Excavations at the Roman Fort of Newstead, 1947. By

EXCAVATION E ROMATH T SNA FOR NEWSTEADF TO , 1947Y B . PROFESSOR IAN A. RICHMOND, M.A., LL.D., F.B.A., F.S.A., F.S.A.ScoT. Read April 12, 1948.

The Roman fort at Newstead was thoroughly and skilfully excavated by the late Dr James Curie 1 in a great continuous series of campaigns covering the years 1905—10. Dr Curie's excavations, which were far in advance of contemporary practice, disentangled a detailed plan of much of the fort and s surroundingsit , produce dcollectioa relicf no notablo s e that they have never since cease excito dt e wonder, and, above all, provide da genera l chronological scheme for the occupation. They demonstrated conclusively that ther bee d maio eha n tw n period occupationf so firse th , t commencing under lulius Agricola in A.r>. 81, the second under Lollius Urbicus in A.D. e principa139—40th f o e . l on distinctionIndeed s wa t i , f Jameo s s Curie's work, and there were many, that it first established in Britain a sound differentiation between pottery of these two epochs. But within the two main periods, whose commencemen welo s s lti determine botn di h casey sb literatur confirmed ean secone th n di d cas epigraphyy eb phasee th , s proved difficult to define. Dr Curie thought that he could distinguish five distinct earle th phases yn i perioo tw ,occupatiof do e th t latee thred Bu th nan .n ei determination was loose and unsatisfactory. The truth was that, while the evidence had manifested itself piecemeal, in many ways and in many parts of the site, no systematic attempt was made to correlate the data stratigraphically thao e s remain, th l tperiod al f o s s coul e viewedb n di direct relationship. To expect such an effort would be to ask for something out of its time, almost thirty years too early. And a generation later, whe tune comd nth eha describo et Royae th site r elth efo Commission no Ancient and Historical Monuments for Scotland, Dr Curie himself felt this strongly. Befor expectatioe eth 193 n i r 8 laiwa f dparalysinna o g hand upon l majoal r archaeological projects worked ha t wite writee h ,dou hth plaa r n to cut trial sections across the fort in order to make a fresh and systematic study of the stratification. All we could then do was to plan and to hope that better times would permit excavatio futuree t therth n Bu ni e. were nine long years to wait, and when they were over Dr Curie was no longer with us. Only his inspiration remained, changing the work from a long- awaited collaboration into the dutiful fulfilment of a pledge. 1 James Curie, A. Roman Frontier Post and its People: the Fort of Newstead in the Parish of Melrose (Glasgow, 1911); hereinafter referred to as Newstead. 1 VOL. L.XXXIV. 2 PROCEEDING E SOCIETYTH F O S , 1949-50.

The work itself was carried out between 26th August and 24th x Septembe o sectionr Tw 1947.s fee0 wer8 e firs tey Th la tcut ) (fig. 1 . middl e easternmose easth th f f o te o t sout t acroshcu whole gatd sth ean e system of defences and well into the interior of the fort, crossing the fourth contubernium from the west in Curie's Barrack I. Its special objective was e secondarth t earlybu y stone building 2 r CurifounD thin y ei db s quarter of the fort. The second section (fig. 2) was taken 230 feet south of the middle of the west gateway hi the so-called reducing-wall, with the intention of defining the relationship of this wall to the occupation levels within the resulte fortTh eac.n i s h case were highly fortunate. Almost everywhere e stratificatioth n remained completely undisturbe y formeb d r word an k supplied, together with much dating material, the clearest possible definition of periods, so as at once to confirm and to simplify Dr Curie's interpretation of the site. THE EXCAVATIONS. Phase I. The Agricolan Occupation. A first contact (figwit) 1 Agricolae . hth n defences madditcs e th wa t he a discovered by Dr Curie just behind the Antonine fort wall. This was a steep- sided V-shaped ditch (PL I, 1) cut in rock, 5 feet deep and 8 feet wide at the top, narrowin channea o gt l abou foo1 t t squarbottome th n t ea Te . feet in front of this ditch lay an outer ditch, not hitherto disclosed on this half of the site, of which the outer lip had been removed by a later and wider s dimensionditchit f I . s were symmetrical t wili , l have measured 9 feet wide by 3 feet deep. hala Foud f an feer t behin innee dth r ditch restind an , g immediately upo e subsoilnth , cam a stone e rampart-foundation 1\ feet wide, formed osingla f e laye mixef ro d stoncobbled bordered re ed an carefullsa an y db y laid front ker f largebo stifn d i r re cobblee stonet f se Th e s. sar (PI 2) , I . clay and are often placed sticking end upwards, so as to form a corrugated surface which would prevent the front of the rampart laid upon it from slidin r squeezingo g forward ove e ditchrampare th r Th . t materiaa s i l stiff whitish clay,3, beaten very hard, and here and there exhibiting layers of trodde nt extend I mud fee3 2 . t r behinsfo outee dth rfounda e edgth f eo - tion or bottoming and then terminates against a small mound or heel of gravel, piled 4^ fee tfoo1 widd t highan e , manifestly intende maro dt k the position chosen for the back of the rampart at the time of erection. Whil rampare bace th eth f ko thus i t s clearly defined frone bees th , ha tn

1 Labour of six German prisoners was kindly supplied through the Roxburgh War Agricultural Executive Council, local labour being unobtainable. 2 Newstead, 65-66. 3 This material was visible from the top of the North Eildon Hill, on 28th April 1951, as a broad white streafreshle th n ki y ploughed fields coverin ramparte sitee gTh forth e . th t f anso d annexed san bath-house were perfectly defined by this notable example of an archaeological soil mark. -LATE DOMITIANIC BUILDING LATE DOMITIANIC -BUILDING -ANTONINE I RAMPAR" AND FORT-WALL- —AGRICOLAN FLOOR-LEVEL REMOVED-* -LATE DOMITIANIC INTERVALLUM -LATE DOMITIANIC RAMPART -*—————AGRICOLAN TIMBER BUILDING- ——AGRICOLAN INTERVALLUM -AGRICOLAN RAMPART- ->AND*-lNNER DITCH-

NEW STEAD 1547 SECTION THROUGH SOUTH DEFENCES AND BUILDINGS

NORTH SOUTH -ANTONINE 1C: THREE DITCHES -ANTONINE I: TWO DITCHES

-TWO \GRICOLAN DITCHES _*LATE DOMITIANIC: ONE DITCH

10 O 10 20 40 100 I I I I ~l— I I I I I —I—- I i 1 FEET

IAN A. RICHMOND. Pig. 1. [To face p. 2. EXCAVATIONS AT THE ROMAN FORT OF NEWSTEAD. 3

obscured by later alterations. Originally, it plainly coincided with the front of the bottoming, but in a later demolition the upper part of the white clay was shovelled forward and a flat working-platform 9 feet wide was e surfacth f o eforme p thuto n sdo created n , whic me churnes e hwa th y db t wora k into dark trodden significancmude Th . thif eo s change, however, concern nexe sth t discusseperios i d dan d. late7) . (p r The Intervallum behind the white clay rampart is represented by a gravel spread inche,6 froo t ms4 thick, covere trampleashey d db an sd dmu and extending for 38 feet. It is then cut off by one of Dr Curie's trenches levelliny b d r latean gfo r Roman building d doe t reappearan sno s e Th . next surviving structural featur f thio e s period occurfee0 5 t t behina s d the rampart back. This is a slot for the foundation of a timber building, and it measures 1 foot square (PI. V, 2). The building itself was 32^ feet overal widthn li norts it , h limit being represente secona y db d slot inche2 1 , s wide and 9 inches deep. Both these slots are only the heels of the foundation-trenches which they represent and of which a complete picture is afforded by a third slot, 12 inches wide by 21 niches deep, which marks the long axis of the building (PL II, 1). A minor subdivision was also found (PL II, 2), a much shallower north-to-south slot, 1 foot square, which turns westward just unde fee5 r t axiae nortth f lho divisio eastward nan d space 5^Th feee sout bordereo . t tit minoe f ho th y drb divisio paves nwa d with light cobbling much worn by treading. Similar cobbling was observed extending farther nort partitione floorinease ho th t sout f th o t o t i t f gho ,bu was of clay only, the cobbling terminating as if at a threshold. In the clay floor there was one small post-hole and an irregular excavation for other timber structure f uncertaio s n significance. Farther north, beyone th d timber building, the ancient ground surface, and any remains of the first period havwhicy ema t hcarriedi agaid ,ha n been remove latey db r buildings. sectione th f norto e almosr ,th d thafo s hen a , feet 7 r is t1 t fa , s thera t eBu wa traco sn timbef eo r buildings t seemi d an s, likely tha site tth e this swa ostreea f t betwee blockso ntw . It will be evident that the remains recovered are no more than a sample from the earliest level at Newstead, but they add much to our knowledge of the Agricolan fort. The discovery of two ditches protecting the eastern half of the fort gets rid of a previous anomaly, whereby one-half of the fort was less well defended tha othere nth ever o , n tha annexese nth . Double postulatee b w ditcheno n d ca severywher e circuit th e massiv n Th o e . e white clay rampart is 25 Roman feet thick, and its rampart-walk can hardly have been less tha fee5 n1 t high, exclusiv e 5-footh f eo t breastword kan merlons which mus addee b t estimatinn di totae gth l height thicknese Th . s at the base is actually greater than that of the contemporary legionary rampart1 at Caerleon; but, when allowance is made for the fact that the Arch. Camb., 1931, 102, figs. 3, 6. 1 4 PROCEEDING E SOCIETYTH F O S , 1949-50.

