th t sa y entrance building e walle th f Th . so , which reste massivn do e foundation f clacobbled o s yan , appeare bettee b o dt r preserved than principia,e thosth f eo survivin excavatee th n gi d portion maximua o st n i n i m6 t heighf 1 f o t three courses. There was no evidence to indicate whether the building had been demolished at lase th t f perioo d r alloweden o e th falo dt l int t possibli os decaywa demonstrato r et no , e that undergond ha t i alteratioy ean r repairno . Nevertheless moss i t i , t probable tha histors it t s ywa the same as that of the principia; its foundation trenches had obviously been cut through the laye burnf ro t wattl daud ean b from Flaviae Buildinth f o nI gfortV , while amongs tumblee tth d stones fro s wallmit s fragment f second-centuro s y coarseware vessels were recovered (see infra, p 195). internae Th lgranare areth f , ao almos1,00c ft y( q s 0 t identical wit he granar thath f o t y at Rough Castle) represents the total grain-storage capacity of the fort, for we must assume thaunexcavatee headquartere th t th f o N de aresth praetorium,buildine o at th site f th eo s gi r residenco commandine th f eo g officer. Accordingl takeindicatioe n b a y s nsiza e ma th e t yf i n o of the garrison of Crawford in the first Antonine period, and, as such, should be compared with the total storage-capacity available at forts whose barrack accommodation is known or may be inferred with reasonable accuracy t CadderA . Balmuildyd an Antonine th n o e Wall, both of which could have held a quingenary3 6garrison37, the total capacity amounted to about 1,700 2,15d t respectivelyf an q st 0s qf single th ; e granar r Hill,Ba t ya wher presence eth f quineo - 38 genary regiment s attestei s inscriptions,y db internan a d ^ha l are f abouao e Th t . 2,10ft q s 0 3 milliary forts of Fendoch and Housesteads on the other hand each had granaries enclosing a total area of about 3,300 sq ft.io It would thus appear that the grain-storage capacity of Antonine I Crawford was appropriate to a vexillation not exceeding half of a quingenary cohort, or about 250 men. The Barracks , III-VIII)(fig13 s d e 1barrackan 1 e AntoninTh . th f o s e for botn i t s hit phases were more slightly built than the granary or headquarters' building. They appear to have 174 I PROCEEDINGS OF THE SOCIETY, 1971-2

been mainly timber-framed with walls mad f wattleo daubd t eunlikan bu , barracke eth d san stable first-centure th f so y fort lowee th , rstructur e partth f so e were encase supported dan y db dwarf-walls of rubble masonry. This technique was used by Roman engineers to protect the main members of timber-framed buildings from damp-rot, and there were two basic methods f applicationo uprighte th : s coul tenonee db d into horizontal timber sills, which were bedded upon carefully-levelled dwarf-walls r elslowee o , th e r uprighte partth f so s coul enclosee db d MAXWELL: EXCAVATION THT SA E ROMAN FOR CRAWFORDF TO , LANARKSHIR5 17 | E by a low wall, which thus supported the framework in much the same way as a post-hole trench, yet without exposinrise f destructioth ko o t t gi damp-rot.y nb lattee 41Th r metho sees do nwa t hav inferree remainine b e th n beer ca d fo n d usegan barracksI Buildingn di V d an ,sV although their remains, where sectioned, were seldo gooa mn i d stat f preservatioeo neved nan r more than two courses or about 1 ft in height. Indeed in many places the course of the walls was indicated solel bana y f tumbleyb do d core-materia facind an l g stones measuringo t frot f m3 widthn i t f originae 4 th ; l thicknes walle . th in f ss6 o varie t f 2 do t fro t f m2 firse th t n AntoninI e perio retenturae dth contained four barracks measurin averagn go e about 28 ft by 65 ft over all; they were disposed per scamna, each pair being separated by a heavily- metalled roadway some 10 ft in width. No internal divisions belonging to this period were discovered. In the praetentura only Buildings VII and VIII have been tested by excavation, but, assuming that their dimension same th thos s ee a sar e just described unexploree th , de th are n ao N side of the via praetoria between Buildings VII and IX could accommodate exactly three more barracks - one double and one single block, separated by a 10-ft wide street. A similar praetoriaa vi arrangemen e sidth S f e woulo th n do t brin totale gth numbe f barracko r e th n si praetentura to six and the fort's full complement would thus have been ten. exaco Thern e t ear parallel Buildingo t s s III-VII Antoninn i I eI Crawford , unlese on s compares them with timber structure f similao s r siz t Raeburnfoota e , Dumfriesshire,42 for, while their width closely approaches the average for centurial barracks in an auxiliary fort, their length clearly preclude possibilite sth f theyo m having house dfula l centuria. Indeed com- parison with excavated example f normao s l auxiliary barracks, which o rangt t f e5 fro17 mc 120 ft in length,*" suggests very strongly that the barracks at Crawford accommodated half a centuria apiece44; thus, if we are correct in our estimate of the number of barracks, the full garriso f Crawforno e firsth t n i Antonind e phase could have amounte fivo dt e centuriesr o , about 400 men. There are, however, two obstacles to this line of argument: firstly, as has been noted above (p 173), the capacity of the granary is not appropriate to such a large garrison; secondly, a garrison of five centuries must be a vexillation of a larger unit and if it is derived from a quin- genary cohorcenturiaex si f o t t representi unusuan sa l fraction; however derives i t i f i ,d from a milliary cohort of ten centuriae, the accommodation would seem to have been excessively cramped, for regiments of this strength rarely appear to have been housed in barracks less than length.15n i t 0f conclusioe 45Th n seems unavoidable tha Antonine tth egarrisoI e nth wasn i s ,a Flavian period vexillatioa , cohorsa f no quingenaria equitata, probably half. Four blocks would have suffice accommodato dt turmaeo etw thirtf o thirty-twr yo o troopers eac theid han r mounts, half the cavalry complement of a composite regiment; the remaining six blocks would have held three centuriae of sixty men apiece, half of the infantry complement.46 The total garrison strength would thus have bee men0 n24 ,numbea whicr rfo knowe hth n granary-capacity would have been adequate comparisoy B . e granary-capacitth n f Antonino y e Birrens, whics wa h garrisone unity db s more tha nestimatee fouth s ra timeg dbi garrisos sa t Crawford,na 47 amounted to c 5,000 sq ft. Only two part-mounted quingenary cohorts are known to have been stationed in Scotland during the Antonine period, the cohors II Thracum at Mumrills48 and the cohors IIII Gallorum t Castlehilla 49; therreasoo n s ei believo nt e that either regimen vexillates wa t t thida s period. The only other fort in Britain known to have been garrisoned by a vexillation of a composite quingenary regimen Hadrian'f o Drumburghs d ti en W s e Wallth t ,a , whic Notitiae hth Dignitatum associates with the cohors II Lingonum.50 The alterations which were made to the defences in the second Antonine period have 176 | PROCEEDINGS OF THE SOCIETY, 1971-2 already been described; their effect was to increase the width of the fort by approximately 70 ft, althoug lengte hth h remaine e samedth generan I . unchangeds layoufore e th th lwa t f o t t bu , whereas the earlier Antonine principal buildings remained in use with only minor modifications, the barracks werextendew eno lengta o d t f abou ho purposft.e 2 8 t5 Th 1 almoss ewa t certainly to enable the fort to accommodate a larger garrison, belonging to a regiment in which the strengt individuae th f ho l, the centurieso t woulni f thas I than . neares wa t dnme 60 swa 0 8 r also have been necessary to increase the grain-storage capacity of the fort by building another granary; spac r thiefo s would knowhave beew th enf no o n available S granar th e o th t r o et yo presumee th f o N d positio praetorium.e th f no Apart from thiargumente sth s already applied to the problem of the Antonine I garrison are equally valid in the quest for their successors. onle Th y units convenientle whicb n hca y vexillate thay sucn di t wa theha y occup equivalene yth t f fivo e barrack r stable-blocko - milliare ar s r composityo e quingenary infantry regimentss A . moss i t i t unlikely tha fore th tt would have been increase aren di a solel alloo yt w another cohors quingenaria equitata o occupt t wityi h more spacious quarters t followi , s thae succeedinth t g garrison may well have been half of a cohors milliaria peditata; that the vexillation may have bee ninfantre halth f o f y complemen cohorsa f o t milliaria equitata possibls i t lesebu s likely. s therefori t I e highly significant thae cohorsth t I Tungrorum, e onlth y milliary infantry regiment whose presenc attestes ei Scotlann di d durin Antonine gth eperioe period,on t da s wa statione t Castlecaryda smalo , whicto holo s t l dhwa much more than halunite th f . Although52 little is known about the layout of the barracks even in the second Antonine period it appears end werE V S f BuildingI tha so e e d th toccupiesquaren an i I 6 rooa sII t f spaca ,y 3 dmb 2 o es much larger than that normally allotte a singl o dt e contubernium justifiee b y n i dma tha e w t recognisin centurion'e th s a t gi s quarters remainine th ; g spac thesn ei e blocks would have been occupie fouy db r contubernia measurin gt internally f abou 3 2 y tb 1t 1f , while Buildingd an sV VI could each have accommodated exactl same morx th ysi f eeo dimensions, thus givin gtotaa l complement of ten. The suggested arrangement bears close comparison with the Hadrianic barrack f Housesteads,o s ^ which were designe unidentifien a r dfo d milliary cohor werd an t e later occupie cohorse th y db I Tungrorum. 5 The barracks contained comparatively few datable small finds which could be associated with one or other period of the Antonine occupation (see infra, p 179); a large majority of these were fragment f pottero s y vessel f black-burnisheo s d ware, mostly beaded-rim cooking-pots and flat-rimmed platters or bowls. The latest sherds are of Castor ware and include more than corniced-ria hal f o f m beaker with barbotine decoration so-callee th , d hunt-cup, which cannot have reached the site much before AD 170; this piece was found on the surface of the floor of Building III immediately below the tumbled debris of its W wall and can thus be associated wit vere hth y last phas f occupationeo . There are, however sherdw fe , black-burnishef so d ware from this or any part of the fort that can be assigned to Gillam's category 2.54 Building IX (fig 11, IX). In the NE corner of the fort beside the intervallum street lay Building IX; it measured about 66 ft in length and 16 ft 6 in in width, over walls which were thicknesn i n i abou8 firss t sf a t2 t constructed. Ther evidencs ewa e tha structure th t beed eha n more stoutly built than the barracks; for the walls were originally 6 in thicker on average than those of Buildings III-VIII and in places appeared to rest upon a shallow foundation trench filled with clay and cobbles. A curious feature of the interior of the building was a stone-built channe wid n aboui d ldeep8 n ean i fro o 6 t m,6 whic observes hwa d running E paralle e th o t l wall at a distance of about 3 ft 9 in from it and interrupting a partition wall. While its purpose ascertainee coulb t dno d without further excavation t betraye,i dclosa e similarit channele th o yt s sacellume i nth principiae th f o t Croa y Hill, which were interprete sleeper-trenches,s da t i d 55an MAXWELL: EXCAVATIONS AT THE ROMAN FORT OF CRAWFORD, LANARKSHIRE | 177 is possible tha purpose tsuppor o th hert s o eto e wa raise a t d wooden angle floorth en I . between the partition wall referred to above and the E side-wall of the building there was a rectangular settin f stonego s whos limiW e t continue e alignmendth e stone-builth f o t t channel a dee; p deposi woof o twhic h das h overla settine yth g suggeste havdy thaema t servei t heartha s da . a certaiTher s wa en amoun f evidenco t o indicatt e e thae buildinth t d undergongha e repair or alteration: the W side-wall had at some time been reduced in thickness to about 2 ft lowese th , 3tin three course earliee th f so r broad-wall being retaine foundatioa s da nd (fian g7 possibls pi it i 16a) d ean , tha lina t largf eo e block stonf so eflooe lyinth rn go besid epartitioa n wall may represent an attempt to strengthen it. A more probable indication that the structure was erected in the first Antonine period and continued in use throughout the second, is the fact that althoug lengts same Antonine hit th thath s ea s f hi o t eI barracks debrie th , s coverins git floor-level contained a fragment of a Castor-ware beaker. The amount of iron-working slag also present in the debris brings us to a consideration of the use to which the building was put. Because it was much narrower than Buildings III-VIII and was not increased in length at the beginning of the second Antonine period we may assume that Building IX was not a barrack. The presence of such quantities of slag and the possible existenc hearta f eo h agains wale on tl mak t probablei e tha t servei t workshoa s da whicn pi h metal-working , f iron involvino possibld e an , us e ygth brass (see infra, 197-8)p p carries wa , d on. The presence of deposits of iron slag in the filling of the large Antonine I ditch on the W front and in the disturbed post-hole pits of the Antonine II N gate, as well as the discovery of crucible fragment differenn i s t sitee partth ,f suggesso t tha outpue fort'e th tth f s o t workshops may have been considerable.