Caerleon rampar verticaa d ha t l back t becomei , s clear tha Newsteae th t d rampart must have been slighter. It is, however, much bigger than the turf rampart at Fendoch, which was 17 feet thick. The scale of the 1 defences thus suggests tha speciae tth l importanc Newsteaf eo pivotaa s da l position, comparable with Corbridge or Carlisle, was realised from the first an a primd s i thae t i texampl f Agricola'o e s sens r opportunitiesfo e locorum.2 The defences were separated from the buildings of the fort by a 50-foot intervallum, surfaced with som inches\oe6 f sharp fine-grade river gravel. The great width of this open belt, almost double the size of the Fendoch intervallum, probabls i y connected wit face hth t tha cavalra t y garrisos ni attested for this period (see below, p. 26). But the need to protect the building fore th tf sfroo m flaming missiles will als borne ob mindn ei o T .

hit the buildings, an enemy woul3 d have to throw over a hundred feet, even fro vere mth y ditchese edg th kulinf f o e o t ;gou whilrange b o et e himself he would hav throo et w over double that distance. This put buildinge sth s well outside normal reach. The intervallum was, however, also the spot where cooking was done, in cook-houses or ovens situated just behind the rampart; and, although the section did not reveal at this level any structures connected with cooking, the fair quantity of ash on the gravelled surface of intervallume th suggests that cooking installation awayr fa t . no y sla The position of the fort gateways indicates that we are here in the retentura, where the buildings to be expected are either barracks or stables, e overalth d lan feewidt2 3 tf ho attachin e timbeth o gt r building here discovered will suit 4 either equally well. But the irregular internal sub- divisions and the differing types of flooring definitely exclude the regularly planned and uniformly floored contubernia or men's quarters of a barrack, so that, if the building were a barrack, the portion discovered must be the irregular rooms of the decurions' or centurion's suite. This interpretation, however, will not accord with the relationship of the building to the general plan of the Agricolan fort. The portion discovered lies too far west of the south-east angle of the fort to be the administrative end of a barrack lying easwestd normaa an t f ; o e souts suppose i whileth d t li e hen b f i , o dt barrack running from nort southo ht totae , thith t ls fi absenc wilt no l f eo foundations at the north end of the trench. According to our data the buildin onln lyinge ca y b g thi n easwesti d sd tan cas ,an e canno regardee b t d barrack-blocka s a t musI . t therefor interpretee b e stabless da whicn i , h

1 PSAS., Ixxiii. 113. 2 Tacitus, Agricola,. 2 , 22 Fo fire-bearinra g arro r dartw o Macdonal e se , Parkd Romane dan Hill,r Th , Ba e Forts116th n ,o 8 fig. 42 , wit2 , h appropriate Greek references red-hor fo : t sling-bolt wells sa Oaese se , . B.G.,. 1 , v43 . For contemporary barracks of timber, of. Fendoch, PSAS., Ixxiii. 134; for timber stables of 4 Antonine date, see Carzield, JUS., xxx.. pi. viii, and Trans. Dumfriesshire and Galloway N.H. and Ant. Soc. xxii. 157. EXCAVATIONS AT THE ROMAN FORT OF NEWSTEAD. 5 e cobbleth d floor e providear s o accommodatt d e horses r neee No th d. occurrence of pottery in the building surprise us, since the grooms, like stable-lads to-day, lived wit animale hth thein si r charge. AGRICOLAN FORT, NEWSTEAD

Pig. 2.

The identificatio stabla f no e lying easwesd tan t raise whole sth e question e dispositiooth f garrisonind nan Agricolae th f go n fort (fig. Thi2) . s fort, unlik Antonine eth e fort which overlie , facesit d west s easwesd It an .t t gates * are opposite and axial, "while its north and south gates, though type gatewayf Th eo , outline Curie'r D s a dditchee s r onlth wor fa y yo b ks went illustrates i , d mor1e full t Tassiesholma y , JBS., xxxix . , figSom20 99 .. y thes da e e Newstead gates will repay examination. 6 PROCEEDING E SOCIETYTH F O S , 1949-50.

opposite, lie considerably west of the axis. The retentura, containing both principal building barrackd san stablesr so , thus lie eastwardo st e th d an ; prcetentura, containing barracks or stables only, lies to westward, facing up the Tweed valley and overlooking the main road descending x to the New- stea bloce dth ford kn I formin. southere gth retentura,ne halth f fo where eth stable discovere situateds di , ther juss ei tnormae rooth r mfo l barracr ko stabl fee 0 quintana,centraa f principaa o r evi t15 fo longw a r ro d l fo ,lan buildings measuring about 100 feet from back to front. The other blocks are larger, and their dimensions afford no clue to how the buildings may have been arranged inside themt therBu abundans e.i t space,e eveth n i retentura, for sixteen single blocks of stables or barracks, and, allotting two turmce to each block, this means that two-thirds of an ala milliaria coul housee db d there morn O e. generous spacing, ther ampls ei e roor mfo 12 block accessord san y building botn si h prcetentura retentura,d an thao s t an ala milliaria may probably be assumed. It will be observed that pits containing Flavian relics were particularly rich in objects expressly associated with cavalry. The splendid trooper's helmets, which were the crowning glory of Dr Curie's excavations, all came 2 from early pits. There must also be mentione e beautifudth l horse-trapping se richl 3 th fro t d Lympi Van decorated leather chamfrons * from pits LXXVII OId ILan The period of this first occupation, which is now seen to be associated wit ha permanen t fort s secureli , y linked with Agricol a fortunat y b a e numismatic discovery. Just beyond the axial division of the building identified wit hstablea sestertiusa , northe finine founs th th f 5wa n go di - ward partition, with other relics of occupation. It was reduced to wretched condition by corrosion, but the surviving portions of the original surface showe dbeed thaha n t i tvirtuall minn yi t conditio ntumbles whewa t ni d int trencdemolitioe e oth th y hb n partyobverse Th . e still exhibite dheaa d of Vespasian, with the legend [IMP CA]ES V[ESPASIANVS], the reverse legene th d AEQV[ITAS AVG^VSTI], both legends reading t inwardno d san outward thereford san e belongin earliee th o gt r part Vespasian'f o s princi-

pate. Such a coin, almost in8 mint condition, is exactly what might be expected to occur as current cash in Agricola's army. To this dating the pottery recovered (see below . 33),p particulan ,i fragmente rth globulaa f so r Samian beake Dragendorff'f ro fina s d eshap broad-flangean 7 e6 d mortarium found in association with the early rampart and intervallum, afford con- firmation without adding anything mor e duratioeth r precisefo f ns o A .

1 Newstead, 2-3. 2 Ibid., pis. xxvii, xxviii, xxix, fro t xximpi i (se. 121—22)epp ; pi. xxx, fro t Ivi. 128-29)mpi (p i . » Ibid., 128, pi. Ixxii. • Ibid., 134, pi. xxi (pit Ixxviii), 155 (pit cii). • Cf. BMC., ii, no. 600, pi. 23, 8, though the coin is not sufficiently well preserved to make quite sur issue A.Df . eo th s e71 .i 6 BMC., a. p. xxviii, A.D. 69-73. EXCAVATIONS AT THE ROMAN FORT OF NEWSTEAD. 7

the occupation, coins from the ditches of the fort, found by Dr Curie, attest occupation unti t leasa l whicr tfo A.D, h 86 .yea dupondiio r thertw e ear 1 minn i t condition. Tha occupatioe th t t lasno t d mucndi he b longe o t s i r inferred fro face mth t tha secone tth d Flavian occupatio itsels somnf wa fo e duration and cannot far have outlasted the turn of the century (see p. 26). n thaI t event ther s hardlei y roo prolonmo t e Agricolagth n occupation beyond A.D. 90, if indeed for so long. It will be recalled that the one and only period of occupation at Fendoch 2 came to an end about this time.

LatePeriodThe Domitianic II, Occupation. e AgricolaTh n defence d buildingan s s were deliberately demolishen di orde stilr makfo o lrt y morewa e massive defence lesr fo s impermanend san t buildings, with stone foundations alreads A e firs. th y t, notestag3) . de(p in the demolition of the Agricolan white clay rampart was marked by digging awa clas fronte yit obtaineo yTh s oute.e uses th fildo p rwa dt u l Agricolan ditch, into which it was pounded so tightly that the filling could be distinguished fro subsoie mth lsoftes onlit stickied y yan rb r naturet I . is thus easy to see how the ditch thus sealed came to be missed on this side of the fort. Below the white clay filling, however, the bottom of the ditch contained about a foot of accumulated growth, covering in turn some nine inche purf so V-shapee apeee th silt th f xt o ,a d bottom furtheo N . r demoli- white tioth f neo clay rampart, however, seem havo st e takene placeth d an , inner Agricolan ditch was filled with a markedly different pinkish-yellow clay, presumably freshly obtained from another source. This clay was deposite evidenn di t lump r spadefulso s , which fille whole dth e ditch right down to the square channel at its base. The channel itselfilles wa fd with yello evidens i wt i siltt tbu , that there was no other accumulation in the ditch at the moment of filling. There was thu smarkea d differenc ditche o time conditioe tw th f th e o t e n sa i e th f no filling outee th , r being choked with inne e growtth rd sild cleahtan an d nan almost empty. This may be partly due to the fact that the inner ditch is here rock-cu would an t d neither make much r favousilno t r much growth. t musi t t Bu als recallee ob d tha innee th t r ditc mucs more hwa - hth eim portant of the two, and that its tidier state may equally well be explained by deliberate routine scouring, otherwise there could hardly fail to have been some white clay wash fro rampare mth e tth jusf i tt behinBu . dit ditch had only recently been scoured, it will follow that the filling took plac estils whilcommissionn i lwa t ei othen I . r words reconstructioe th , n whic filline hth g represents looks lik remodelline eth fora commissiof n gi to n as distinct from the reoccupation of a deserted or evacuated site. As the pinkish clay filling nears the top of the inner ditch, it is indis- tinguishably merge drampart-fron w witne ha beatef o t n reddish-pink clay, 1 Newstead, 405-406, coins nos. 163, 170. * PSAS., Ixxiii. 146, 153. 8 PROCEEDING E SOCIETYTH F O S , 1949-50.