CONCLUSIONS Whil e datinth e g evidenc e variouth r efo s phase f occupatioo s t Crawforna d agreen i s general with that obtained from other Roman site Scotlandn si respecto , thertw e whicn ear si h it appears to differ significantly: firstly, there is nothing to prove that the first-century occupation was prolonged much beyond the middle of the ninth decade, and secondly very little later Antonine coars buildingfoune e eb th potter lase o n t di th t s f occupationso y wa . Admittedl botn yi h cases we are dealing with a very small number of stratified finds, but the absence of any trace of structural alteration buildingo t s r defence-worko s firse th t f perioo s d would ten supporo dt t the hypothesis that Crawford was an Agricolan foundation, built in AD 80 or 81 and abandoned, like Easter Happrew56 and possibly Castledykes,57 and Gatehouse of Fleet58 soon after AD 87. Moreover face th , t tha woodee th t n building Agricolaf so n Crawford were carefully dismantled, their main timbers being considered capable of further use, shows that the occupation cannot have continued for the whole of the Flavian period. The matter is, however, put beyond specula- bronzo facf e Domitiao th tw tio ts y a thad nb e n an coina t n6 8 f Vespasiaminteo sD A n di n (see infra, 183-4p p ) coul directle db y associated wit demolitioe hth first-perioe th f no d fort, as they wer l discovereeal debris-layee th n di r immediately overlying Flavian street-levelst No . very much worn when lost coine th , s invite comparison with example similaa f o s r datd ean mint condition which constitute the main terminal dating-evidence for the Agricolan sites at Inchtuthil,59 Newstead d Stracathro.6an 0 t 6I woul1 d seem possible, therefore- ,re thae th t organisation of the northern Limes which followed the withdrawal of the Legio XX from Inch- tuthil c AD 86 may have involved not only the strengthening or refurbishment of such forts as Newstead, Dalswinton, Glenlochar, Oakwoo Miltond dan t alsabandonine bu ,oth f smallego r posts, like Crawford, Gatehouse of Fleet and perhaps Castle Greg,62 nor is there any certain 178 | PROCEEDINGS OF THE SOCIETY, 1971-2 evidence that the larger forts of Castledykes or Easter Happrew were re-occupied or replaced in the late Domitianic period. Presumably at this time the vexillations which had manned the small forts and fortlets were re-unite63 d with their parent regiments and the whole frontier defence force re-organised so that garrisons of maximum strength and mobility could be stationed in those regions where the threat to security was greatest, i.e. the triangular area whose angles are Newstead, Glenlochar and Corbridge. To the N and W of this zone only a very few sites, Loudoun Hill, Castlecar (?d )yan Ardoch believee ar , havo dt e been then occupied, althoug t hmusi e b t remembered tha smale tth l numbe fortf ro thin si s categor reflecy lace yma archaeologicaf th tk o l research as much as a hypothetical Roman evacuation. That Crawford was next occupied in the middle of the second century is proved by the presence of Antonine coarse pottery and samian, and it is more than likely that this re-occupation was associated with the campaigns of Lollius Urbicus in or around AD 140. The fort was now given an orthodox layout, although without increasing its size, and its new garrison was probably a composithal f o f e mounted regiment s interestini t I . o speculatgt Crawfory t no ewh s wa d made big enough to accommodate a full regiment. One explanation may be that the shortage of manpower resulting from heavy troop concentration Antonine aree th f th ao n si e Wall com- pelle Romae dth n army onc evexillatee morus o et d garrisons. Another explanatio n- alread y suggested by S N Miller64 - is that Crawford was garrisoned by half a quingenary regiment, the rest of which was detached for duty in adjacent fortlets in Clydesdale, Annandale and Nithsdale. The system to which these fortlets belong has been shown to originate in the first Antonine period, althoug individuae hth l sites appea havo t r e been modelle milecastlee th n do s mile-fortlete th d an f Hadrian'o s s Wall. size theif 6e5o Th r garrisons will have varied from site siteo t probablt bu , y averaged about forty men suggestioe vien th i ; f wo n thagarrisoe th t t na Crawfor firse th t n d perioi cohorsa s dwa equitata significans i t i nearbe tth thaf o yt leas a to tw t fortlet havy ma es been occupie part-mountey db d units. t woulI d howeve improbable b r e that all of the fortlets known in SW Scotlan66 d were manned by detachments of the Crawford garrison f onli , y because the spreae yar d over suc hwida e area. Moreover assumptioe th n o , n that at any given time half of the cohort remained in garrison at Crawford, the number of men available for detached service would not have been sufficient to man more than five or six fortlets. e territoriumTh r areo jurisdictiof ao n allotte Crawforo dt therefory ma d e have containee dth fortlets of Wandel, Redshaw Burn, Milton, Durisdeer, Carronbridge and Kirkconnel, together with such road-side signalling installation havy ma e s beesa n associated with them.67 impossibls i t I ascertaieo t precise nth e duratio firse th t f nAntonino e occupatio Crawfordf no , but the evidence suggests that it was relatively short and may well have come to an end c AD 154-158 whe Brigantese nth mose th , t powerful confederac tribef yEnglando N n si believee ar , d to have been in a state of unrest. Whatever the actual date, there can be no doubt that the interval betweeAntonino tw e nth e occupation extremels wa s yinnee th brief r r fo Antonin, eI frontE ditche S N e d sWth containean n so d silte o littln r deo material when they were deliberately filled to provide a foundation for the extended defences of the last period. As similar evidence briea f o f interval Antoninbetweeo tw e nth e occupation recoveres wa s t Newstead,da s i t ^i 6 reasonable to expect that most of the other forts in S Scotland were abandoned and re-occupied in the same short space of time, many of them doubtless in the same peaceful and deliberate manne s Crawford.a r e mosTh ^ t probable occasio r sucnfo ha re-allocatio e frontieth f o n r garrison is not an assault by Caledonian tribes, nor even a panic reaction to internal revolt, but a cool re-appraisal of the political situation in Brigantia; actual troop movements need not have taken place until two or three years after peace had been restored. Now, if the unrest which prompted this associatereviee b n wca d with coin type Ac Df so 15 5 showing Britannia subdued MAXWELL: EXCAVATIONS AT THE ROMAN FORT OF CRAWFORD, LANARKSHIRE | 179 and with an inscription referring to reinforcements for the Roman army in Britain under Julius Verus, garrisone 70th Scottise th f so h forts nee t havdno e been withdrawn re-allocatedr o , , until AD 157 or even 158, at which latter date we know Birrens was re-built to accommodate the cohors I Tungrorum^I curtaisectoe a th d f Hadrian'an f o rn o s Wal undes wa l r repair.72 More- over t wouli , reasonable db assumo et e tha changee th t frontief so r policy which were then made would have been reflecte revisee th n i dd orde f battleo r , and, whil clearls i t ei y impossiblo et reconstruct such a list, there are sufficient indications of the new troop dispositions to allow a general pictur revisee th f eo r dexample polic Fo buile . b e up to y t th , f Ardoc o t leas a e d on t han forts on the Antonine Wall appear to have accommodated smaller garrisons,73 while Broomholm, possibly Carzield7* and Raeburnfoot as well as some of the south-western fortlets75 were abandoned; legionary troops were replaced at Newstead by a large cavalry unit, and may well have been withdrawn from other sites which they occupie e firsth t n di Antonin e occupation. On the other hand Crawford was strengthened, a new fort was built at Lyne, possibly to house a cohors milliaria equitata,' strikine th d g16an poweNewsteae th f o r d garrison probablw no , y an ala milliaria, was greatly increased. However, although there is no evidence that building activit Scottisn yi h forts continue reige th f Antoninu no f o d d beyonen e s dth Pius cleas i t i ,r tha Hadrian'n o t Penninee th sn i Wal d s lan generall y reconstructio refurbishinr no f auxiliargo y forts started at least as early as AD 158 and continued until the governorship of Calpurnius Agricola.77 Consideration of the number of troops available in Britain at this period has led some to doubt whether Scotland could have been adequately garrisoned, while so many sites in N England were apparently being prepared for new occupation, and it has been suggested that Scotlan facn i s t abandonedwa t re-occupieAn di D no 15d 5an d unti campaige th l Ulpiuf no s Marcellus in 184.78 Attention has however also been drawn to the scarcity on Roman sites in Scotland of examples of samian ware made after AD 160, although such types abound in the Period Ib leve Hadrian'n o l s Wall. conclusioe Th n would obviouse seeb mevidence o t th f i ; e samiae oth f accepteds ni t followi , s that eithe secone 79th r d Antonine occupatio f Scotlanno d t lonno gd afteen camn r a A o eDt 160 r lato ,e Antonine garrison Scotlann si d received supplies of pottery differing markedly from those reaching forts in N England. The former alternative would more seeb mo et probable would an , d accord well wit evidence hth Crawfordf eo , where lace latf th ko e Antonine black-burnished-ware (see infra, 193-6p p ) appear indicato st relaea - tively short secondary occupation, althoug presence hth sherdf eo Castof so r ware shows thae tth evacuation did not take place much before 170. While similar evidence of a brief occupation in the second Antonine period was forthcoming at Lyne, and possibly Glenlochar, ! and 80 8 Loudoun Hill82 there is reason to believe that some forts, particularly those on the Antonine Wall, were at least held, if not fully garrisoned, for a little longer;83 how much longer is not thae b ty muccertainma f t Scotlani ho t bu evacuate,s dwa 175y db , rather tha t somna e time reige in th f Commodus. no explanatioe 8Th 4 f sucno h disparity betwee terminae nth l datef so Crawford or Lyne and forts on the Antonine Wall may be explained by the nature of the policy which appears to have been adopted in 158; for it seems clear, in view of the evidence of rebuilding of the Pennine forts under Marcus, that the decision had been taken to re-instate the Hadrianic frontier. What occurre Scotlann di d from that time onwar havy dma e bee moro n e thana selective running down of garrisons and installations in the forward area together with the temporary provision of extra strong-points and mobile units in the vicinity of those areas affected by the Brigantian revolt. The role of Crawford in the second Antonine period must be seen in the context of a partial disintegration of the fortlet system. For as there is no evidence either of the construction of new fortlets or of any increase in the size of existing fortlets, it is difficult to believe that the augmented 180 | PROCEEDINGS OF THE SOCIETY, 1971-2 garrison at Crawford, now possibly 400 strong, was intended to do other than increase the security importanon fa t nodal Romapoine th n ti n road-system. Accordingl less i t syi likely tha unie th t t of which part was stationed at Crawford during the later Antonine period also supplied the garrisons for the neighbouring fortlets. Betweefinalls fore th wa t 0 y n17 abandoned 16d 5an , apparentl deliberata s pary a f o t e withdrawal which left ample time for a thorough and tidy demolition of the installations on the same th immediatelt esiten a i tim f d I . ean y succeeding year same sth e kin f withdrawado s wa l taking place at other sites in S Scotland the 'bellum Britannicum' which threatened to break out in the early 170s85 may have been the response of hostile tribes from beyond the Antonine Wall to whom the growing weakness of the frontier defences was an invitation to invade. Before many years had passed the invitation had become too compelling to decline.

NOTES 1 Military Antiquities Romanse oth f Britain,n i 61. Romane Th 2 Occupation of South-Western Scotland, Miller (ed)N S , , 11. 3f 3 The southern slopes of Castle Hill to the NW of the Roman fort are occupied by at least five pre- historic habitation sites: a fort (951218), a settlement (945215), a homestead (952220) and two enclosures (944215, 946216). 4 Wilson ContributionA , A J , Historye th o t of Lanarkshire, ff2 . 27 , II 5 A fort intended to accommodate a cohors quingenaria would normally occupy from 3-0 to 3-5 acres (measured over the ramparts); Brough on Noe, however, with an area of just over 2 acres, was garrisoned by a full quingenary cohort (Derbyshire ArchaeolJ, LXXXVH (1967), 154-8). € For evidence of similar constructional technique in sections of Roman road at Turf Law, Berwick- shire, and on the Cask Ridge, Perthshire, cf DBS 1964, 24; ibid 7967, 37. horizontaf o e us e lTh timbe 7 r strappin ensurgo t stabilite e th earthe n a f yo turr no f rampar attestes ti d varioun o s exampln sitesa r fo ; t Caeea r Montgomery e SwYorkd Th san f c , Collections, , vol59 . pts 1 and 2 (1965-6), 112 ff; ibid, 60, pts 1 and 2 (1967-8), 64; RCHM, The City of York: Eboracum, 12 f, pi 12. 8 PSAS, xxxrx (1904-5), 45. 6ff 9 DBS 1960, 29Stud,m ; /Ro XLI (1951). 60 , 10 Thi t sthroug trenchcu e fort' e b firse hth o th ,t t s defences durin excavatione gth f 1961-66so s wa , t right-anglea lait fortde ou visible th sidthis th a W ;f seS o o t s efeatur e scarth n peo doet sno exactly follow the alignment of the SW defences a slightly oblique section was thus obtained. The measurements relating to the ditches and rampart observed in this section, where they appear in the text, represent the true measurements, not the apparent dimensions visible on the section wall (fig 6). 11 Archaeol Aeliana, v (1909), 213 ff. 2 1 Antiquity, (1966). L ff X 0 30 , 13 cf Current Research in Romano-British Coarse Pottery, (CBA Research Report 10), 67 ff. Stud,m Ro J XLI 4 V1 (1954). 84 , 15 Richmond, I A, Hod Hill, II, 71. 16 PSAS, xxxv (1900-1). f 6 17 , 17 Yorkshire ArchaeolJ, xxvi (1922), 13 f. 18 ibid, xxvm (1926), 159 f. 19 Trans Birmingham Archaeol Stud, m Soc, (1969) 3 /; Ro 8 ff LI 5 X6 , (1969), 21. 7f 0 PSAS,2 LXXXVI (1951-2). f 4 9 , 21 ibid, Lxxm (1938-9), 115 ff. 22 ArchaeolJ, LXXXIX (1932), 70 ff. 3 2 Trans Cumberland Westmorland Antiq Archaeol Soc, xxi (1921). ff 2 , 24 Richmond. f citp 6 o 6 , , A I , 25 Van Giffen, A, Jaaresverlag XXV-XXVIII, XXXIII-XXXVIII Van de Vereiniging voor Terpenonder- zoek, RomeinseDe Castella dorpsheuvelde in Valkenburgte Rijnde (ZH),aan Praetorium Agrippinae. MAXWELL: EXCAVATION THT SA E ROMAN FOR CRAWFORDF TO , LANARKSHIR | E18 1 6 PSAS,2 LXXHI (1938-9). ff 9 12 , 27 ibid, xxxix (1904-5), 474 ff. turmae th f 2I 8 comprise d thirty trooper Cheesmanf Auxilias(c e Th , L Romane oG ,th f Imperial Army, contuberniax si e th 2 ) 6f would have held five troopers each alse ;se o Epigraphischen Daviesi , W R , Studien, , HOf4 . 29 There is evidence that the fort at Birrens in the Flavian period may have been too small to have accommodated more than a vexillation of a cohort (DES 1967, 21). 30 Cheesman citp o , , 113L G ;, CIL, m, 600 alsf ;Stud,c m o /Ro LIX (1969) ff3 ;5 , Archaeol Aeliana, rx (1932), 207. principiae th n i t Brough-by-Bainbridgs a 3A 1 Stud,em (/Ro LIX . (1969)f) 7 20 , 32 cf Archaeol Aeliana, xm (1936), 221 ff. alternate 3Th 3 e layer 'powderef so d brick' blacd ,an k material containing scrap woonumbef a s o d dan r of small nails which filled the cellar of the sacellum at Old Kilpatrick may be evidence of similar partitions (Miller Romane NS , Th , Kilpatrick,d FortOl t a 26). 34 Examination of the S corner of the principia was inconclusive owing to later disturbance, but it seems unlikely that a buttress was built there. 35 PSAS, Lxxra (1938-9), 126 f. 6 Clarke3 Romane Th , J , Fort t Cadder,a . f 1 4 37 Miller, S N, The Roman Fort at Balmuildy, 26 ff. 38 Macdonald, G and Park, A, The Roman Forts on the Bar Hill, 40 ff. 39 RIB, 2167, 2169-70, 2172. 40 cf Nash-Williams, V E, The Roman Frontier in Wales (2nd edition, M G Jarrett, ed), 160 ff. 41 Richmond Studiesn i , A Building n I i , History, Jope (ed)M ,. 24 , 42 Trans Dumfriesshire Galloway Natur Hist Antiq Soc, xxxix (1960-1), 24 ff. barracke 4Th 3 t Caddesa r measured lengtn 12i t 0f h (Clarke citp ff)8 o 4 , , Kilpatricd J ,thos Ol t ea k Romane t (Millerf Th measure 5 , Kilpatrick,d N 17 FortOl S ,o t t a p dff)u 5 1 . 44 For half-century barracks at Martinhoe, Devon, and Gatehouse of Fleet, Kirkcudbright, see Fox, A, in Studien Militargrenzenn zude Roms (Vortrag . Internationale6 s de e n Limeskongressen i s Suddeutschland) 1961,S . DE 35 d an f 7 1 , f Archaeolc 5 4 Aeliana, XLV. Hff (1969)7 2 , 19 , 46 Cheesman, G L, op cit, 28 ff. 47 PSAS, (1895-6)x xx ; RIB, f 2 11 ,2092-4 , 2097, 2100, 2104 etc. 8 RIB,4 2142. 49 ibid, 2195; for a recent estimate of the size of Castlehill, cf / Rom Stud, XLI (1951), 61 f. 50 cf Birley, E, Research on Hadrian's Wall, 209 ff. barracke 5Th 1 closele sar y matche respecn di f lengto t thosy hb t Gatehousea f Fleeeo t (DES 1961, 35). 52 RIB, 2155, and possibly 2135, but cf ArchaeolJ, cxxv (1968), 96 f. 53 Archaeol Aeliana, xxxvni (1960), 61 ff; ibid, xxxix (1961), 279 ff. The presence of arrowheads in the principia Buildind an t Crawfor a X gI d make possiblt si e tha lase tth t garriso parcohorss a f nto wa sagittariorum; alreads ha t i y been suggested that suc hregimena t forme originae dth l garrisot na Housesteads (ibid, XLVI (1968) ff)4 28 ., 54 cf Current Research in Romano-British Coarse Pottery, (CBA Research Report 10), 67 ff. 55 ibid, LXXI (1936-7), 41 ff. 56 ibid (1956-7)c x , . ff 3 9 , 57 J Rom Stud, LVI (1966), 269 f. 58 DES 1961, 35. 5 Stud,9m /Ro XLIX (1959), 104. 60 PSAS, LXXXW (1949-50), 6 f. 1969,S 6DE ; PSAS,12 (1970-1)3 10 . f 1 13 , 62 The fortlet of Castle Greg (RCAMS, Inventory of Midlothian and West Lothian, 140) occupies approximately the same area as Gatehouse of Fleet and is nearly twice as big as the average Antonine fortlet absence Th .roa a f eo d linkin t witgi Forte Clyde h th th fort r hn eo o ssuggest s that, like Easter Happrew (PSAS, (1956-7)c x havy , 101)ema t beei , n buil evacuated an t d before the construction of the road system in the late-Domitianic period. 63 Castledykes may well have been unoccupied during the later Flavian period, but it seems most 182 | PROCEEDINGS OF THE SOCIETY, 1971-2 unlikely that, as was suggested in / Rom Stud, LVI (1966), 270, a new fort was then built on less favourable ground more NE e thamile th . no on eEvidenct Flaviaa r efo n II occupatio t Lynna e consist fragmena f so first-centurf o t y samian; however recene vien th i , f wo t up-datin ultimatf go e GaulisS h wares from Montan Wacherf s(c , ExcavationsS J , t Brough-on-Humbera 1958-61, 129) muco to testimone hth weighn singlo a t f yo tpu e shoul e sherdb t alse dno se ;o not. e76 Romane Th 4 6 Occupation of South-Western Scotland, 220. 65 The sites at Burnswark, Durisdeer, Milton (South) and possibly Dalmakethar appear to be descended typologically from milefortlet e Cardurnocth f o s k type (Trans Cumberland Westmorland Antiq Archaeol Soc, XLVH (1948), 78 ff); other fortlets, including those on the Antonine Wall, betray a greater affinit milecastleo yt Hadrian'n so s Wall although they see favoumo t squara r e rather than a short-axis or long-axis plan (cf Birley, E, op cit, 98). For evidence of a centuria in garrison at Barburgh Mill Britannia,e se , m (1972). f 4 30 , 66 Find f horse-equipmenso t were mad t botea h Milto Durisdeed nan r (Roman Occupation of South- western Scotland, 106, 126). 67 e.g watch-towee .th Beattoct ra k Summit (DES1966,47) which, althoug Antonint datee hth no o dt e period by any finds, seems from its low-lying position to be better suited to the surveillance of the road syste conjunctiomn i n wit fortlete hth s than long-range signalling. 68 PSAS, LXXXIV (1949-50). 14 , 69 ibid, xcrv (1960-1). f 7 ,9 70 RIB, 1322. 71 ibid, 2110. 72 ibid, 1389. r ArdocFo e Crawford3 hse , 7 Topography S G O , of Roman Kilpatricd Ol r Scotland,fo e ; kff se 5 3 Romane MillerTh , Kilpatrick,d N FortOl S , t a . 58 74 Trans Dumfriesshire Galloway Natur Hist Antiq Soc, xxiv (1947), 70, 77 f. 75 Lyne fortlet appears to have been superseded by the large fort (PSAS, xcv (1961-2), 208 ff); the site at Burnswark was overlain by a practice-camp, while Wandel, if ever completed (DES 1966, 47), does not appear to have been in use for more than a brief period. 76 PSAS, xcv (1961-2), 217. Since this was first written it has been shown, on the evidence of samian ware, tha large th t e for t Lyn a tprobabls ewa y buil A(Britannia,n 0 i tD 14 (1972)m , 44-5. 77 The evidence is conveniently summarised in Frere, S S, Britannia, 156. f Banner7c 8 Jahrbuch, (1970)0 17 , 194. Britannia,w no e 7Se 9 m (1972). ff 1 , 80 PSAS, (1961-2)v xc not e . f7 se ;e 76 21 , 1 Trans8 Dumfriesshire Galloway Natur Hist Antiq Soc, (1951-2)x xx . 16 , 82 ibid, xxxix (1960-1), 48. 83 Macdonald, G, The Roman Wall in Scotland, 479; PSAS, 103 (1970-1), 133. 84 ibid, 245; The denarius of Crispina of c AD 180 which was found at Newstead need not necessarily be associated wit Antonine hth I occupationeI . Recent inspectio samiae th f no n from Newstead HartleR B reveales r byyha M presence dth earlf eo y third-century material morf .O e significance absence isth e fro otherwise mth e complete coin-series from Traprai emperor e coinf o th f w so nLa s Marcus Aurelius, Commodus, Pertinax, Severus and Caracalla (PSAS, xcrv (1960-1), 161-2). 85 Scriptores Historiae Augustae, Marcus, xxn, 1.