which extends backwards over the dirty trampled surface of the shelf cut e earlieith n r white clay rampar d forwardan t s right acros e ditchth s . Beyonditce th t resthf i outee o dth s p li rupo nstria stonf po e bottoming fee3 t vern widei t y se ,har d blue clay (see . figThi1) . s bottomind gha evidently been laid before the inner ditch was filled, since a tongue of the associated blue clay extends wel over s ditci e l o'veth d sout-f e han o rth p hli lapped by the pinkish clay filling; and the function of the bottoming is to underpin the front of the pink clay rampart in exactly the same way as the earlier bottoming supporte e whit e th fron th df e o t clay ramparte Th . finished arrangement is, however, almost eclipsed by the Antonine fort-wall, inchepine w th fe k f a so cla t t awaybu whic cu frontl ys al describehs ha a , d in further detail below (p. 13). Beyond the front, immediately in front of the bottoming, a wood fire had been made by a working party, but its embers were quickly covered by an ever-increasing spread of washed silt from the pink clay. The spread extends for 13-| feet and tails out to nothing ove edge greaa rth f eo t ditc t leasha fee6 1 tt wid 1\d e an fee t deep. This imposing ditc manifestls hi y later tha outee nth r Agricolan ditch whicf o , h it has removed the south lip: but it is equally plainly contemporary with pine th k clay rampart, from whic washee hth d deposit spreads ove lips rit . In fact, the construction of the pink clay rampart, which is an extension or enlargement of the Agricolan white clay rampart, demanded a new ditch- system, since it obliterated the old one. The scale of the new works is in each case much bigger than that of the Agricolan defences, massive though thesramparw e ne som s aree fee5 wa teTh 4 . t s thickbackit r dirtn i ,fo , y white clay revetted with large stones, occur fee6 1 s t behine th taie f dth o l Agricolan rampart, and the ditch is correspondingly large. The object of thickenine th onln gca y have bee gaio nt n suggestioe heightth d an , n con- proportionw ne e veyeth s thay i sde heighb th t t must have been almost doubled, being calculable now at some 28 feet as against some 15 feet before. desige gatewaye th th f n nI o s suc hheigha rampart-walf o t k would imply three-storeyed towers and a rampart-walk level with the second storey and coinciding with the top of a roofed chamber over the gate. The arrangement is in this respect comparable with the upper tier of Csesar's elaborate fortifica- tions agains Bellovacie th t , wher towere eth s were three-storeyee th d dan upper rampart-walk levea n o ,l wit e seconhth d storetowerse th s f yo wa , designed1 so that the defenders were "safer because of their height, and for that reason might throw their missiles more boldl further.d yan t A " Newstead, however, nothin knows gi n abou e structurth t e gateth r f so eo interval-towers of this period., and the question, which is of limited scope and great interest, may well seem, appropriate to future exploration, the more so becaus e souteth h gate f thio s s perio isolatee dar separatd dan e frol mal others. 1 E.G., viii. 9 . EXCAVATIONS AT THE ROMAN FORT OF NEWSTEAD. 9

e intervallum,Th behind these imposin defencew gne s markei s a y b d sprea dirtf do y white clay covere ovenn a y d.b Then follow intervallumn sa roa t leasda fee4 1 t t wide, whic t awanorts it cu t s hyi a h Curie'r edgD y eb s. diagonal trench * crossing the Antonine contubernium from N.W. to S.E. Both the true edge of this road and the wall of a building bounding it must, however, have occurred withi widte nth thif ho s trenchnorte th r h fo ;sid e of the trench cuts into the floor of a building covered with occupation-earth, which continues northwards for 8 feet. The floor is formed by the natural e subsoilclath f yo , from surfac p whicto bees e heha th n removed, together with any Agricolan remains which it may have carried. A reason for the drastic levelling is provided by the north wall of the building, which is represente singla y db e layer inche4 , s thic 2d jk an fee t wide smalf o , l blue cobbles set on end and packed in red clay. This is plainly the foundation for a stone wall, and the fact that it overlaps the heel of an Agricolan sleeper- trench (Plat) show2 , n strikinei V s g fashion thae woodeth t n buildings had now been replaced in stone. Further, the planning of the new buildings is different northe Th stone . walth f eo l building just described narroa , w buildin t morgno e than 12 feet wide overall, overlap soute sth he walth f o l Agricolan building, whic largels hi y covere cobblea y db d roadwa fee3 y2 t wide. This roadway does not cut into the subsoil, as does the stone building, but rides on top of the Agricolan level, preserving and sealing it. Its north edge has, however, been removed, once more by one of Dr Curie's trenches, whic followins hwa soute gth h wal stona f l o (Plate) 2 buildin , eII fee7 g5 t wide and 224 feet long, discovered and planned in 1908. This foundation

is 3^ feet wide and 6 inches deep2 , composed of two rows of cobbles and broken trachyte, again set in red clay. Like the stone foundation already described t welse ls i int t subsoile i ,o th . Similarly surfacp e to th e f o e,th subsoil, together with any Agricolan floor or street that it may have carried, bees t awaha ncu ordeyn i furniso rt buildine hth g wit hnaturaa l clay floor, in which, as before, only the heel of an Agricolan sleeper-trench is left. The floor of natural clay thus formed is covered with a thick layer of occupation-earth, among which was much wattle-and-daub, large lumps of which also bestrewe road-surface dth e buildingsoute th f ho . This material clearly indicates tha stonewore tth k carrie dtimber-framea d superstructure largely buil wattle-and-daubf o t , whic turn hi n explain sucy hswh cars ewa take securo nt elevea l stone foundatio settiny nb stonewore gth k deen po the upper side of the slope. The main sills for a framed building, as all erector f prefabricateo s d buildings know, mus e laib t d dead e leveth f i l superstructure is to be successfully erected at all. A further point here emerges Agricolae Th . n buildings also were built with timber-framind gan wattle-and-daub, but their foundations were wholly of timber, as is shown remainy b s already describe . Thus3) e . differencth (p d, e betweee th n Newstead,. 42 Ibid., 65-66. 1 a 10 PROCEEDINGS OF THE SOCIETY, 1949-50.

Agricolan buildings and those which followed them is not in their super- structur thein i t r ebu foundation reasoe Th sucr n.fo hchanga evidents ei . Wooden building dama n i pt ssubsoise l rapidl inevitabld yan y begi roto nt . LATE DOMITIANIC FORT, NEWSTEAD

500 100 FEET Pig. 3. t buildingBu s constructed wit hwoodea n framework raisey ddr higd han upon stone sills will last for generations if the woodwork is adequately weather-proofed achievee b d thin an ca s, d easil d cheaplyan limey b y - washin exteriore gth . The, chang thus ei structurasa l economy oned an , , moreover, which would quickly be called for in a soil so damp as that of EXCAVATION E ROMATH T NA S FOR F NEWSTEADTO 1 1 .

Newstead. Its object is to eliminate heavy and wholesale routine repairs, polica y jus indicativs a t e intention oa f settlo nt e e th sit e dows th ea n no creatio e massivth f no e rampar ditchd an t . Equally e heavilth , y burnt wattle-and-daub suggests that the buildings in due course perished by fire. The buildings discovered in this section do not stand alone. As is well known, they form, part (fig. 3) of a series of early stone buildings discovered recognisem hi y Curier b b maie y D par s d th d a f nan , o t central seriese Th . first, lying below the south granary of the Antonine fort and measuring 80 feet wide by 115 feet long, is itself pretty certainly a twin granary. The

second f whico , h only fragmentary remains were discovered s acutelwa , y 1 identifie e sacellumr Curith D s a westward-facindy a e b f 2o g principia. Ther nothins ei g surprising abou lese t th sthis f importanI . t buildingn si the retentura had stone foundations, principal buildings thus constructed may be expected as a matter of course, and these too may well have been half timbered. garrisoe Th f thino s for , howeveris teas o s define o yt t stone no ,Th e. building with which contact was made was 57 feet wide and must surely be recognised as a double barrack-block. Its clay floor, with abundant

occupation-material and pottery, will no3 t suit a stable. In this case, however, its length of 224 feet is quite abnormal for a barrack in an auxiliary fort; 4 it comes so much closer to the legionary barrack,6some 250 feet long, as to raise the question whether legionary soldiers were there in garrison. The ample evidence for their presence during the Antonine occupation (see allowee musb ) t 19 obscurno t. o d t p existence eth evidencf eo same th f eeo kind in relation to the first-century garrison. Legionary swords and a helmet came from pit LVII, below the Baths, in association with first- century pottery.6 On the other hand, this pit also produced 7 unmistakable auxiliaries' equipment. Thus, if legionaries were in garrison at this epoch, they wer t aloneconditioe eno Th armoure . th f no , moreover eloquens wa , t of disaster thid an s, correspond evidence th o st e pointin e firino gt th f o g buildings, in the form of burnt wattle-and-daub.

Period First III.The Antonine Occupation. The third stage of the occupation clearly follows a long break, marked by the notable accumulation of 3^ feet of black silt in the great ditch of Period II. It is evident that this silt accumulated gradually, since the sides 1 Twin granaries occur at South Shields, Ambleside, Hardknott, Brecon Gaer, Housesteads, and back to back, at Birrens. » Newstead, 52. 3 For dimensions cf. Pendoch, PS AS., Ixxiii. 135, barracks 8 and 9; Benwell, AA.*, xix. 25. r normaFo 4 l dimension . Fendochscf , PS AS., Ixxiii. 134-36; Housesteads, AA.1, xxv. 231; Gelly- gaer, J. Ward, Roman Fort of Gellygaer, 65. « Cf. Novsesium, 240 feet; Caerleon, 242 feet (Arch. Camb., 1931, 136). • Newstead, 129, pi. xxxiv(gladii).4 1 , ,13 nos, 11 . ' Ibid., 129 ,(visor-mask) pisx xx . , xxxiv (spaiJia).0 1 , 12 PROCEEDING E SOCIETYTH F O S , 1949-50.

of the waterlogged ditch caved in at two distinct levels during the process. Further sile associates th i ,t d wit branchee hth growinf so g hazels, witha heavy deposi vivianitf to ribhorsee pictura e f th s o Th d .e ean thus afforded waterloggea f iso d ditch gradually becoming choked with decaying rubbish, e coursdu an n edi overgrown with vegetation among which young hazel trees sprouted. Compariso mosy nma t readil made yb e wit ditchee hth f so the Agricolan fort at the Bar Hill, which were found by their excavators * o havt e been similarly choke d overgrowan d n whe e buildernth e th f o s Antonine Wal e sitel th t lai theifor.n w dou o t Facene r thesy db e conditions JRomae th , n engineer t Newsteadsa Hillr ,Ba muc , t begaa s ha n their wor sealiny kb choked gol ofe dfth ditches wit hlayea clayf ro , her feee5 t thic whitisd kan h yello colourwn i . Their next action was to dig two new ditches (see fig. 1), the inner 8 feet wide and 3-^ feet deep, the outer 6 feet wide and 3 feet deep. The profiles of these ditche e abnormalar s . They accord neither wite normath h l V-shaped ditch or fossa fastigata,2 nor with the fossa punica,3 with its vertical outer scar slopind pan g inner scarp eacn I thef .h o outee mth r t scarno s pi vertical t verye , y much steeper tha e innenth r scarp, anda ter f f i m, o classificatio wanteds ni , they could legitimatel categorisee yb semi-Punics da . Moreover, like the Punic ditches at Cawthorn,4 they are associated with a field of fire situated some distance away from the defences with which they e connectedar fiele question Th i d .fee 2 3 e inne t s nth i wide d r ditcan , h is 25 feet in front of the contemporary fort-wall. But the most remarkable featur thesf eo e ditche theis si r fillinglowee Th r. par composethif s to swa d of small coagulated lumps of very heavily percolated soil, neither silted nor compressed, resembling most closel kine yth heavilf do y leached lumpy soil which gathers under a privet hedge in a neglected garden. It must be taken in connection with the fact that the steep scarp of the outer ditch showed distinct trace wooa f o sd deposit suggestioe Th . therefory nma e be made that the function of these deep and narrow channels was to hold an obstacle 5 composed of entangled branches, the barbed-wire entanglement s dayosuggestioe it f Th . mads ni e wit e morhth e confidence whes i t ni recalled tha t Vetera t Claudio-Neroniae th a n legionary fortress exhibited

remains of just thi6 s kind of entanglement (Astverhau), though the ditches holding it had much narrower proportions. The supersession of the great ditch of Period II by the inner semi-Punic ditc evidens hi t fro face mth t tha clay-filline th t botd outee an h th f p go rli strictlgreae A th . t awatit y ditccu y comparablye b har e relationshis pi

Macdonald and Park, The Roman Forts on the Bar Hill (Glasgow, 1906), 14-15. De mun. castror., 49. Ibid. Arch. Journ., Ixxxix. 72-73. Livy, xxxiil. 5, 9; Polybius, xviii. 18. . LehnerH Romerlagers Da , Vetera Xanteni be (Bonn, 1926. 8 , Abb, 7 19 ) , 6 . EXCAVATIONS AT THE ROMAN FORT OF NEWSTEAD. 13

discernible between the rampart of Period II and the new fort-wall. The trench in which the wall-foundation is laid is dug not merely through the rampart itself t alsbu o, throug accumulatee hth d silt whic gathered hha d in front of it in a layer 2^ feet deep. It will be observed, in particular, that, if the pink clay rampart of Period II had been built against the back of an existing stone wall foundation-trence th , wale th lf ho mus t itself have been dug in the subsoil and could not have been dug in the rampart. Secondly, the clay of the rampart would have been packed tight against the stonework, wherea facn i s ti separatet si d darfroy b mkt i earth whic slippes hha d down e trench th e sid f th n shortI o e .e relationshi th , f wall-foundatiopo o t n rampart clearly indicates that the wall was erected only after the rampart falled ha n into such decay that gatheresild depte tha th f o ht o front n di i f o t thu2n demonstrate^e ca sfeetb t I . dfort-wall w tha erectioe ne tth e th ,f no like diggine semi-Punieth th f go c ditches, followe a lengthd y periof do neglec abandonmentr o t . Ther accordingln eca doubo n e ye b t th tha n o t rampart ditchese dealine th ,ar n jusi e s ,w ga t wit Antonine hth e occupation, which followed well over thirty years of neglect. The builders of the new wall plainly no longer appreciated what was subsoil and what was not, but worked from the surface of the collapsed rampart-front as they found it, digging a foundation trench 3 feet deep and 7 feet wide (Plate III, 1). foundatioe Th n whic trence hth h containe exceptionalls dwa y carefully formes wa laid t dI .partl locae th f ly o trachyte t mostlbu , largf yo e river- cobble vern i t y sfee se toug 6 s tfeeclayi d 3 wid t hI re td .deepe an . Nothing remained, however, of the masonry superstructure which the foundation had carried, except a scatter of broken sandstone blocks immediately below the plough, including one fragment of a chamfered plinth. It is, then, clear tha masonre th twal e th lf ypropeo oncd rha e begun immediately above eth surviving foundation, though all stonework in position had vanished. Allowin r offsetsgfo , this walhavn ca l e been little more tha feen5 t thick, thud an s matche masonre sth y wallforte Hadrian'th n sf o o x s Wall rather than those of the Antonine Wall. These walls were all designed with an

earth bank to rear, carrying a lower rampart-walk supplementin 2 g the parapet-walwale th lf itselfo p to . n ko And , althoug t Newsteaha d nothing of an earth bank recognisably contemporary with the fort-wall now remains, owinremovae th o gt everythinf o l g above foundation level occurrence th , e Antoninon fa e oven, inserted just beyon rampare dth t backin distanca t ga e of 39 feet to rear of the wall, is in itself good enough evidence for the original existence of such a bank. The bank itself was doubtless a refurbishing of ramparts I and II, for the new earth bank is not likely to have been as high

1 Benwell, 5 feet, AA', xix. 8; Halton, 5 feet 9 inches, A A.*, xiv. 155, 159; Birdoswald, 5 feet, CW.2, xxxii, 255, fig. 10; Balmuild 7s J yfeewa t thick. NS . e Millerse :Roman e Th , Fort Balmuildyt a (Glasgow, 1922), 7; Castlecary, 7i feet thick, PS AS., xxxvii. 289. This questio firss ntwa precisely discusse Britisr dfo h example . NS . Milley sb JRS.,n ri . 177-78xv . z 14 PROCEEDINGS OF THE SOCIETY, 1949-50.

as the estimated height for rampart II since it is related to stone defences. The height of the stone fort-wall will hardly have exceeded that of the York fortress-wall,1 of which the parapet-walk is 15^- feet high, while the rearward earth bank still exhibit wal s heigha sit feett 9 ka f o t. well-buil a ovee s Th nwa t structur f ston eo clayd ean , abou fee5 tn i t diameter whicf o , ash-coveree hth d floor remained saucera front n i I f . o t- like depression serve s stoke-holda r ash-piteo . Then followe a vacand t strip of ground, 7 feet wide, providing elbow-room near the centre of cooking. The strip was delimited by a road 12 feet wide, made of broken red sandstone bottoming carryin g3-inca h laye f gravelo r . intervallume Thith s si road recorded 2 by Dr Curie, and it borders his barrack-block I, of which the walls were recovere drecorded m preciselhi y b ,s ya togethe r wit e diagonahth l trench which it was his custom3 to cut across each division within the block, this being the fourth room or contubernium from the west. The walls are carried upon a double layer of river-cobbles, very carefully laid 4 in thick red clay, and are themselves built of well-dressed sandstone blocks, 5 inches high, from, which the masons' fragments provided the road-bottoming mentioned above soute Th h. wall fee2 , inche8 t s wide buils i , ) t1 (PI, IV . directl e intervallumth n yo offsetso n s doubo n ha ,roa f Period tdo an I dI becaus fire th m f eo surfac e upon whic t hrestsnorte i Th h. wal) l2 (PI, IV . is laid directly upon the wall-foundation of Period II, from which the stonework must have been systematically stripped, and is furnished with a three-inch offse eacn o t h side (PI , 1)V . Both wall entirele sar clan i t yyse as opposed to mortar and presumably belong to half-timbered buildings.5 From this point northwards the plough and, perhaps, systematic clearing site agriculturr ofth e fo e 6 have left Antonine littlth f eo e structures. Only soute th f ho wale undebarracf le o l e th r , whickII itsels hha f been totally removed thers i , e preserve dsmala l fragmen roadwae th f o t y betweee nth barracks.

SecondThe Period AntonineIV. Occupation. Clear evidenc a secon r fo e d Antonine occupatio s furnisheni e th y b d ditch system. The two semi-Punic ditches were now deliberately filled with upcast fro mnewly-dua g serie thref so e ditches filline shos Th gwa .t in layers on top of a small accumulation of the coagulated earth which has already been described. growto Thern s siltingr ht eseemo wa i d an s, clear that no long gap separated the two occupations. The new ditches were

> S. N. Miller, JRS., xv. 177-78. " Newstead, 33. 3 Ibid., 42. 4 Mr S. N. Miller, who saw these remains exposed, considered them to be the finest work of the kind seed heha n in Scotland. t Chestera s A 6 e Journalse : Chester e oth f Northd an Wales Architectural Archceologicald an Society, N.S., xxxviii , pi5 ,. ii. 1. 8 For agricultural clearing about 1820 see Newstead, 5, quoting Archceologia Scotica, iv. 422. EXCAVATION E ROMATH T NA S FOR F NEWSTEADO T 5 1 .

all V-shaped fossae fastigatce. The outermost, 13 feet wide and 5 feet deep, contained half a sandstone quern and the rim of a late Antonine cooking-pot (p. 36, no. 17), and these objects lay in a bed of silt which half filled the ditch turn i s n coverewa d wasan y db h fro e upcasmth t moun southo dt d an , finally by material ploughed down from the mound at a much later date. thus i t sI clear that this ditch remained mor lesr eo s open until ploughing began, and so falls into place as the latest ditch in use on the site, which gradually silted up during long centuries of desuetude. Similar conditions appeamiddle th innermosd n i rean t ditches, which thus plainly belono gt the same series (see fig. 1). The middle ditch is 10^- feet wide and 3^- feet deep, and the innermost 13 feet wide and 3 feet deep, But, while the outermost and middle ditches were dug in normal boulder-clay subsoil, innermose finee th th rn i grad g t du ditce s clay-fillinhwa g sealin greae gth t ditc s profilf It Perioo h s therefor. i eII d e less shar mucd an p h more weathered than that of a ditch cut in natural subsoil, and it is the difference in colour, between the chocolate-toned filling and the surrounding mass of whitish-yellow clay, rather than a crisply defined profile, that makes the V-shaped form distinct. No trace of work contemporary with the three ditches survives on the fort-wall. But at the rear of the earth bank a mass of clean material coverin e demolisheth g de firs oveth tf o nAntonin e period representa s backward extension of the rampart. This shelters part of the new intervallum road, whicligha s thha gravel surfac inche3 e s thick resting upon inche9 r o heavf so 8 y cobbled bottoming survivine Th . g portion, 1\ feet wide, represents all that is left of an original width of 11 feet. It should be emphasised that both the new rampart-backing and the new intervallum t routinno roa e dear repair existino st g structure t taksbu e their places a 1 entirely new creations, which are as distinctive as the triple ditch system already described. Of the buildings of this period, however, nothing had been left by the plough. Dr Curie assumed that the barracks of Period III had been re-used, and this is the more likely now that proof has been obtained of their timber-and-daub superstructure (see p. 14). If such buildings were destroyed t wouli , ease b de burntlifo yt th tr tumbleo d framewor e stonth f e of k sille sillsth s ,e whiceitheus o ht rcarried an , dit unaltere framinheightenedr dw o nexe ne th f te go periodth r fo , . The Period of the " Reducing-wall." The section so far described may be regarded as having produced abundant and satisfactory evidence for two Flavian and two Antonine structural periodst couli t t shedno Bu d . light upo e relationshinth f po these four phases to the so-called reducing-wall in the retentura of the

That is to say, they blot out and extend over the structures of the previous period; they do not refurbis1 h them. NEWSTEAD 1947 THE ANTONINE DIVIDING-WALL E W

ANT.n WALL^ -ANTONINE H COBBLED SURFACING- FOUNDATION -ANTONINE I FLOOR & STREET—OF DIVIDING*- -ANTONINE I STREET AND CLAY FLOOR-- WALL

-LATE DOMITIANIC STREET AND CLAY FLOOR- ——————AGRICOLAN OCCUPATION-LAYER.———————

10 4o FEET

Pig. 4. EXCAVATION E ROMATH T NA S FOR F NEWSTEADO T 7 1 .