ABBREVIATIONS USED IN THE CATALOGUE OF SMALL FINDS MillerN S RomanBalmuildye Th , Fort t Balmuildya Antoninee th n o Wall (1922) Bar Hill G Macdonald and A Park, The Roman Forts on the Bar Hill, Dumbartonshire (1906) Birdoswald I A Richmond and E B Birley, 'Excavations on Hadrian's Wall in the Birdoswald - Pike Hill Sector, 1929', Trans Cumberland Westmorland Antiq Archaeol Soc, xxx (1930), 169 ff CordeP RomansT d Broughan r , Excavations Romane th t a Town Brough Petuaria- 1937 (1938) CGP J A Stanfield and G Simpson, Central Gaulish Potters (1958) Camelon D Christison, et al, 'Account of the Excavation of the Roman Station of Camelon ... in 1900,' PSAS, xxxv (1900-1), 329 ff MAXWELL: EXCAVATIONS AT THE ROMAN FORT OF CRAWFORD, LANARKSHIRE 183 Camulodunum Hull R Hawke C M ,F Camulodunumd C san (1947) Cardurnock F G Simpson and K S Hodgson, The Coastal Mile-Fortlet at Cardurnock', Trans Cumberland Westmorland Antiq Archaeol Soc, XLVH (1948), 78 ff Castledykes A S Robertson, The Roman Fort at Castledykes (1964) Duntocher A S Robertson, An Antonine Fort, Golden Hill, Duntocher (1957) Forden Gaer F N Pryce and T Davies Pryce, 'The Forden Gaer, Second Interim Report', Archaeol Cambrensis, 1929, 100 ff Gillam J P Gillam, Types of Roman Coarse Pottery Vessels in Northern Britain (1968) BrailsfordW J Ho d Hill,d Ho ,Hill Volume1 One: Antiquities e Durden th fromd Hill n i Ho Collection (1962) Hofheim E Ritterling, Das Friihromische Lager bei Hofheim in Taunus (1913) HartleyR B , 'ThIlkleye Roma n For t Ilkleya t : Excavation f 1962'so , Proc Leeds Philosoph Lit Soc, xn, 2 (1966), 23 ff Knorr R Knorr, Tdpfer und Fabriken verzierter Terra Sigillata des ersten Jahrhunderts (1919) Kirk L CordeP J RomanA Langtond , ran Villa Langton,t a near Malton, . YorkshireE (1932) CorderP Defencese Romane MaltonTh ,th f o I I Fort t Maltona (undated) Mumrills I G Macdonald and A O Curie, 'The Roman Fort at Mumrills, near Falkirk', PSAS, LXIII (1928-9), 396 ff Mumrills II K A Steer, 'Excavations at Mumrills Roman Fort', PSAS, xcrv (1960-1), 86 ff CurieJ Newstead RomanA , Frontier Peoples it Postd an (1911) Old Kilpatrick MillerN S Romane Th , Kilpatrick d FortOl t a (1928) KarnitschP Ovilava Reliefsigillatae Di , van Ovilava (1959) Richborough V B W Cunliffe ed, Fifth Report on the Excavations of the Roman Fort at Richborough, Kent (1968) Trentholme L P Wenham, The Rotnano-British Cemetery at Trentholme Drive, York (1968) Bushe-FoxP J Wroxeter , I Excavations Sitee th Romane n oo th f Town t Wroxeter,a Shropshire, in 1912 (1913) Bushe-FoxP J Wroxeter ,I I Second Excavationse Reportth n t Wroxeter,o a . . . Shropshire 1913 (1914) Wroxeter 1923-7 AtkinsonD , ExcavationsReporte th n o t Wroxetera (The Roman City f Viroconiuni)o in the County of Salop 1923-1927 (1942)

THE FINDS The Coins1 Nine Roman coins were found durin excavatione gth f 1961-6o s . They wer l submitteal e o t d Dr Anne Robertson of the Hunterian Museum, University of Glasgow, who kindly reported on them in the following terms. 1 From the surface of the Flavian intervallum street on the N front. Denarius of Nero AD 54-68 Undated, c AD 65-68 AR Siz1-8 mm. 8 . e91 Axig Wt . s \ Obv. [NERO CAESAR] AVGVSTVS (1. up, r. down) Head of Nero, laureate, bearded, r. Rev. IVPPITER [CVSTOS . downr , up ). (1 ] Jupiter, nake waisto dt , seate thronen o . d1 , holding thunderbol sceptrd an t e Min Romf o t e 8 2 74-7C C HC 6BM 9 RI 11 5 C4 C Much worn 2 From the first-century debris layer immediately S of Flavian Building IV. As of Vespasian AD 69-79 COS VIII AD 77-78 AE Size 26 mm. Wt. 6-66 g. Axis j Obv. (IMP CAES] VESPAS [IAN AVG COS VIII P P] (1. to r.) Head of Vespasian, laureate, r., globe at point of neck 184 | PROCEEDINGS OF THE SOCIETY, 1971-2 Rev. IVDAEA CAPTA (1. up, r. down) S C (in exergue) Jewess seate cuirasn o . dr foot s a palm-tree f o t , head whic proppem ar . h1 . restknen r d o n seo In front and behind, shields Min f Gauo t l (Lyon) ? s 3 18 C HC 5 84 C BM 0 RI2 24 C 76 C Corroded, not much worn 3 Fro same m. th 2 e. leveno s a l Sestertius of Titus (under Vespasian) AD 72-79 COS VI AD 77-78 AE Size 35 mm. Wt. 20-90 g. Axis j I CENSO) V r. S o t . CO R (1 P R T N PO F G Obv. AV CAET P SIM Head of Titus, laureate, bearded, r., globe at point of neck Rev. ROM) up . A(1 S C (1. and r. in field) Roma, helmeted militarn i , y dress, standing 1., holding Victor verticad yan l spear Mint of Gaul (Lyons ?) 8 5 C HC 5 85 C BM 2 77 C RI 4 18 C Corroded, not much worn 4 From loose soil immediately to W of Antonine Building IX in Section E. Denarius of Titus AD 79-81 TR P VIIII IMP Xtlll COS VII P P Late AD 79 . 1-8 . 7Wt g AxiA R. s jSizmm 8 e1 Obv. IMP TITVS CAES VESPASIAN AVG P M (r. to 1., outwardly) Head of Titus, laureate, bearded, r. Rev. TR P VIIII IMP XIIII COS [VII P P] (1. to r.) Rostral column surmounte statuy db f maleo e figure, radiate, standing front, holding spear and parazonium Min f Romo t e 3 1 C BM 0 1 C RI 2 27 C Corroded t greatlno , y worn 5 Fro layee m th debrif o r s overlyin Flaviae gth n intervallum street immediately belowalE f e o lw th Antonine Building III. f Domitia6 o 8 s D A A n I AXI DS 81-9CO 6 . 10-6 Wt . A2g Axi E. s Sizmm j 8 e2 Obv. IMP CAES DOMIT AVG GERM COS XII CENS [PER P P] (1. to r.) Head of Domitian, laureate, bearded, r., aegis at point of neck Rev. [FOR] TVNAE AVGVSTI (1. up, r. down) S C (1. and r. in field) Fortuna, draped, standing 1., holding rudde cornucopiad an r e Min f Romo t e C 122 RIC 333 BMC 386-8 HCC 147 Corroded, very little worn ia 6 From the same level as nos 2 and 3. Fragment of a small brass coin, possibly a Flavian As. 7 From the floor of the sacellum of the Antonine principia. 112-11DenariuD A I 7 f TrajaV so 98-11 D S A nCO 7 AR Siz . eWt 19mm . 2-1 . 6g Axi s ^ ) r. o t . (1 ] P P [COP I R V ST M P C DA TRAIANObv.P R IM GE G OAV Bust of Trajan, laureate, r., slight drapery on 1. shoulder (1. up, r. down) QRev.P OPTIMR S O PRINCIPI exerguen i [ARA Q BAD ] Arabia, draped, standing 1., holding branch over camel walkin bundla gd 1. an cane,f eo ) s(? (bu6 2 Cf t C .rev. camel t ostrich) 474-6 no ,C 160-16C 245C BM ,RI HC , 1 Corroded, not much worn 8 From loose soil above the NE corner of the Antonine principia. Denariu Trajaf so n AD 98-117 From portrait probabl (AI V D S 112-117yCO ) AR Sizel7imm l-94g. Wt . . Axis j MAXWELL: EXCAVATIONS AT THE ROMAN FORT OF CRAWFORD, LANARKSHIRE | 185 P P I V S CO Obv. P P C R MProbablT DA TRAIANP R yIM GE G OAV Bus Trajanf o t , laureate , sligh,r. t draper shoulde. 1 n yo r Rev. [S P Q R OPTIMO PRINCIPI] Mars, helmeted, naked except for cloak round waist, advancing r., holding transverse spear and trophy over 1. shoulder C372C f , RIC 415-17 C 270 146-C BM , HC ,7 Much corroded t mucno , h worn From the same level as no. 7. Denariu 119-13D A f Hadrias o 8 I II 117-13D S A n CO 8 2-4. . 3Wt g AxiA R. s jSizmm 0 e2 Obv. Probably HADRIANVS AVGVSTVS Hea Hadrianf do , laureate . (slighr , t draper shoulder?. 1 n yo ) Rev. Probably COS III Concordia (?), draped, seated 1., holding patera Cf C 328 (but obv. with slight drapery on 1 shoulder ?), RIC 172d, BMC 391-2, HCC 125-6 The coin has a heavy deposit on its surface, but does not seem to be much worn

Objects of bronze (fig 14) From the surface of the gravelled berm in front of the Antonine II W rampart. A small fragment probably constituting about one-quarter of a bronze ring of oval section, measuring c 0-2 in thick and 1-1 in diameter; its purpose is unknown. From loose soil over the SW angle of the Antonine granary. An incomplete object comprising a moulded stem c 1-0 in long and 0-3 in maximum diameter with a small shank at one end and a cross-bar 1,c -2 longn 5i , wit projectino htw possibl uncertaiy s gi other arme e ma t th us i t s yt a ; it n bu have served as a guide-loop for a piece of horse harness. From the debris layer overlying Flavian Building VI. Several very badly corroded fragments prob- ably of a small bell. The stump of what may have been its attachment loop measures about 0-3 in across; when entire, the bell, if such it was, would not have measured much more than 0-8 in across the mouth. Such bells are commonly found on Roman sites and may have been originally attache harnese th o d t horses f so necessarilt ,no y cavalry mounts; cfLangton, 18,13g fi ; Castledykes, p. 16114 . ,no

FIG 14 Objects of bronze (above) and lead (below), numbere catalogun i s da e (1:2)

Objects of iron From loose soil above the Antonine II intervallum street on the S front. A very badly corroded object with pyramidal head 2- c ,long n 2i , resemblin gpilum-head,a althoug possibilite hth t i f yo bein gspear-buta r eve o tshank-terminae nth carpenter'a f o l t rulee shoulb bi s t d dno out f ;c Ho; Newstead,17 d HillG , I p pid iLIX an XXXVII 2 1 , ; Mumrills19 , 125g . fi , ,I 26 N | PROCEEDING 6 18 THF SO E SOCIETY, 1971-2 2 From beneath Antonine cobbled layer immediately S of Flavian Building IV. A spearhead in very poor state of preservation, the surviving part being just over 6 in long; the dimensions of the blade were probably about 4- lonn 5i 1- y widegn b 1i Newstead,f c ; pi XXXVII. 7 , 3 From the final demolition-pit in the floor of the Antonine sacellum. An arrowhead of hollow-sided- triangular section, 2-longn 1i , averaging 0-6 acrosn 5i barbse th s ; much corroded, found with fragments of window glass; cf Hod Hill I, 6, B107; Bar Hill, 115 ff; Newstead, pi XXXVIII, 1-7. 4 Fro uppee mth r leve debrif lo s inside Antonine Buildin arrowhean A . gIX mosdn i t respects similar above3 . no , o t 1- lon n 70-i d acrosn 6gi an barbse sth . 5 Fro finae mth l laye f debriro s overlyin Antonine flooe th gth f o r e sacellum. arrowhean A d similar above4 d an , 3 1- lon s n 80-i d acrosn no 6gi an o barbst e sth . Foun remainde th wit lean d hru d an s of three iron nails ranging fro lattee mapparentl s longth -n 41 i 0-f rwa o o 6t leas t ; a e on t y beny b t extraction. 6 From the upper filling of the Antonine I middle ditch on the W front. A bolt, 5-4 in long and 0-6 in maximum diameter; purpose unknown. 7 From the final layer of debris overlying the floors of the rooms on either side of the Antonine sacellum. Three nail-fragments ranging from 1-1 to 1-3 in long; the shanks appear to be of rectan- gula l slightlal r e sectioar y d bentnan . 8 From loose soil over the forecourt of the Antonine prindpia. A fragment of a nail, 2-5 in in overall length shane th , k bein approximatelf go y square section, 0- acrossn 2i markedld an , y bent. 9 Fro debrie mth s layer overlyin Antonine gth front eN I rampar e fragmenA .th n heava o t f o t y bolt 2- longn 4i , wit hspatulata e hea f Castledykes,dc . 12 p, 8 i 10 Fro same m aboveth 9 e. leve fragmenno A .s la fairla f o ty heavy nail, 2- longn 4i shane th , k being slightly bent. 11 From beneat tumblee hth d stone wal W Antoninf f lo s o e Building III nail-fragment.A long-n 41 ,i , the surviving part of the shank being unbent. 12 Fro layee m th f debri o r s overlying Flavian Buildin . ThregVI e fragment f nailso s , ranging from 1-0 to 1-3 in long; the shanks of all were bent and one had traces of charcoal adhering to it. 13 Fro surfac e floome th th Flaviaf o rf eo n Buildin fragmenA . gnaia VI f lo t measuring 1- longn 1i ; the shank, which is slightly bent, has traces of burned wood adhering to it. 14 From beneath the Antonine cobbled layer immediately S of Flavian Building IV. Several nail- fragments bese th ,t preserved havin gbena t shank 2- longn 3i . 15 Fro uppee mth r fillin draia f Flaviae go n th immediatelf o n prindpia.W e th o y t fragmen A f o t nail, 1-5 in long, the shank, which is of rectangular section, being sharply bent.

Objects leadof 1 From beneath the Antonine I intervallum street on the N front. A curved strip of lead of varying octagonal section maximu,n i 0- n 4i m thicknes aboud long n si an strie 6 t Th p. appear havo st e been twisteplaceso tw t whethe n di bu , r thi intentionas si r accidentao l uncertains i l havy ema t I . servehandle th s dr stea lamp-holderea o f mo Newstead,f ;c pi LXXIX. 7 d an 5 , 2 From beneath the Antonine I intervallum street on the W front. A curled strip diminishing in width from 1- 0-o 1, originallt 4in y abou fairlta 2-7f o ylon n 5 i d constangan t thicknes abouf so t 0-. 2in mutilatio e vien th I f wo subsequend nan t corrosio objecte th f purposns o it , difficules i ascertaino t t , represeny buma t i tside r handlth te o leaa f deo lamp-holder Newstead,f c ; pi LXXIX. 9 d an 4 , 3 From beneath layer of debris overlying the Flavian prindpia. Hemispherical playing-man measuring averag0-7n i n 5i e diamete heightfouns n 0-4d i rn wa an 5i dt I . wit e glashth s paste playinn gma described belo 000w p . ( 11) Newstead,f ,no c ; pi XCIII , 21-212 , 2 etc. Althoug conceivabls i t hi e tha piece tth e serve weigha s da t (its present weigh 24-f o t 7 gm being reasonably close thath o et f to Roman undo), associatios it n wit countee hth r make t almosi s t certain that, latterl t leastya t i , was use gamingr dfo . 4 Antonine Froth f mo d loose en granary eW soi e irregularln lth A . immediatelf o S ye th o yt shaped lead disc, probably a weight, measuring 0-75 in in diameter and 0-45 in in thickness; it weighed 24-2 gm; see no. 6 below for comments. 5 From loos Antoninf eo soid len abovN e e Buildineth thiA n. discgIX , reasonabl sidee yon ,flan o t the other side bearing the impressions of the apparently square-headed implement used to hammer it flat; the average diameter is 1-9 in, the thickness about 0-15 in, and the weight 54-4 gm. MAXWELL: EXCAVATIONS AT THE ROMAN FORT OF CRAWFORD, LANARKSHIRE | 187 6 From the N edge of the Antonine via praetoria. A well-made disc, 1-1 in in average diameter, thicknesn i 0- n 3i 47-d san 2 g mweightn i . Although items 4, 5 and 6 have all been subject to a fair amount of damage and wear, it is clear that they have originally served as weights for use with scales. No. 5 approaches very closely the equivalen Romao tw f to n unciae 55-c ( 5 falgm)6 l d equivalent e shor,an th whil 4 f e to s on eno r sfo and two unciae respectively to an extent which, although appreciable, is not without parallel in the field of Roman weights and measures; cf Newstead, 309 f; Hofheim, 66, fig 26. It may also be noted that the weight of the hemispherical piece, no. 3, corresponds fairly closely with a weight from Newstead (cf Newstead, pi LXXXII, 17) as well as with no. 4 above. Certain unstratified fragments of lead found during the excavation at Crawford were submitted r HugD o ht McKerrel Researce th f lo h Laborator Nationae th f yo l Museu Antiquitiemf o Scotland,f so 2 who reporte followss da : fragmenA partialle th f o t y corroded carefullleas dwa y cleaned dow freso nt h smalmetaa d lan l sample was weighed, dissolved in nitric acid and made up to a precise volume with deionised water. The silver content of this solution was determined by atomic absorption spectrophotometry and correspondine th g silver conten calculateleae s th d wa f o t 0-001s da 0 ±0-0003%. Values for the silver content of Roman lead can be as high as 0-05 % and the low value would certainly seem to indicate that desilverisation had taken place. However it is not at all impossible that the silver content of the lead was at this level without desilveration and whilst the result obtained certainly indicates desilverisatio t cannoni regardee b t proos da f positive. nearese Th t sourc leaf e o Crawforo dt Leadhille th s di s field, only three mile. sSW distan e th o t t t seemI s more than likely arebeine s thatth awa f gi , worke Roman di n times, Crawford would have playe t inconsiderablno da e part f Castledykes,c ; . f 7 16 Objects of stone (fig 15, pi 16b) 1 From the make-up level of the floor of Antonine Building III. About one-third of an upper quern stone of local material, originally bun-shaped in section and measuring about 10 or 11 in in diameter and just over 4 in in maximum thickness. It is remarkable in that the grinding-face is slightly convex. 2 cross-hale th f Froo Antonin d e thmn li th en e prindpia.N e leveI eroughlI o th t la Tw y rectangular blocks of sandstone (pi 16b), one measuring about 13-5 by 10 by 4-75 in, the other about 13-5 by 10-5 by 4-0 in. The first has been finely dressed on one of its major faces, and less well on the other, whil edge eon e exhibit pronouncesa d bevel othee Th . r stone, whic roughls hi y dresse botn do h major faces and bevelled on two adjacent edges appears to have served as a corner stone in the structure to which it belonged, probably the tribunal (see supra, p 170). The former stone seems smoothess it n o beaV o lettero t I t tw rface) t whethes(? bu , r this represent tally-marsa k incised before the building of the structure or merely the chance crossing of two lines of stone-dressing is difficult to determine. Both blocks appea composee b o t r f Carboniferoudo s Sandstone, ^roca k t whicfounno s hdi locally possibilite th ; y arises thastructure th t whico et h they belonge importans dwa t enougo ht merit the import of suitable stone from distant quarries.

FIG 15 Quernstone (1:4)

Objects of flint chertor 1 From loose soil over the Antonine intervallum street on the W front. An end-scraper of greyish- white flint with steeply flaked working-edge. 188 | PROCEEDINGS OF THE SOCIETY, 1971-2 2 From loose soil ove Antonine th r e II intervallu frontN m e streeth . much-abradePara n o tf o t d struck flake of veined grey chert with no evidence of use. 3 From the old ground surface beneath the W rampart in section B. Over thirty pieces of dark grey veined chert varyin sizn gi e from irregularly-shaped lumps measuring about 2- 1-y 0-7y 0b 0b n 5i to tiny flakes and scraps no more than 0-3 in long; the majority are flakes measuring about O75 by 0-, non6in e wit definity han e secondary working. The items were found scattered over a fairly restricted area, not more than about 3 ft across, and it seems clear that they represent waste material from tool-manufacture gravew lo site a l e- Th . plateau immediately overlooking the flood-plain of the Clyde - is similar to others on the banks of the Clyde and Tweed where Mesolithic industries have flourished4; cf also Castledykes, 153 f. The end-scraper (no. 1), however, would lie most readily in a Bronze Age context, and it is also possible that the waste material is to be associated with activity in the second millennium BC. Objects f leathero (pi 17) The moist conditions obtaining in the peaty fill of the middle Antonine I ditch on the W front proved extremely favourable for the preservation of organic materials. Among the items recovered from this leve Section i l wern C following e eth : 1 The remains of a left shoe (calceus) comprising: two parts of the upper, each exhibiting three lace- loops t lackinbu , heele gth , which appear havo st e bee t offncu cleanly wit hshara p instrument; the inner sole, measuring 9- lon n 6maximu3-6i n d i gan n 5 i m width ,oute e wit firsth e f hrth o t soles tongue-shapea d an d stiffening heee layeth lt attachea rmeany b t i f narro so o dt w thongso tw ; shaped strips which were originally stitched to the inner side of the upper and were designed to give added strength aroun lace-eyeletse dth individuae Th . l parts were found together, crushed flat by the superincumbent mass of infill; the thread which originally held them together had perished. hob-nailo N s were found attache leathee solese th t th o dr,bu t itsel speckles fwa d wit hdeposia f o t vivianite.

shoe Th e belong well-knowa o st n categor Romaf yo n shoes represente Hillr forte Ba f th so t d a and Newstead,6 as well as on the continent.7 In5 style it was very similar to examples found at Balmuildy8; its size would be appropriate for either a man or a woman. 2 A strip of leather measuring 11-9 in long and varying from 0-9 to 2-3 in wide. It has been stitched impossibls aloni t sidei e t gon ,bu identifo et type ygarmenf th e o equipmenr to whico tt belongedht i . A similarly shaped but larger piece found at Balmuildy was described as a neckband.9 Objects of wood (pi 18). In the waterlogged peaty fill of the middle Antonine I ditch on the W front (section Q among the object f organio s c material which survive almosn di t pristine conditio bloca s f woodkno wa , shaped roughl fore rectangulaa th f m o n y i r prism t measureI . d approximately 4- longn 8i averag2-7n y i b , n 5i e breadt thick-fron d 7i 1 h showan -s mo illustration2 e 1 t A . th n n i beed ha n t i ,pierce rectangulaa y db r hole measuring 1 -7 by 1 -1 in, which contained when first discovered a piece of a rounded stave, 3-3 in long stavjuse d beed th obliquel t tan f e diameterha nunde n o cu i d n i en r1 ye wit;on hshara p instrument, while the other terminated in a ragged fracture. bloce ende th woof f kTh o so originall d dha y been saw squaref nof t werbu , e later subjectea o dt slight degre f weaeo poundin y b r r hammeringgo . Clearl t somya e time bloce th , k must have served as a mallet, and indeed its general appearance, if not its size, invites comparison with the mallet found at

usedo s Newstead. s t musi , wa t ti havf I e beecarpentea y nb r engage lighn di t work wit hchisea r o l

0 gouge. 1 t shoulI d howeve notee rb d tha stave tth emallet-heae holfoune th th f eo n di d canno originae th e tb l wrone th bots f smalo go wa hcross-section to d haft t i lan s a , . Another curious featur shalloe th s ei w longi- tudinal furro r groovwo e whic block'e th h f appearo s middle e majoth on n si f eo r faceso . n Thi s sha obvious connection with the object's use as a mallet, and may indicate that such use was secondary, the wood perhaps deriving originally from the framework of a timber building. Objects of glass and glass paste (fig 16) 1 From loose soil to the W and S end of Antonine Building III. Fragment of a ring-pendant or bangle of translucent light green glass with opaque white looped inlay (Kilbride Jones Type III, F)»; original diameter probably c 2-0 in. Such objects are commonly found on native and Roman sites between Tyn Forthd ean . MAXWELL: EXCAVATION THT SA E ROMAN FOR CRAWFORDF TO , LANARKSHIR9 18 | E 2 Fro Antonin e bode mth th f yo frontramparI eN I e th . Cornen to r fragmen rectangulaa f o t r bottle in thick, light green glass; parraisea f o t d ova circular lo r mouldin basee seeth e b n . no n gca 3 From the floor of Antonine Building V. Small fragment of light green glass, probably from wall of rectangular bottle. 4 From loos portic S Antonine e th th soif oo n i l e principia. smalo Tw l fragment f ligho s t green window-glass; average thicknes below)7 . no sf .0-1(c n 5i 5 From the upper level of the cross-hall in the Antonine principia. Fragment of fairly thick, light green glass, probably from wal rectangulaf o l r bottle. 6 From the same level as no. 5. Fragment of thick, light green glass, probably from shoulder of cylindrical bottle diameter n aboui n i 6 t Newstead,f c ; . 36 g fi 7 Fro debrie mth s layer overlyin sacellume flooe th e Antonin gth f th o r n i e principia. Over forty fragment lighf so t green window glass measuring from 0-1 thickness0-2n o 0i t n 0i 12; each piecs eha mora lesr eo s level matt undersurfac glossa d eyan upper surface, often exhibiting slight undulations. The assemblage includes eight pieces from the edge of a window pane as well as one corner-piece. edgee Th s exhibi well-knowe th t n thumb-shape U-shaper do d cross-section, which would indicate tha glas e cylindes th t swa r blown. corner-piece 13Th e bears small indentations whic represeny hma t the marks of the implement used to uncurl and flatten the blown cylinder in the final stages of the manufacturing process. 8 From beneath Antonin granarye th f eo cobblinS . Fragmene th o gt f thicko t , light green glass, probably from the wall of a cylindrical bottle (cf no. 6 above). 9 Fro debrie mth s layer overlyin Flaviae gth n granary. Fragmen thickf to , light green glass, probably from the base of a rectangular bottle. 10 From loose soil between the S ends of Antonine Buildings III and IV. About one-third of a blue 'melon' bea f vitreoudo s paste, abou axiatn i 0- n 7i l length. Such bead commonle sar y founn do Roman military sites throughout Britain; cf Newstead, 336 f. 11 Fro debrie mth s layer overlyin Flaviane gth principia. bun-shapeA d playing-ma whitf no e vitreous paste; average diameter c 0-65 in, height c 0-25 in. The item was found along with the lead hemi- sphere described above (p 000, no. 3), which may also have been used as a playing-man; cf Newstead, pi XCIII, passim.