Antonine fort t wasI . , however existence th , thif eo s walld whicle d hha Dr Curi postulato et t fouet fiv 1no bu re period occupationf so whicn i , e hth Antonine fort, built to full size in a third period, was reduced by cutting off larga e portio retenturas it f n o fourt a n coursi e hdu periodn ei throwd an , n open agai fulno t l sizfifta finad n ehi an l phase seconA . d sectios n (figwa ) 4 . therefor t acros reducing-walle e cu linth e f sth eo poina fee 0 t a ,t 12 t north sout e feedesigneos 0 th f7 t wa h e fort longe b t wal I th o dordef .t n i ,lo o rt ascertain whether, as Dr Curie thought, there had been neither an earth bank behind the wall nor any ditch in front of it. The topmost layer in the trench comprised 3 feet of imported earth, strewn over the whole site in order to ensure growth after the digging of

1910. Below this layer cam2 econtinuoua s ban heavf do y pitching, formed of cobble broked san n building stone surfacp , to fro e f whicmeo th h many e walloth f s recorde r Curi D d bee y ha eb n d stripped e opeTh n. drain e wesclosth to et edg f buildino e g XXII was, however, stil positionn i l , together wit h wala t hithert no l o recorded pitchine Th s quit. - gwa e un disturbed except by the trench cut in 1908 for the purpose of tracing the reducing-wall. It yielded no pottery; but a chink in the east side of the open drain produced an aureusz of Trajan, with the rare reverse legend REX PARTHVS and the accompanying scene of proskynesis before Trajan and his army by the puppet-king Parthamasiris in A.D. 116. The com was in good preservation, though considerably rubbed before it was lost—a condition "very like that of the Corbridge gold hoard of about A.D. 160," as Mr Mattingly remarked upon seeing the piece. Thus, so far as it went, this coin tende confiro dt Antoninn ma layere datth t shoulI r .e fo d also be observed that the condition in which this surfacing was found, stripped l wallingoal f , help explaio st state n th affairf eo s observe r CuriD n y ei db e northerth nretentura,e th hal f o f wher foune eh d 4 nothin pitchea t gbu d spac assumed ean d that this represente open da n space storesa use r dfo - lighe dump presenth f o t n I . t informatio regardee b n pitchina ca s dt a ni g covering the whole area, upon which the latest buildings of the Antonine period rested until they were remove stone-robbery db agriculturalistsr so . The dating of this pitching to a second period of the Antonine occupation was amply confirme e potterth y b dy found belor embodieo , ws it it n di make-up f earlieo .l al r Thi s Antoninswa e dat earle e ff.) 4 Th (sey3 . . ep Antonine layer, too, was evidently contemporary with the so-called reducing- s notewa wallds A almos. t everywhere r CurieD y b 5, whose trencs hwa clearly defined, no superstructure of this wall remained. But there was a beautifully laid foundation (PI .f larg o Ill ) e2 , blue river-cobblen i t se s

1 Newstead,. 85 face tTh that this happene clearls dwa y recollecte Porteousr M y remembereo db wh , evene dth t as a1 youth. » BMC., iii. 106. « Newstead, 71. 5 Ibid., 34. 2 VOL. LXXXIV. 18 . PROCEEDINGS OF THE SOCIETY, 1949-50.

stiff red clay, 6^ feet wide and 3 feet deep, so very closely resembling the fort-wall discovere Section di nI tha t there th e s doubo couln wa e t di b t work of the same builders. This observation was the first clue to the early Antonine associatio reducing-walle th f no t exacbu ; t stratigraphical proof was shortly forthcoming. In laying the foundation, from which nothing had been removed except the superincumbent masonry, the soldiers had s outerit f r westerno o , p proofeto e , th dedg e wit ha broa d fille f stifo t f yellow clay. This fillet rested upo e cleanth n Antonine surfac f graveo e l and cobbles, and was in turn overlaid by the occupation-layer of ash and debris which had accumulated on that surface. It was thus evident that the finishing touches to the foundation followed the laying of a metalled surface against precedebu , it toccupatioe dth r whicn fo surface hth s wa e intended. In other words, the laying of the so-called reducing-wall was exactly contemporary wit layine hth g earldowe th f yno Antonine surface. Below this surface occurred a condition strikingly different from any yet encountered on the site. The surface itself was bedded in a thick mass of laithica df o clayk p layeto , peatf whicn ro o t yh sa mud . Below this again wer layero etw f occupationso , sometimes distinc sometimed an t s merging into one another. A considerable amount of timber and boarding was associated with the lowest layer. Neither layer, however, was productive of pottery botd an ,h were difficul examino t t e owin thico gt k blacd kmu a bris d kan flo f wateo w r which require da pump frequenf o e . us t Despite these hindrances, it was possible to recognise hi these two lower layers, so close to one another, the remains of the two pre-Antonine occupa- understano tionst d ,an d also wha produced ha t marsh-like dth e deposin i t which both were engulfed. "When the ramparts at the west end of the fort were laid down they encircled wha alreads wa t ynaturaa l hollow frod fe ,m appreciablsite d th ean f o ease d th en te upo groune nth d e to-dayth n I . Roman design drams were certainly provide deao dt l with accumulating water. But when the fort was deserted and the drains became choked the whole area would be converted into a waterlogged pool or marsh. The development of such a condition between the late Domitianic and Antonine occupation thus si s excellent abandonmene prooth f o f sitee goed th an ,f s o t with the gradual silting up of the great ditch of the late Domitianic fort. It is now clear that the Antonine occupation exactly contemporary with the so-called reducing-wall was the first Antonine phase, and that in this e soutth n ho e site rampartparth s a ,followef s furthe e o t wa on t y i ,r b d phase of occupation only. It is, then, no longer possible to view the reducing-wal a fourt s a l h phase, representin ga reductio e th sizn ni f o e Antonine fort. It is a feature which belongs to the Antonine fort from its inception. It is, moreover, clear that Dr Curie was right in thinking1 that bano n unbrokekn d a ditco behinan r n ther t frons n fo i hi , f ewa dit o n t 1 Newstead, 34. EXCAVATIONS AT THE ROMAN FORT OF NEWSTEAD. 19

occupied surface in fact runs up to either side of the feature. There is, on the other hand, no doubt that it is the wall of a fortification; its very thickness guarantees this conclusion wese plae th th f t no gatd an ,e included t confirmi n i pointe sth . Further closs it , e resemblanc soute th ho et walf lo the fort, as revealed in our Section I, implies that the two works were erected by the same builders. A striking difference, however, now emerges between these walls and the west wall of the Antonine fort. The south wall Antonine ofth e for carries i t d upo nfoundatioa feen6 t fee 3 wid td deepean , reducing-wale th d an wese similars th i l carrie s t fore walth wa Bu t f .o ld on a foundation1 8^ feet wide and only 9 inches deep. Dr Curie's photo- grap thif ho s wall2 also show reconstructiona s thawa t ti , sinc masonrs eit y plainly lies upo building-materia nd masa ol f so upot no nd systematicalllan y laid footings , then If wese .phases o th , ttw wald , ha linvolvin g complete reconstruction apparents i s ,a , this stat affairf eo s matcheAntonino tw e sth e occupations proved in Sections I and II, and the first phase should belong to the first Antonine occupation. In other words, during the first phase of the Antonine fort the full circuit existed as well as the reducing-wall, the only difference being tha e defence th te easter th f o sn two-thirde th d san reducing-wall wer estylee builon ,n i twhil defensive eth ewestere walth f lo n thirs buil anothern dwa i t westere Th . n third thusn a emerge s a t no s abandone a reducing-wall y b a separatf d s of areaa t t cu e,bu , occupied area with its own build of wall, additional to the eastern two-thirds yet forming a continuous whole with the main circuit. The so-called reducing- wall thus effectreductioa t divisiona no s t nbu faca , t which completely explains the absence of bank or ditch. It must henceforward be known as the "dividing-wall." Before attempting to define the purpose of the western enclave, it will be wel examino t l e maieth nenclosure parth f o t e (fig . Thi.5) s contains twelve barracks, capable of accommodating either the twenty-four turmce alaon fa milliaria twelve th r o e centurice cohorteso tw f o quingenarice brigaded together. Ther larga s ei e prindpia, granarieg bi commandant'a o d tw san s house; in addition, there is space3 for at least another commandant's house and other buildings. From the prindpia well4 came a variety of objects, seale deposia y db building-stonef o t altarn a d , san eithe r thrown there durin gdestructioa r deliberatelno y deposite attempn a n di tido t t y the site. The altar5 was dedicated by a legionary centurion, C. Arrius Domitianus of the Twentieth Legion, who was also responsible for extra- mural dedications 6 to Diana and Silvanus. This comprehensive series of- dedications, of which it may be assumed 7 we have now only a part, indicates, Newstead, . 33 2 Ibid., . 1 pi, vi . Newstead,e Se 3 , plan38 . Newstead, 48-49, 116-117. EE., ix. 1235 =Newstead, pi. xvi. GIL., vil. 1081 (Silvanus); EE., . 123ix 4 = Newstead, (Diana)3 14 . For a comprehensive series of dedications by a new commandant of. Maryport, CW.2 xxxix, 23. 20 PROCEEDING E SOCIETYTH F O S , 1949-50.