FIG 16 Objects of glass and glass paste (1:2)

The pottery Only a relatively small number of datable stratified sherds were found during the excavations, and man thesf yo e were smal badlr o l y wor n- specimen samiaf so n ware being particularly poorly preserved owing to the somewhat acid nature of the soil. The following list consists of all the identifiable stratified sherds with a representative selection of significant pieces from an unstratified context. The presence of an asterisk after the number of any item indicate t beeno nss illustratedthaha t i t . Samian Ware - Decorated (fig 17) Form29 1 From the fill of the innermost Flavian ditch on the S front. Wall fragment showing part of the lower frieze of decoration consisting of joined semicircular festoons enclosing spirals with rosette- terminals; a trilobate pendant and rosettes separate the festoons; the upper border is represented b ywava y line. 190 PROCEEDINGS OF THE SOCIETY, 1971-2 Althoug exacn ha t paralle difficuls li find o motit generalte n i th ,, fis , common amon GaulisgS h potters (cf Knorr, Taf. 37g and Richborough V, p 153 no. 26). Mr B R Hartley suggests a date of Ac D 65-75. piece 1 4Th presumabls ei ysurvivala . From immediately above subsoil to the S of Flavian Building IV. Small wall-fragment showing the central mouldin bead-rod gan w with scrolpara f o t l design below. Date: Flavian. Fro same m aboveth 2 e. leveno .s Smalla l wall-fragment showin centrae gth l mouldin beadd gan - row with parscrola f o t l design. Date: Flavian. From loose soil abov Antonine eth I intervallueI frontS m e streeth . Smaln o t l wall-fragment showing the central moulding and bead-row. Date: Flavian. Form30 From beneath cla yAntonine flooth f o r e granary. Small badly-worn wall-fragment, possibly form 30. Relatively hard fabric, good glaze. Date: ? Flavian. From debris layer immediately overlyin Flaviae gth n granary. Small much worn wall-fragment from just belo witm wri h inner face exhibiting horizontal flutings. Date: Flavian? . 7* Fro easternmose mth t post-hol porticS e th Antoninf of o e o e prindpia. Three conjoining fragments of rim with inner face exhibiting horizontal flutings. Date: ? Flavian. Form37 From the surface of the cobbled area to the W of the Antonine prindpia. Two conjoining fragments of base and wall, the internal diameter of the footring being c 2-0 in. The decoration consists of a vine-scroll with interspersed circles motia , f very popula middl e secone th th n rf i e o d centure th n yi hands of the potters Cinnamus, Paternus and Laxtucissa (cf Newstead, p 225, 4 (Cinnamus); Ovilava, . Date 7) Taf d , :an 73 . AntonineTaf2 , 70 . . From the upper debris level in Antonine Building IX. Small, very worn wall-fragment, possibly showing part of belly and rear legs of galloping horse with the foot of its rider; similar to type used by Sacer Centraa , l Gaulish potter operatin gAc D 125-15 Hartle0R (CGP,B r ff)1 yM . suggest16 s that it may be South Gaulish, although not from La Graufesenque, possibly Montans; found with fragments of Castor ware (infra, p 193, no. 10*) and black-burnished ware (infra, p 191, no. 22). Date Earl? : y Antonine. 10* From beneat Antonine hth I intervallueI frontmN e stree.th Fragmenn o t basf o t e with footring, the internal diameter being c 2-5 in; both the inside and lower edge of the footring show distinct concavity of profile, and the lower edge is tilted up quite sharply (cf CGP, pi 168,15). Date: Antonine. 11* From the floor of Antonine Building IX. Small rim-fragment with more or less rectangular hole for lead rivet used in the repair of the vessel; rim diameter uncertain but probably c 9 in. Date: Antonine. 12* Fro same mth e. 1 1leveno above s a l . Small rim-fragment; diameter uncertai probablt nbu . in 9 yc Date: Antonine.

FI 7 GDecorate1 d samian ware (1:2) Miscellaneous 13* Fro floor-levee mth Antoninf o l e Buildin . WalgIV l fragment possibly fro showin7 m3 r foro 9 mg2 part of a scroll design; reasonably hard fabric and bright glaze, but very much worn. Date: ? Flavian. MAXWELL: EXCAVATIONS AT THE ROMAN FORT OF CRAWFORD, LANARKSHIR1 19 | E 14 Fro same abovem3 th 1 e. leve.no Vers a l y small wall-fragment fro showin7 m3 r foro 9 mg2 part narroa f o w basal wreat uncommof ho n design f Wroxeterc ; , piII ; Camelon, XIV16 . Date, 10 g :fi ? Flavian.

Samian Ware-plain Form27 15* From Antonin eramparI t materia wale sectiof th lo n li n B. Small rim-fragment with typical rounded undercu internad an m lri t groove just below rim; diameter uncertain. 16* From the upper fill of the channel immediately to the W of the Flavian principia. Small fragment of rim and wall, with external and internal groove just below level of rim; diameter c 3-5 in. 17* From loose soil above the SE corner of the Antonine granary. Small rim-fragment with incised line on inner face just below level of rim; flattening of upper part of outer rim surface gives it a nearly triangular section; diamete 3-r. c 5in 18* From loos Antoninef o soi d l en abov S e e Buildineth g III. Small rim-fragment with internal groove below rim; diameter uncertai t possiblnbu y nea in5 r ; very badly abraded. Form 33 19* From loose soil above the N wall of the Antonine granary. Small rim-fragment exhibiting internal groove, jus tmarke d beloan p wli d external concavity; diamete 4-7c r . Date5in : Antonine. Forms 18/31 or 31 20* From between tumbled stones of W wall of Antonine Building III. Small fragment of the footring of a platter, form 31 or 18/31; outer diameter of footring c 4 in; found with sherds of black-burnished ware (infra, p 195, no. 12). Date: ? Antonine. 21* From surface of Antonine II intervallum street on the E front. Small rim-fragment of a platter, probably form 18/31R; diameter uncertain, probabl y9-; founc 5in d with wall fragmen blackf o t - burnished cooking-pot (infra, p 195, no. 20), Date: Antonine. 22* From upper leve f debrio l Antoninn si e smalBuildino Tw l . conjoiningIX g fragment base th ef o s of a platter, of form 31 or 18/31, exhibiting internal rouletting and the incised letter V or A (see infra, p 197); found with fragments of black-burnished bowl and cooking-pot (infra, p 196, no. 39 and p 195, no. 16). Date: Antonine. 23* From beneath gravel layer which seal fillee sth d Antonin emiddlI e ditc the-n ho W frontsmalo Tw .l conjoining fragment basf uppeso d ean footrinre parth platterf a to f go , form 3r 18/311o ; internal diameter of footring about 2-5 in; found with fragments of a black-burnished cooking-pot (infra, . 23)195p no . , Date: Antonine. 24* From between the tumbled stones of the W wall of Antonine Building III. Small rim-fragment of a platter, probably form 18/31; diameter c 11 in; found with fragment of a black-burnished bowl or platter (infra, p 196, no. 36). Date: Antonine. Mortaria (fig 18)15 1 From the layer of debris overlying the Flavian intervallum street on the W front. About one-quarter o vessefa friabln li e orange-buff fabric with variegate incomplets i dm gritri e ; diameteTh e . in 2 1 rc rudimentare buth t y bead suggests kinship with Bushe-Fox's types 10-18 (Wroxeter 19)g fi , ;I cf also Brough, . Date15g fi 22 , Flavian? : . 2 From immediately above subsoil to the E of Flavian Building V. Rim fragment in very friable sandy buff fabric with mainly white trituration grit extending overime th r; diameter uncertait nbu c 9 in; profile resembling Forden Goer, fig 18, 1; poor fabric similar to nos 5, 6 and 7 below. Date: Flavian. 3* Fro make-ue mth pCross-hale flooe levelth th f Antonin e o rf so th n li e principia. Small wall-frag- ment in soft orange-pink fabric with fairly large opaque white trituration grit. Date: probably Flavian. 4 From the layer of debris immediately overlying the Flavian granary. Fragment of rim and wall in coarse greyish-buff ware with variegated trituration grit extending over rim; burnt; diameter c 12- . 'Possibl5in 95-120t certain'yD no EnglandA madE t c N bu ,f , Richborough n c e;i so f i ; , V pi LXXXVII . Date37 , Flavian? : . 19 | 2PROCEEDING S OF THE SOCIETY, 1971-2 5 From the same level as no. 4 above. Small rim-fragment in sandy buff fabric with orange-pink core; diameter unknown; flat rim with small bead. 'From Gillam 239; manufacture of some of these form Gallin si a Belgic possibles ai vase th tt majoritbu , froe yar m Britain therd an ,ever s ei y likeli- hood that many, if not all, in this fabric were made in SE Britain, perhaps Kent. Date: c AD 70-100.' 6 From the layer of debris immediately overlying the Flavian via principalis in front of the granary. Flange fragment from Gillam 239 in sandy buff fabric with white trituration grit extending over rim; diameter unknown differena ;t fabribu sourcd 6 . tcan sam e vesselno e th s ea . 'The extreme friability vers i 7 y ofted an n6 , noticeabl5 s ofno mortarin ei a from this pottery which have bee acin i d soil fo lona r g period. Date Ac : D 70-100'. 7* From the upper fill of the channel to the NW of the Flavian principia. 'Several small wall-fragments of similar fabric to nos 5 and 6. Date: c AD 70-100.' 8 From loose soil betwee middle n th outermos d an e t Antonin frontW e I ditche. th 'Rimn o s - fragmen harn i t d buff ware exhibitin letter laso e gth tw t diagonaf so f o l stamp A , fro 3 e mdi V ] , Albinus who worked in potteries near Watling Street between Verulamium and London c AD 65-95. latese th f ford to productsan me e on suggesdi t e no .Th s tMor thawa t i et thastamp0 n25 f so Albinus are known from sites throughout England, Wales and occupied areas of Scotland (Inch- tuthil, Castledykes, Newstead, Loudoun Hill (3), Camelon (2).) No other potter is attested by as many as 100 stamps!' Date: Flavian. 9 From the filling of the Antonine I innermost ditch on the S front. Four conjoining wall- and rim- fragments, three conjoining base-fragment harn i s d smooth pink-buff fabric with ochre core; variegated trituration grit; base diameter c 4-3 in; rim diameter c 12 in. 'This piece could have been probabls i , so mad Britaif N i y n ed i earl nan y second-century t doe i ;t fal no sl inteasily oan y recognisable category'; cf Balmuildy, pi XLII, 36. Date: ? Hadrianic-Antonine.

FIG 18 Mortaria (1:4) MAXWELL: EXCAVATION THT SA E ROMAN FOR CRAWFORDF TO , LANARKSHIR3 19 | E 10 From upper occupation Antonine levelth n si e principia. Thirty-one fragment f wallo s , basd ean smootn i m ri h hard buff fabric; variegated trituration grit, much worn diamete; m in ri ; 5 1 c r base diamete 'Hartshill-Mancette. in 6 rc r potteries for;a m particularly favoure luniuy db s Loccius and Locciu . Date- o sPr : AD 140-180.' 11 Fro Antoninfilline e mth th f go emiddlI front fragmenW m ee ditcRi . th harn n hi o t d yellow-cream fabric grio n ; t visible; diamete . Date5 , Mumrillsf c 10-c r 91 : ; Antonineg 5in fi , I . 12* From the surface of the Antonine II intervallum street on the W front. Base fragment in fairly hard pale yellow-buff fabric with variegated grit. Date Antonine? : . 13 From loose soil above Antonine intervallu front fragmenW m e Ri stree. th fairln n i o tt y hard pinkish buff fabric with a grey core and variegated trituration grit extending over rim; diameter c 12-5 in probably; badly abraded; cf Old Kilpatrick, pi XIX, 15; Balmuildy, pi XLI, 23. Date: Antonine.

Flagons and Jars (fig 19) 1 From the layer of debris immediately overlying the Flavian intervallum street on the W front. Seven fragments constituting most of the rim, neck and ribbed handle of a ring-necked flagon in hard pinkish orange fabric with grey core; cf Newstead, fig 33, 1-4; Wroxeter 1923-7, fig 44, A22; . Date:1 , 2 Malton g Flavianfi , II . 2 From the 'lower layer of rubbish' below the floor of Antonine Building IV.1* Small rim-fragment of ring-necked flagon in hard buff fabric with grey-brown core; cf Wroxeter I, fig 17, 1. Date: probably Flavian. 3* From below Antonine Antonine clath yf o laye d e ren granary immediatelW e th . f Baso yS e fragmen f coarso t e pink fabric with internal groovin pronounced gan d footring, possibly froma flagonKilpatrick,d Ol f c ; pi XXI . Date:,17 Flavian? . 4* From the layer of debris overlying laid gravel in the verandah of the Flavian principia. Two small fragment wal e necd th lan f kso probabl narrow-neckea f yo sandf o r dyja buff ware with red-orange core; mouldin t junctioga n between nec wald groovind kan an l shoulden go r visible. Date? : Flavian. 5 From beneath the Antonine II rampart cheek of the W front. Small very worn rim-fragment, probably fro mnarrow-neckea flagor o r bufdn nja i f fabric with darker surface; diameter uncertain, but probably c 2-5 in; cf Gillam 22 and Wroxeter I, fig 17, 5. Date: ? Flavian. 6 Fro layee m th debri f ro s overlyin vie gath prindpalis Flaviae fronn th i f o t n granary fragmentm .Ri , probabl wide-neckea f yo d jar harn ,i d orange fabric with light grey core diametem ;ri r uncertain, t probablbu profil 7-yc m 5ri uncertaiins e ei ;th n becaus damagef eo stria t raisef po bu , d clan yo the neck may be the result of intentional rustication; found with fragments of mortarium of Ist-century date (supra, p 192, no, 8). Date: Flavian. 7 From loose soil between middle and outermost Antonine I ditches on the W front. Rim fragment r witja o hfa everte harn i dm dgreri y fabric with darker slip fragmen;e diameteth ; in 6 t rc exhibit s a shallow external groove 1 in below the neck and traces of rustication; cf Ilkley, fig 9, 8. Date: Flavian. 8 From loose soil above the SE corner of the Antonine granary. Small rim-fragment of a narrow- necke grittn i r ydja grey fabric with trace f bufso f slip r ;shap diametefo ; f Duntocher,ec in 4 c r . Date15 , Antoninefi? : g16 .

Beakers and Cooking-pots (fig 19) 9 From between tumbled stone wals W froAntoninf lo e mth e Building III. Several conjoining frag- ments composing about one-third of a small corniced-rim beaker in Castor ware, with barbotine decoration, the so-called 'hunt cup'; cf Gillam 84/85. Date: Antonine, not much before AD 170 (see supra, 176)p . 10* From the upper level of debris inside Antonine Building IX. Very small wall-fragment of a beaker in Castor ware with barbotine decoration; found with wall fragment black-burnishef so d ware cooking- pot (infra, p 195, no. 22) and fragment of samian ware (p 190, no. 9). Date: as for no. 9 above. 11 From the surface of the floor of Antonine Building IV. Two conjoining rim-fragments and a wall fragmen short-rimmea f o t d cooking-po black-burnishen i t d ware, exhibiting rivet-holes made during repair of the vessel; diameter c 4-5 in; cf Wroxeter 1923-7, fig 46, C6. Date: Antonine. 194 I PROCEEDINGS OF THE SOCIETY, 1971-2

v~\ 2==&? f—m " \

I* T 15

161 V 26 s 19

27 28' 729

32 30

33

r 35 31

36 I 37 38

40 \41 \ 42 \ 43

\46 45 \ 44

47 FIG 19 Coarse ware (1:4) MAXWELL: EXCAVATIONS AT THE ROMAN FORT OF CRAWFORD, LANARKSHIRE | 195 12* From the same level as no. 9 above. Several small wall-fragments of a cooking-pot in black- burnished ware; found with fragmen footrinf o t f samiago n platter (supra, 191p . 20),no . Date: probably Antonine. 13 From the Antonine II intervallum street on the S front. Very small, badly abraded rim-fragment of a cooking-pot in black-burnished ware, with fairly upright beaded rim; diameter uncertain; cf Mumritts 11 g . Datefi 7 , , II : Antonine. 14 From below the Antonine I intervallum street on the N front. Two very worn fragments of the rim and walcooking-poa f o l black-burnishen i t dbein m ware ri fairlf ge o th , y upright beaded profile; diameter uncertain; cf Gillam 122. Date: Antonine. 15 From the Antonine II intervallum street on the N front. Rim fragment of cooking-pot in black- burnishe dbein m wareri f fairlge o th , y upright beaded profile; diameter uncertain t probablbu , y c 6-5 in; cf Gillam 119. Date: Antonine. 16 From the same level as no. 10 above. Rim fragment and three lattice-decorated wall-fragments of a short-rimmed cooking-pot in black-burnished ware; diameter uncertain, but probably c 4-75 in; cf Gillam 118; found with fragment f samiao s nf blacko platte m -ri rd (supra,an ) 22 191p . no , burnished bow platter o l r (infra, 195p . 38) ,no . Date: Antonine. 17 From the same level as no. 13 above. Small rim-fragment of a cooking-pot in black-burnished ware with wavcavettd an y m linori e decoratio necke th n ;n o diamete r uncertain, Mumrillsf ;c 12 g fi , II 36. Date: Antonine. 18 Fro same mth e. 1 leve3no aboves a l . Small rim-fragmen cooking-poa f o t morn i t e friable black ware with greyish core; the rim is similar to that of no. 17, immediately above, but slightly less curved; soot-encrusted; diameter uncertain. Date: Antonine. 19 From the same level as no. 10 above. Six wall-fragments and one rim-fragment of cooking-pot in coarse orange-brown ware; the rim exhibits an everted flaring profile and there is lattice decoration on the wall; rim diameter c 4-5 in; cf Trentholme, fig 29, 8; Mumrills II, fig 12, 36. Date: Antonine. 20* From the surface of the Antonine II intervallum street on the E front. Very small wall-fragment of cooking-po black-burnishen i t d ware exhibiting lattice decoration; found witfragmenm hri f o t samian platter (supra, p 191, no. 21). Date: Antonine. 21* From beneath tumbled stone walN Antoninf f o lso e granary. Wall fragmen cooking-poa f o t n i t fine orange-brown ware exhibiting lattice decoration. Date: probably Antonine. 22* Fro sam e mabove0 th wall-fragments 1 o e. leve.no Tw s la cooking-poa f o , black-burnishen ti d ware, one exhibiting lattice decoration; found with a fragment of a Castor-ware beaker (supra, p 193, no. 10) and a fragment of samian ware (supra, p 190, no. 9). Date: probably Antonine. 23* From below the gravel layer sealing the filled Antonine I middle ditch on the W front. Two fragments wal e bascooking-pod oa th ff an le o black-burnishen i t d ware exhibitine on , g lattice decoration; found with basal fragments from a samian platter (supra, p 191, no. 23). Date: probably Antonine. 24* From the easternmost post-hole of the S portico in the Antonine.pnncipfa. Small worn wall-fragment ocooking-poa f r beakeo t black-burnishen i r d ware; found witfragmentm hri f samiao s n bowl, for (supra0 m3 . Date 7) 190p , . : no ,probabl y Antonine. 25* From the floor of Antonine Building V. Fragment of base of a vessel, probably a cooking-pot, in black-burnished ware. Date: probably Antonine. 26 From loose soil above the S end of Antonine Building III. Small rim-fragment of a cooking-pot in black-burnished ware, with short beaded rim; slightly soot-encrusted; diameter probabl. in 5 yc Date: Antonine. 27 From Antonin loose th f eo soid el en granaryabovW e conjoininex th .Si g rim wall-fragmentd -an f so short-rimmea d cooking-po beaker to black-burnishen ri d ware diametem ;ri 3-cr; c f 5 in Birdoswald, fig 15, 35. Date: Antonine. 28 From loose soil in the same area as no. 26 above. Three small rim- and wall-fragments of a cooking- pot in black-burnished ware with fairly upright beaded rim; neck soot-encrusted; diameter probably c 5 in. Date: Antonine. 29 From loose soil in the same area as no. 26 above. Rim fragment of a cooking-pot in black-burnished ware with plain outcurving rim; diameter uncertain f Castledykes,c ; . Date3 , 42 : Antonineg fi . 30 From loose soil immediately to the E of Antonine Building LX. Nine fragments (six conjoining) of the rim and body of a soot-encrusted cooking-pot with fairly upright beaded rim; the body exhibits lattice decoiation nece th , kpattera f obliquno e scoring diametem ri ; f Mumrills c 5-7rc ; 5in II, . Date12 , fi: g11 Antonine . 196 | PROCEEDINGS OF THE SOCIETY, 1971-2 31 From loos Antoninf eo soid len abovN e Buildine eth . EighgIX t fragments (five conjoininge th f )o rim and body of a cooking-pot in orange gritty ware with dark inner surface; the rim is beaded and fairly upright; decoration consist lattica f so e wavpattera bode d th necke yyan n th nlin o n ;eo diametem ri 6-rc 5 in f Mumrillsc ; 11 g . Date9 fi , , II : Antonine. 32 From loos same soith . en 3i l 1 areno aboves aa . Three fragmentbasd a f an walle o th m f ri , o s soot-encrusted cooking-po black-burnishen i t d ware fairlm ri ; y upright diametem ri ; r uncertain t probablbu y5-5c in f Trentholme,;c . Date6 , 27 : Antonineg fi . 33 From loose soil in the same area as no. 27 above. Very small rim-fragment of a cooking-pot in black-burnished ware with plain outcurving rim; diameter uncertain. Date: probably Antonine. 34 From Antoninf looso E e soith eo t lBuildin g fragmen VIm Ri . cooking-pof o t black-burnishen i t d ware with plain outcurving rim; diameter uncertain; cf Trentholme, fig 28, 11. Date: Antonine. 35 From loose soil above the S end of Antonine Building IV. Rim fragment of cooking-pot in black- burnished ware with plain outcurving rim; diameter Balmuildy,f probablc ; in 5 yc pi XLV, 1-4. Date: Antonine.