FIRST ANTONINE FORT AT NEWSTEAD BATH-HOUSE OF THE EQU1TES

EQVITES ALAE VOCONTIOR.VNV

COMB NED ADMINISTRATION ETC.

iiiiiiiiiiiiiiiiiiiiniiiimiip ^//illlllllUllllllUlllllllllllllM.

500 FEET Pig. 5. EXCAVATIONS AT THE ROMAN FORT OF NEWSTEAD. 21

as surely as the dedication to Jupiter Optimus Maximus, that Domitianus was commanding office t Newsteada r . Since, too e stateh , s unin hi s o t each occasio Twentiete th s na h othero Legion cannote d h , nan ,a likt eno few legionary centurions,1 have been placed in charge of an auxiliary cohort. muse H regardee b t havins da g commande ddetachmena unit n ow — s hi f o t that is to say, a vexillatio comprising two cohortes quingenarice—as Cheesman2 lon conjectureo gag Curies a d dan 3 also suggested. This explanation satisfactorily dispose eastere th f o s n twelve barracks. What purpose , thenth westere s th f wa ,eo n enclav Curir inclines D eewa ? d to think that it served partly for stabling and partly as a barrack-yard.

But his ston4 e buildings indicative of stabling were of the second Antonine phase, and it can now be stated with confidence that the gravelled surface which underlay them and which alone remained in the northern half of this part of the site belonged to this period also. It is further clear that the gravelled surfac preparatora s ewa harde yth bottomint - no site d eth an f go stana stores-dump r dfo e building Th e firs.th tf o sAntonin e phase li e deepe fac n weri d t t ran recovere eno d Curir eitheD ourselvesr y eo rb . Only the size of the area itself is available as evidence for what it may have contained structuras a r fa o s , l evidenc concerneds ei valuabla t Bu . e clue of another kind comes from outsid areae 178en th I alta.n 3 a foun s r wa 5 d som hundreo etw d yard e sfort th eas ,f o tdedicate e Campestresth o dt r o goddesses of the Parade Ground, by a decurion of the ala Augusta Vocon- tiorum. The style of the lettering and of the man's name, .ZElius Marcus, betoke secone nth d century musd an , t therefore refeunia garrisoo n t i rt n during the Antonine period. But the ala Vocontiorum6 is a quingenary ala, cannod an t have occupie largwhole e dth th f ee o are a throwne opeth n ni second Antonine phase, which is in fact suited to an ala milliaria. An ala quingenaria mucs aleeo i smallo tw hquingenariceto d an , muc largeo hto o t , fill the space. The ala quingenaria must therefore belong to the first Antonine phase, and experiment shows that if the appropriate accommoda- tion of four double barrack-blocks and four double stable-blocks is super- imposed upon the western area of the fort, it will exactly fit the area. This leave administratioe sth stored nan compann si y with legionariethose th f eo s in the eastern two-thirds of the fort, where there is ample room for them at range th principaf f enorte o o th d hen l buildings, beyon norte dth h granary buildinge (sitth e w xvi)Ho s . were planne t clearunite y no Th s ma di . possibly have share dheadquartersa ; indeed exceptionalle th , y large sizf eo buildine combinatioe th th d gan legionarf no auxiliard yan y armour found in the well suggests very strongly that they did so. But a commandant's

Cf. EE., vu. 1071, EB., ix. 1157 =ILS., 9151, where Dessau cites other examples. 1

Auxiliae Th 2 Romane oth f Imperial Army (Oxford, 1914) ; hereinafte27 , r referre Auxilia.s a o dt 3 Newstead, 74-75. • Ibid., 71-72. 5 OIL., vii. 1080= Newstead, 140-141. • Auxilia, 147. 22 PROCEEDING E SOCIETYTH F O S , 1949-50.

house would probably be required, and perhaps extra granaries, and the space available is nicely adequate to both needs. But despite the feasibility of combining or juxtaposing on the official side administrativeth e accommodatio r botnfo h units e marketh , d differ- ences between them must not be overlooked. As between a vexillatio legionis and an ala the distinctions were many and sometimes invidious. The commandant of an ala, already at least an equestrian by birth or Imperial favour, was senior1 in social and military standing to the legionary centurion, who became an equestrian only upon leaving the centurionate. e otheOnth r hand e ranketh , r legionaries, already Roman citizens, were superior in social status to the auxiliaries, who were normally peregrini only and Roman citizens only in prospect; the legionary also received more pay2 subjecs differenwa a o d t t an t discipline, with different degree f punishso - ment.3 Thus, while collaboration and understanding may be assumed at officee th r level t cannoi , takee b t granter nfo d amon lowee gth r ranks. The brigading of legionaries and auxiliary troopers together is thus in every way adequate to explain the arrangement by which they are contained within one fortification and yet divided from one another as regards the men's lines. But it also makes it possible to explain for the first time certain structural peculiarities. When Dr Curie faithfull methodicalld yan y traced the whole surviving length of the dividing-wall, he discovered 4 that to north and south it petered out just where the Antonine rampart-backing begins. So long as the structure is considered as a reducing-wall this dis- appearance of the feature was a stumbling-block, since it was impossible to think that the angles of a reduced fort should not have been contained by a continuous wall and most difficult to suppose that the foundations of such wala l should have been removed when they were preserved everywhere else. But a dividing-wall could very properly have been arranged to terminate against the rampart-backing by running up its slope;6 and there can be no doubt now that, although the precise arrangement is beyond recovery, this is substantially what happened, and the whole arrangement becomes in- telligible. A second fact which is now susceptible of a ready explanation difference th s i construction i e n betwee wale nth l surroundin enclave gth e assigned to the alares and the legionaries' wall. It is now evident that the external wall of the eastern two-thirds and the dividing-wall are exactly alike, because they do in fact together form the circuit of the legionary accordingly enclavma d ean expectee yb havo dt e been constructed together same th ed walan legionariesy e lb on s a . This will accoun vere th yr greatfo t technical ability discernible in the foundations of the walls in question, which

1 Auxilia, 93-94. » Ibid., 35. 3 Cf. Mommsen, BE., vii. p. 465. ' Newstead, 34. Curis A 6 e conjectured (loc. cit.), though without perceiving that this made nonsens cross-wale th f eo l except as a dividing-wall pure and simple. EXCAVATIONS AT THE ROMAN CAMP OF NEWSTEAD. 23

are at least a class above the normal walling of auxiliary forts, and for the comparable efficiency shown in the structure of the internal buildings. On •£he other hand, while nothing is known of the internal buildings of the first Antonine perioe enclav e alares,th th n d i f s surroundineo it go s wals ha l inferior a foundation that it cannot be assigned to the legionaries, and would be readily explicabl a differen s ea t preciselbu t y contemporary wall con- structe auxiliary db y troopers difference Th . statun i e s betweeo tw e nth unit agais i n reflecte choice th n grounddf i e o . Whil legionare eth y barracks sitee th upo liy ,f ee summie o thosdr easnhige th th d d f th een o ht an t a t auxiliaries occupy a swampy hollow, which demonstrably required much preparatory levelling to make it fit for habitation, if hi fact that object was ever quite satisfactorily achieved. These are among the more obvious explanations of hitherto unexplained anomalies within the fort at Newstead which follow automatically from the recognition of the first Antonine occupation as a joint establishment shared between legionarie auxiliariesd san t outsidBu for.e eth t ther furthea s ei r puzzl whicf eo hsolutioa readile b n nyca base interpretationdw upone e nth . The little bath-house l of the Antonine period, which lies outside the west auxiliariese th ford e gatth an t f eo ' enclave t onlno yreconstructioa s i , n no largea e sit f earlied th eo an r r building alss owa encloset i ; tima r edfo within a special rampart of its own, unlike any other bath-house in Britain

or elsewhere. This uniqu2 e arrangement can hardly be interpreted as a special defence for the bath-house, for such a conception has no place in the tactic forf so trampare defenceth t s reaBu ha tl . meanin ring-wala s ga l intended to isolate the bath-house and fuel stores reserved for the use of a particular unit. At this point the size of the building also becomes relevant. t largno es i enoug t I servo ht neede eth legionarief so auxiliaried san s com- bined, or even the joint needs of two legionary cohorts; but it would serve troopere th theid san r stable-lads specifis it d c,an reservation would prevent quarrellin promotd gan e concord betweebodieo tw f troopse o se nth Th . corollary of this conclusion is that a legionary bath-house still remains to be discovered elsewher sitee th .n eo

The Second Antonine Occupation. The remain e seconth f o sd Antonine perio s discloseda Section di nI have already bee e relatenb describedw o othet dno t then r Bu ca y. structural features. They revea a verl y complete reconstructioe th f o n ditch-syste e rampart-backingth f o d man , whil r Curie'D e s photograph 3 of the west wall shows a complete rebuilding of this feature from its very

foundations internae Th . l building e easterth f o sn two-third e sitth e f o s

2 Newstead,1 93, fig. 7. Ibid., 95, fig. 8; for description, 97-98. • Ibid., pi. vi, 1. 24 PROCEEDINGS. OF THE SOCIETY, 1949-50.

appea havo rt e bee somn i e degree usable headquartere Th . rebuilts swa 1 upon very much the same lines, though the colonnade of the forecourt was entirely renewed, while the sacellum received a strong-room below ground, fore-hala d s addedan wa l principalis.a astridvi e e solidleth Th y built granaries survived the reconstruction without fundamental alteration of plan.2 The same can be postulated of the twelve barrack-blocks, whose well-laid foundations would emerge little scathed on clearing the wreck of e dauth b walls thaburied ha t d themwestere Th . n portioe fortth ,f no however, was no longer separated by a dividing-wall. This was now removed buildingd an , s rightly identified r CuriD stables y ea b 3 s were laid t acros ou beyond san . dit Thes e buildings, however hardln ca , y have been confine e southerth o t d ne retentura, th hal r f Curio fD s ea thoughte Th . cobbled area in its northern half can be reinterpreted, in the light of Section II, as the surface upon which buildings were erected and from which they have since been ruthlessly robbed. Further, a full complement of stabling is needed to fit the use of the twelve barracks in the prcetentura, since, in relation to cavalry, these can only have housed the twenty-four turmce of alan a milliaria. equivalenn A t twelve block stablinf so thue gar s required, and this would call for all the space available in the retentura. Incidentally, it is of interest to observe that the addition of a fore-hall to the headquarters building synchronises t Haltona s a , , wit postine hth cavalrf go forte th ,o yt and thereby increases the likelihood of the identification of this building as a basilica equestris exercitatoria, to borrow the term used upon the Netherby stone 4 of Severus Alexander. The t identit reveale inscriptiony e unino th an s i tf y o db t alee Bu . milliarice were rare in the British province. The best known, and perhaps the most likely candidate for the position, is the ala Petriana, stationed at 5 Corbridge in Flavian times and at Stanwix under Hadrian and his successors. The choice of a regiment of such a site for the position marks out later Antonine Newstea keypoina s da networe th Lowlandse n th i t f ko t alsI .o imparts new interest to the earlier Antonine arrangement, by which both an vexillatioa d alaan legionarief no s were brigaded together legionariee th , s being detache purpose th r dfo e fro legatuse mth legionis vicesimce t Chestera . It seems to single out Antonine Newstead as a virtual advanced headquarters from which this part of the frontier was run. It also explains why the mileage upon the Ingliston milestone was recorded as measured from

Trimontium. If Newstead was an advance6 d headquarters the road-system would naturally focus upon it.

1 Newstead, , fig; als2 .43 o 44-52. 2 Ibid.,. 59 3 Ibid., 70. 4 OIL., vii. 965. Be . Se Birley5 , CWf, xxxix descriptioa . r 219fo , thif no s regimen histors it d Britainn yan ti . 6 OIL., vii. 1085. EXCAVATION E ROMATH T NA S FOR F NEWSTEADO T 5 2 .

CONCLUSIONS. The picture of Antonine Newstead thus emerges tolerably clear. It is a place-forte of great strength, containing a garrison of exceptional size, big enough to act as a miniature field-army, whatever other routine calls may have been mad es garriso t It firsupoa s mixea . t nit nwa d collectiof no legionaries and auxiliary troops brigaded together, the former representing approximately two-third totae th f l o sstrength . Later numbere th , s were cut down by one-third, and the legionary arm was eliminated in favour of more mobil onle th yet sigcavalry no parinf ns o i t I g. down evident upon Antonine th e frontier musd comparedan e , b t perhapd an , s connected, with the reduction in size of the Antonine fort at Ardoch. It is also consonant

with the growing tendency to exalt the cavalry arm,1 which was to reach its climalatee th rn x i thir d century. buildinge Th f botso h Antonine phases were massive therd an , goos ei d reason to think that the principal buildings were throughout of stone. It is e otheclearth n r o ,hand , thae stonth t e foundation e barrackth f o sr o s stables did not carry stone superstructures but -were the sill-walls for half- timbered buildings. This is an important point, because it seems to make it certain that many barracks hitherto described as of stone and assumed to be entirely stone-built were in fact less permanent structures, thoug thoughe b t hneei t dno t that they wer facn ei t less comfortablee Th . distinctio n qualiti n y observable betwee e legionarth n d auxiliaran y y buildin s alsgi f speciao l interest s rarelseee a , b juxtapositionn i no y t . The appearance of the fore-hall in direct connection with a change to a wholly mounted garrison is also significant for the interpretation of such buildings. Lastly evidence th , e fro ditchee mth s suggestiv shora p f eo ga t only between the first and second Antonine phases is of high importance; while the fact that these amount to two phases only and not three, as on the Antonine Wall alss ,i ascertaine n oa d fac greaf o t t value, evehistoricas it f ni l significance is not as yet to be assessed. The second Flavian for t Newsteaa t undoubtedls dwa e modeyth r fo l the first Antonine arrangement, though the ratio between legionary and auxiliary troops mixeit n i sd garrison canno e stateb t d wit e samth h e confidence s barracIt . k buildings appea havo t r e bee half-timberf no d an , this syste mconstructiof o wely nlma have bee nprincipae useth r dfo l buildings also. Its defences appear to have been exceptionally massive and high emphasisd an , importance eth e attache Britise th y dhb commano dt the possession of this advanced base. The most valuable result of the present excavations, however, is the demonstration that the second Flavian

1 O. G. S. Crawford, Roman Scotland, 37, fig. 9, the "early fort" being the first Antonine fort. A comparable reduction is also seen at Cappuck, where the rampart containing the gate is plainly a reduction (PSAS., xlvi, 450). 26 PROCEEDING E SOCIETYTH F O S , 1949-50. occupation, beginning soon facn aftefinae i th s t lr wa phase A.D, th 86 f . eo earlier stage in the history of the site. A third phase is now eliminated, and with it vanishes the best ground for a late Trajanic occupation. On this poin Samiae tth n ware recovere thin do s occasion offers little helpt bu , amon coarse gth e pottery, fragmen furnishe4 . No t s some suggestive evidence (see p. 34). It is difficult indeed to suppose that a piece of this kind could have lain undisturbe a roa r intn e secondfa o do th d century, while th e previously noted scarcity of Trajanic Samian ware from the site as a whole must carry weight. It would be rash to carry the occupation much beyond 1 A.D. 100 on existing evidence. The Agricolan fors alsha ot produce featuresw ddouble ne Th . e ditch can now be deemed to have run round the entire circuit. The rampart has been define massiva s da e wor permanenf ko alss i t ot i type clead an ,r that it enclosed timber buildings founded in sleeper-trenches and that stables were included among them. Mounted troops were thus present in this earliest phase. These discoveries eliminat e oldeth e r idea thae firsth t t fortification at Newstead was a semi-permanent bivouac, whose garrison lived in tents. The fragments of leather tents or the wooden tent-pegs ditchee founth n di s mus occupante tth o t belon o forte t t th t bu ,f gso no the working-parties who reconstructed it and deposited their rubbish in the ditches which they were obliterating. It is thus clear that the Agricolan designes forwa t d upon permanent lines alsfros i firste ot mI th clea . r that its massive rampart of beaten clay, once constructed, set the line to be followe l subsequenal n i d t reconstruction e defencesth rampare y b sTh . t might be enlarged or remodelled, but no Roman engineer ever faced the task of levelling it so as to plan the site on completely new lines. In this respect the fort contrasts sharply with such sites 2 as Tassiesholm or Loudon Hill, where a less uniform and lighter subsoil and ramparts of turfwork were an invitation to rebuild and remodel drastically. The source of the Newstead clay, whic s obviouslhi y local s probabli , ya ver y large bowl- shaped excavatio north-easo nt sitee th ,f o tfro m which many hundredf so ton claf so y lon e havth gn ei distan t past been removed. It remains to thank those who so kindly helped with the requirements worke foth r . Permissio excavato nt readils ewa y grante ownee th f y do rb the site, Mr David Colville of Chapel-on-Leader, and his tenant, Mr Thomas Porteous of Leaderfoot Farm, afforded us every facility for our work, includin ggeneroua s suppl fencinf yo g material excavatioe cose th f Th to . n was e Christianburbornth y eb y Trust. Tent equipmenr sfo shelted an t r were generously lent by the 2nd Galashiels Troop of Boy Scouts through

1 JRS., xxv. 70-80. This differenc mucs i e h more importan histore th r sitef fo yto s tha s oftenha n been realisedt I . may2 furthe addee b r d tha preciselr fo t y this reaso Newsteae nth d phase reflecw sno t themselves more noticeabl ditch-systee th n yi m rampart e thath n i . EXCAVATION E ROMATH T NSA FOR NEWSTEADF TO 7 2 .

their Scoutmaster . J Gaw r d surveyinM ,an , . G g . equipmenR r M y b t Bruce, the Galashiels Burgh Surveyor. Local arrangements were also much facilitate e kindnes Gordons th y Mr db f o s , the Newsteae nth d post- mistress.

THE FIGURED SAMIAN WABB. By ERIC BIRLEY, F.S.A. e smalTh l grou f figurepo d Samian ware includes some piece f moro s e than usual interest t merelno , evidencs y a phasee th r occupatiof esfo o t na

Newstead, but for the study of this category of material generally. In my study of the Lezoux pieces I have had the advantage of access to the unpublished Stanfield. paper late A . th e J f so , whose untimely deat bees hha n suc hheava y blo Britiswo t h archaeology pleasur a dut a d s i y an t i o et d an , record the kindness of Mrs Stanfield and Mr Michael Stanfield in placing instanceo tw r advantagdisposale o y the th e sI havm d mt on a e ha n i ;f eo Miss Grace Simpson's hel settlinn pi g doubtful pointsindebtem a I . o dt Mr Geoff Hardy for the drawings, here reproduced as fig. 6. All the bowls 28 PROCEEDING E SOCIETYTH F O S , 1949-50.

Dragendorff'f o e ar s form figure-typese th . r 37fo ; followe. O , numbea y db r r referOswald'D o st s Index of Figure-types Terran o Sigillata.1 1. Flavian I. Fragment from a bowl by the "Potter of the Large Rosette," first studie Professoy db r Atkinso s papea hoarhi n n f o ni rd o Samian ware from particula n Pompeii d an , . 51f. i r no —7 (JRS.,pi 2 , .x . iv completa e bowl, probably fro same mth e moulNewsteae th s da d fragment). The Pompeii hoard, as he pointed out, must be regarded as a consignment received there shortly befor destructioe eth greae citth e y tth y b eruptiof no n of Mount Vesuviu A.Dn si . 79; that give unusualln sa y close datine th r gfo manufactur presene th f eo t piece, whic wely hma l have reached Britain ni e sam s th obviousli et i year t bu y: impossibl lon w telo t ega perio ho l d elapsed before it was broken and thrown away. The potter is probably identifiee b o t d with PAVLLV a GraufesenqueL f So , whose ovolo seems e identicab o t l wit e presenhth e (cf.on t Roger, Augsburg, pi. xxxia , 6 , piece from Giinzburg). 2. Flavian II. Part only drawn. Portion of a rather small, neatly modelled bowl: the ovolo (its egg bordered by two fine lines, the tongue bent somewhat to left), wavy lines and S-ornament in series, recur on a bowl same oth f ePompeie forth mn i i hoard (JRS., , piiv . xiv . 74)whicn o no ,, h cursive th e signature MIIMORIS retrograde appears belo decoratione wth ; the present bowl may therefore be assigned with confidence to MEMOR, a a potteGraufesenquL f o r e who r OswalmD d (Potters' Stamps, . 201p ) assigns to the period Claudius-Vespasian. .The only examples of his stamps fro e nortmth f Britainho , com27 . e, froDr bot n mho Carlisl d Yorean k respectively; both pieces, like that from Newstead, were presumably imported during the decade A.D. 70—80, and the Pompeian parallel shows that the Newstead bowl is likely to have been imported a year or two before the latter date; its appearance in the later Flavian level can throw no light latter'e onth s termina initiar o l l dates showingy , botan n h(o ) later thas nit manufacture. 3. Antonine I. Worn scrap from a South Gaulish bowl. Below the ovolo d coarsean wavy e seeb ne y characteristiline th par ma f o t c anchor motif (first so interpreted by Roger, Augsburg, 1914, p. 100). This is plainly straa y fro Flaviae mth n occupatio sitee th .f no 4. Antonine I. Two pieces from a bowl of unusual interest. The a copygladiator s i . ,r mucho t , reduce e Sout sizen th di f ho , Gaulish type, O. 1020, which was used most frequently by BIRAGILLVS and M. CRESTIO, and the dog to I. is a reduced copy of another South Gaulish type, O. 2008; the reversed S-ornament, too, reminds one of South Gaulish