Bowls plattersd an ) (fi19 g 36 From between tumbled stones of the W wall of Antonine Building III. Rim fragment of a flat- rimmed bowl or platter in black burnished-ware with lattice decoration; rim diameter uncertain f Balmuildy,t probablc ; bu in 7 yc pi XLVII Mumrillsd ; an foun101 g 9 22 ,fi , , df I o witm hri samian platter (supra, p 191, no. 24). Date: Antonine. 37 Fro surfac e mAntoninee th th f eo intervalluI I front conjoininmS o e streeth Tw . n to g rim-fragments oflat-rimmea f d platte black-burnishen i r d ware diametem ri ; r uncertain t probabl; bu , in 8 yc cf Gillam 308; Cardurnock, fig 12, 28. Date: Antonine. 38 From the surface of the Antonine II intervallum street on the N front. Very badly worn rim frag- ment of a flat-rimmed platter or bowl in black-burnished ware; diameter uncertain; cf Cardurnock, fig 12, 28. Date: Antonine. 39* Fro uppee mth r laye debrif o r s inside Antonine Buildin . SmalgIX l very badly worn rim-fragment oflat-rimmea f d platte r bowo r black-burnishen i l d ware; lattice decoratio bodyn no ; diameter uncertain; possibly similar to Mumrills II, fig 11, 17; found with fragments of samian platter (supra, . 22)191p no . , Date: Antonine. 0 4 From loose soil abov Antonine eth e intervallu front fragmenW m e Ri stree.flata th f n -o to t rimmed platte bowr ro black-burnishen li d ware with lattice decoratio bodyn no ; diameter uncertain; cf Balmuildy, pi XLVII . Date7 , : Antonine. 41 From loose soil in the same area as no. 35 above. Rim fragment of a flat-rimmed platter or bowl in black-burnished ware with slight trace f latticso e decoratio bodyn no diamete m ;ri r uncertait nbu possibly c 9-5 in; cf Castledykes, fig 47, 13; Cardurnock, fig 12, 34. Date: Antonine. 42 From loose soil above the W end of the Antonine granary. Rim fragment of flat-rimmed bowl with wavy decoration on the body; rim diameter uncertain but probably c 9-5 in; cf Mumrills II, fig 11, 14. Date: Antonine. 43 From loos same e above 2 soith 4 fragmen e. n m li areno .Ri flat-rimmea s aa f o t d platte bowr ro n li black-burnished ware with lattice decoration on body; rim diameter c 7 in; cf Castledykes, fig 47,10. Date: Antonine. 44 From loose soil above Antonin front E I rampare I e th .flata sid d Fragmenn f an e-o o t m ri f o t rimmed bowl in a gritty burnished fabric with brown-orange surface exhibiting lattice decoration and chamfer; diameter probably c 8 in; cf Mumrills II, fig 11, 15. Date: Antonine. 5 4 Fro layee m th debrif ro s immediately overlying Flavian Buildinfragmenm Ri curvinga . f gVI o t - sided platter in soft red-orange fabric; diameter c 7 in; cf Gillam 324. Date: Flavian. 46 From below the Antonine cobble-layer immediately S of Flavian Building IV. Rim fragment of a curving-sided platter in chalky white buff fabric with traces of orange-tawny slip; diameter uncer- Gilla f taipossiblt c ; nbu in m 8 y325c . Date Flavian? : . 47 From the layer of debris immediately overlying Flavian Building VI. Seven conjoining fragments of a curving-sided platter in fine hard greyish ware with red-brown mica-dusted surface externally and trace f burninso g internally; diameter uncertai 7-; c difficul 5t in nbu parallel o t clearle t th t ybu , forme indicatiodn a footrine b y f earlno gma y date f Camulodunum,;c pi XLIX, 16A-C. Date: ? Flavian. MAXWELL: EXCAVATIONS AT THE ROMAN FORT OF CRAWFORD, LANARKSHIRE 197

Lids 48* From below Antonin I intervallueI frontN m e stree.th Smaln o t l fragmen outee th f ro tcircum - buff-orangn i ferencd li a f eo e fabric with mor lesr eo s squared edge Gillaf ;c m 339. Date: Flavian or Antonine. * Fro49 uppee mth r Antoninlevele th f so e intervallu frontmE e streeth . Vern o t y small worn fragment outee oth f r circumferenc lighn i d t li orange-bufa f eo f fabric with simila above8 r4 profil. no . o et Date: Flavian or Antonine.

Graffiti (pi 19a) Two of the fragments of pottery found during excavation bore incised graffiti; the pieces were referred to Mr R P Wright, whose comments are reproduced below.16" 1 From loose soil at the N end of Antonine Building IX. One of two conjoining base fragments of a samian dish, probably form 18/31 R, with the letter V or A incised upon its interior face. 'If it is V it may indicate the number 5. Otherwise a single letter sufficed to give it a distinguishing beginnine th s mark texa wa f r gto o ,neve r completed.' 2 From the rilling of the Flavian innermost ditch on the S front. A wall fragment of an amphora bearing the incomplete graffito A Q [ on its exterior. 'The graffito on the amphora is a simple one, probably indicating the purpose for which the vesse re-useds wa lexcavato e Th . suggestes rha VA[ thid Q san , dA seem more sth e likely inter- pretation CallendeH M r citeds D .ha r 17 some instances where secondara ther s f eo wa e yus amphorae as water-containers. It is less likely that the graffito cited the personal name of an owner, such as Aquila or Aquilinus. As it came from the debris thrown into the Flavian innermost ditch it cannot have been used to contain the ashes from a cremation'.

outer surface FIG20 Crucible fragment: top, viewed from above; lower left, external surface showing pincer impression; lower right, section (1:2) 198 I PROCEEDINGS OF THE SOCIETY, 1971-2

APPENDIX I Metal-working at Crawford There is evidence in the form of slag to suggest that iron-working was being carried on at Crawford in the Antonine II period on a fairly extensive scale. The hollow of the already sealed Antonine I broad ditch on the W front contained a considerable quantity of iron slag which had clearly been deposited there with other rubbish when the ditch had been out of use for some time (see supra, p 158); a smaller quantity of slag was found amongst debris filling the emptied post-hole pits of the Antonine II N gate (p 165). The source of this material may have been the narrow building (No. IX, see p 176) which stood at the eastern extremity of the praetentura and has been tentatively identified as a fabrica or workshop. If that identificatio corrects ni possibls i t ,i e that Buildin als s scene owa th brass-workingf eo X g I associater fo , d with the slag found in the N gate post-hole was a fragment of a crucible which had apparently been used for that purpose. This ite bees mha n subjecte intensivo dt e examinatio J McKerrel H e r th D f y o nlb National Museum, upon whose report the following description is largely based.« e fragmenTh t , p(fii20 g19b) appearc , belono t s roughla o gt y hemispherica r bag-shapeo l d crucible, probably with an external rim-diameter of about 4 in and an original capacity of at least 250 cc; maximue th m surviving thicknes sucs 0-6A c woulht . s i i 5 in d rank amon largese gth t example cruciblef so s from Roman or Romano-British sites.19 The outside surface of the fragment is partially covered with a red and black glass-like layer in which the impression of the tongs used to hold the crucible is clearly visible. From this it would appear that the crucible was removed from the source of heat, from above, by means of fairly narrow tongs, perhap t unliksno e those foun t Newsteadda cruciblfabrie e . 2th " Th f co e appear havo st e bee nfireclaya - like material, thren builti p eu layer differinf so g compositio tha y mos e sucn i tth wa hta durabln o s ewa

the mosoutside th td resistanean actioe th moltef o nt o n inside. e metath n o l

1 X-ray fluorescence analysi outee 2 inned th f ran so r surfaces provided evidence tha metae e us th t n i l was a copper-zinc alloy (brass). In view of the fact that a furnace heat of about 1100°C would be required for the working of this metal - a temperature at which, teste the tong-impressions described above, the outer surfac crucible th f eo e would begi softeno nt cleas i t i ,r tha degree th t skilf eo l involve usinn di e gth metal t inconsiderablewoulno e db . The rim fragment of a second crucible was found in loose soil above the W end of the Antonine granary. Although similar in most respects to that already described, it appears to belong to a somewhat smaller vessel, in which the layered structure of the former example is not so clearly defined.

APPENDIX II Identificatio Botanicaf no l Specimens The waterlogged conditions of the middle Antonine I ditch on the northern and western fronts ensured the presentation of a number of botanical specimens as well as artefacts made of organic materials (see supra, p 188). The floral remains were submitted to the Royal Botanic Garden, Edinburgh, and the following identifications were made: 1 Fro peae mth tmiddl e stacth n ki e Antonin eI ditch , sectio . 'MossnA : Hylocomium splendens (Hedw.) B. & S., a fairly common bryophyte found growing in woodlands and amongst grass and heather.' 2 From the peat-block filling of the middle Antonine I ditch, section C. 'Moss: Polytrichum Commune Nut: Hazel (Corylus Avellana) Twig Branchesd san : Hazel (Corylus Avellana) Birch (Betula sp.) Willow (Salix Alba)' Alwooe lth d fragment beed t witsha n cu sharha p instrument largese th , t pieces being exclusively birch. As such they constitute evidenc activite t onlth eno f yo y whic beginnine hth t wenlase a th tn f to periog o d of occupation - the cutting and trimming of materials used in wattle screens for barrack walls and rampart breastworks - but also of the local vegetation in Roman times, mixed deciduous forest, steadily MAXWELL: EXCAVATIONS AT THE ROMAN FORT OF CRAWFORD, LANARKSHIRE | 199 being encroached upon by heath and peat-bog. Such natural reduction of the local timber would obviously have been accelerate presence th areRoma e a y th f db ao n ei n fort, wit constans hit t foragin r fuegfo l (lignatio) recurrens it d an t nee smalf do l timbe reconstructior rfo r repaino r work.

NOTES 1 Reference Coheno t e sar , Description historique medailless de frappes sous I''empire remain (1880)o t , Roman Imperial Coinage, i (1923), ii (1926), to Coins of the Roman Empire in the British Museum, i (1923) i (1930)i , i (1936)Romano ,ii t d ,an Imperial Huntere Coinsth n i Coin Cabinet, i (1962) i i , (1971). la For the significance of the as and the sestertius of AD 77-8 and the as of AD 86, see 'Two Groups of Roman asses from Northern Britain Numismaticn i ' Chronicle, PSAS,d an 3 ; 196810 ff 1 6 , alse (1970-1)se o ; supra,ff 1 13 177,p . authoe Th 2 r wishe acknowledgo st r McKerrel Nationadebe s D estafth hi e o t f th o f d llan Museum of Antiquitie f Scotlanso wero dwh e concerne examinatioe th n di r preservationo f smalno l finds from Crawford. authoe CollinH Th indebtes G i rr thi 3 r sfo s M information o dt . 4 Lacaille, A D, The Stone Age in Scotland, 187 ff. 5 Bar Hill, 101 ff, fig 36, 7. 6 Newstead, 150 ff, pi XX. 7 Berichten Rijksdienste d n va voor het Oudheidkundig Bodemonderzoek, ix (1959). ff 8 6 , 8 Balmuildy, p, iff LVII8 9 . 2 , 9 Ibid, pi LVI. 10 , 0 Newstead,1 280p p , 311, pi LXXXIII alsf c o; 3 ,Jahrbuch Romisch-Germanischens de Zentralmuseums Mainz, 8 (1961), 98. 11 PSAS, Lxxn (1937-8), 386 ff. 2 c1 f PSAS, LXXI (1936-7). 68 , 13 Jope (ed)M E ,, Studies Buildingn i History,. ff 9 3 14 The author gratefully acknowledges Mr Hartley's assistance in identifying this piece and no. 9. 15 The author is grateful to Mrs K F Hartley who kindly inspected the majority of mortarium fragments; relevane r reporth he n o t t item bees sha n include individuae th n di l descriptio eachf no . 16 This sher recoveres dwa d durin excavatione gth t ProfessoJosepf so S S 1938n hi K J r . authoe 16 Wrigh assistancP aTh s gratefuR hi s r i r r compilatioe tfo M th o t eln i thif no reportse parth f to . 17 Callender, M H, Roman Amphorae, 35 18 A detailed report on the crucible fragment has been prepared by Dr McKerrell for publication in a forthcoming volum Archaeometry.f eo 19 Tylecote , MetallurgyF ,R Archaeology,n i . ff 0 13 20 Newstead, 286p ; pi LXffl. 4 , 21 cfll/cley, fig 14, 9.

ACKNOWLEDGMENTS The writer is indebted to Mr J T Haddow, then proprietor of Castlemains farm, for ready permission to excavate at Crawford over such a prolonged period, in spite of the many incon- veniences this must have caused. Particular thanks are due to Dr A S Robertson of the Hunterian Museum, University of Glasgow, without whose assistance and advice the first season of the investigation could not have been carried out. Dr Robertson has furthermore contributed a coie reporth n n findo tprovided an s d facilitie inspectioe th r fo s f thano materiae t th par f o t l from Crawford which has already been deposited in the Hunterian Museum. Thanks are also due to Dr K A Steer, Secretary of the Royal Commission on the Ancient and Historical Monuments of Scotland, whose counsel and encouragement, especially in the early stages of the work, were much appreciated. 200 | PROCEEDINGS OF THE SOCIETY, 1971-2 So many people have participate excavatioe th n di t variouna s times tha t wouli t e db impossible to mention all by name, but the following deserve a special word of thanks for the part they have played in the supervision or organisation of different phases of the investigation: my wife, Kathleen; my colleagues, A MacLaren, J N G Ritchie, and I G Scott; Messrs R R Craig GoldsackM I , Henderso W ,I Parker D d addition I .n an n generous financia othed an l r assistanc receives ewa d from FifLanarkshird ean e County Council furtheo st e wore th th r n ki seasons 1961-3 196n i investigatioe d 6th ,an headquartere th f no s buildin materialls gwa y aided by a grant from the Society of Antiquaries of Scotland. The line-drawings which appear in this report are the work of S Scott and D Boyd. Finally and not least to all those who contributed to the success of the enterprise by digging or travelling, often in the most inclement of weathers^ including the pupils and staff of schools in Fife and Lanarkshire, as well as members of the Extra-mural Class in Archaeology at the University of Edinburgh, the writer wishes to record his sincere gratitude.

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