1 Other abbreviations employed are as follows: Oswald, Potters' Stamps—F. Oswald, Index of Potters' Stamps on Terra Sigillata (1931); Roger, Augsburg = O., Roger's studies of Augsburg material, Zeitschrift des Historischen Vereins fur Schwaben und Neuburg, 1913-15; JUS. = Journal of Roman Studies. EXCAVATIONS AT THE ROMAN FORT OF NEWSTEAD. 29

influence does a ,e straight sth wreath betwee e mai th e ovol nth d fieloan d e decorationoth f e ovolo th e dividin t th itsel bu d ; an fg row f squarisho s beads have no parallels on South Gaulish ware. The present bowl seems rather to belong to a class which as yet is only sparsely represented in Britain comparativels i t bu , y commo e southerRaetin ni th d aan n parf o t Upper Germany, where the pottery which made it is perhaps to be sought (though it is customary to assign it to the fabric of Banassac in Southern Gaul) I hav. e noticed unpublished example t Augssa Switzerlann i t d dan Bregenz in Austria, and Roger has figured several in his reports on figured Samian from Augsburg in Bavaria.1 On vessels of this class, the figure-types are for the most part derived by mechanical reproduction (involving reduction in size and some loss of sharpness in detail) from South Gaulish originals, t sombu e types have been borrowed from Lezoux, ovoloe th othe d d san 2an r decorative details, usually rather crudelocaf o e l ar ,design yets A , . dating evidence (other than that suggeste typologyy db virtualls i ) y lackingt bu ; exampln a same th f eeo class, though attributabl differena o et t potters ha , recentl Antoninyn a turnen i p du eI deposi t Corbridgea t Newsteae th ; d bowl has therefore probably been found in its correct level, for all that its figure-types hark Flaviae bacth o kt n period. 5. Antonine . I Ovolo, bead-row below portiond an , larga f so e d leafan bird I.o t looking f thos. 2306)o (O e e ovol e . th on backr ; use s o oi t y b d ATTIANV f LezouxSo bowln o , s carryin s stamghi p retrogradeOFT .AT , which may be assigned to the later phases of this potter's activity. Oswald's dating of ATTIANVS (Potters' Stamps, p. 29) to the period Trajan-Hadrian does not accord with the distribution of his products; at Corbridge, for example, they have come uniformly from Antonine deposits presene Th . t piece seems likel havo yt e been made neare tha0 16 rn 140. 6. Antonine I. Portion of the rim, ovolo and upper part of the decorated e "Potte zonbowa th e Smal f th y eo b f l o lr S," perhap mose th s t easily recognisable anonymous potter of the Antonine period. Note the ovolo wit s knohit b attache e eggth , o dt lackin independenn ga t tonguee th ; figure-type, obscure carelesy db s remova bowe th f lo lfro e mouldmth a s i , variant of Apollo to r., O. 83—a favourite motif with this Antonine potter and many of his contemporaries. For other instances of his work found on Scottish site . Balmuildy,cf s nos. d 9143-45an ; 1 Mumrills,5 , 4 5 , nos48 . and 56; Newstead, pp. 221, 6 and 223, 2; Curie's Inventory (PSAS., Ixvi.) (fro1 1 figs , m(fro1 2 . 1 Ardifuar , m 41 Traprain) d an ) I hav. e also noted examples from Camelon (National Museu Antiquitiesf mo frod )an m Carzield (Dumfries Burgh Museum). 7. Antonine I. Rim fragment, showing ovolo and bead-row as used by

Cf. Augsburg, pis. vi, 1, vii, 1, viii, 4 (with an ovolo and bead-row not unlike those on the Newstead 1 piece), xvi xxiii, ,2 . 1 , 2 E.g. 0.1317 and 0.1491. 30 PROCEEDING E SOCIETYTH F O S , 1949-50.

pottee th r DRVSV Lezouf So bowla n xo , carryin signaturs ghi e beloe wth decoration, in the Lancaster Museum; I owe the identification to Miss Grace Simpson. Belo bead-roe wth horsee wpars th i f to galloping L, looking . 1910)(O . backr , o whict uses severay hwa db l Lezoux potters numbeA . r of pieces attributable to DRVSVS have been noted amongst Antonine I material at Corbridge; the dating suggested by Oswald (Potters' Stamps, p. 112, "Trajan-Hadrian") must therefore be revised, as in the case of No. 5 above and many other potters. 8. Antonine I. Portion of a large bowl, with elaborate panel decoration i nstyla e characteristi earliee th f co r wor CINNAMVSf ko y whoo t , ma mt i be assigned without question. The Vulcan to 1. (O. 66), bear to I. (O. 1666) . 1781stagd 1.(O o an t l occu )al signen o r d bowl thay sb e t th potter o d s a , various decorative details. To judge by the occurrence of his products in both Antonine levels at Corbridge, the working period of CINNAMVS may be assessed at 150—190; he was by far the most prolific of the Antonine potter Lezouxf so signed unsignean ,d dan dhav m bowlhi e y beesb n noted at Balmuildy, Mumrills, Newstead, and many other sites in Scotland. 9. Antonine . I Fragmen t showin ga portio ne decorativ th onl f yo e scheme, a continuous winding scroll pattern. The trifid ornament, the outer arms of which end in two detached beads, while its central leaf is curvilinea t angularno d s matchean ri , a Silcheste n do r piece (May, Sil~ Chester Pottery, pi. xxvii . e sam69)th no . e,d whicgeneraha s hha l pattern evey nma havd an e come fro same mth e mould, thoug t fro samhe no m th e part of it as the Newstead fragment; it, too, shows a neat horizontal bead- similaa rof o w e abover on serie d plaif an ,so n rings, while belo decorae wth - tio t ncarriei signature sth e (written free-han moulde th thereford n i dan , e reproduced retrograde on the bowl) CR of the Lezoux potter CRICIRO.1 We may therefore be confident in assigning the present vessel to that potter; but it may be noted that there is another Lezoux potter, MAKTIALIS, who uses the same trifid ornament, rings in series and bead-row bordering them (Knorr, Rottweil 1907, pi. 28, 6), and if it had not been for the close parallel to the design provided by the Silchester piece, it might have been necessary leavo t e ope questioe nth n whether CRICIR r MARTIALIOo madd Sha e the Newstead one. CRICIRO, too, is dated "Trajan-Hadrian" by Dr Oswald (Potters' Stamps, pp. 96, 378), but his products turn up regularly in Antonine deposits examplr fo , t Mumrillsea . 504)p d (no.an 0 .2 10. Antonine II. Portion of a bowl, showing the lower part only of e decorationth , which consiste a continuou f o d s straight wreath formed repetitioy b serien i trifid a f so leaf, terminatin designe gth ; abov t ecomei s a series of medallions, each to contain a figure-type. All three elements leae oth ff hav esimilaa r raised borde depressed an r d centre, thouge hth

same 1Th e trifid ornament appears, together with plain ring serien si s bordere neay db t bead-rows, on a bowl with panel decoration at Mumrills (no. 16); that bowl, too, is best attributed to CRICIRO. EXCAVATION E ROMATH T NA S FOR F NEWSTEADO T 1 3 .

bowl comes fro ma wor n mouldn mosi d t an ,case e detai th ss bee ha l n obscured t remindi ; t firsa e t sglancon thaf eo t use somy db e pottere th f so period Hadrian-Pius (e.g. QVINTILIANV s associates)hi d an S , which, however, has a. notched or corded central leaf. The figure-type in the central surviving medallion may be O. 417 or O. 667, but cannot be identified with certainty; only a fragment remains of that in the medallion on the left, and none of that on the right has survived. The bowl is unquestionably Lezou xs genera it ware d an ,l styl decoratiof eo n suggest relativelsa y late date, in the second half of the second century; it is to be regretted that piece on f figurethis o ee th , d Samian fro e seconmth d Antonine levet a l Newstead, cannot be assigned to a specific potter, and that it is at present impossibl produco et eclosa e paralle decorativs it o lt e scheme.

THE COARSE . GILLAMPOTTERYP . J y B , .M.A . (a) Introduction. Before discussin coarse gth e pottery recovere 194n di t 7wili wel e b l o t l consider briefly in what way it may throw light on the history of Newstead. alreads i t I y known withou hels it t p that there wero Flaviao tw e tw d nan Antonine forts thad an ,t there wa lonsa g intervae th f lo betweed en e nth second Flavian occupatio beginnine th firse d th n an tf go Antonine e Th . date when each of the two main occupations began can be closely inferred. The first Antonine fort was divided by a wall which was demolished when secone th d Antonine built s forwa t . Coarse potter hely determino pyt ma e the dates of the end of the second Flavian occupation, of the dividing-wall, of the beginning of the second Antonine occupation, and of the final abandonment. Pottery vessel easile sar y broken t seemi d san , reasonabl reckoo et e nth average life of a second-century vessel as less than ten years; authentic instances of survival for a much longer period are rare. The odds are always agains singly an t e vessel bein gsurvivala . Strays fro earlien ma r occupatio e oftenar n foun laten i d r deposits t vesselbu , s representey b d unworn pieces, or by a number of associated fragments, are unlikely to be strays.1 About two hundred separate vessels are represented by the fragments discovered; but the seventeen that have been drawn include all the best- preserved pieces. In a few instances a vessel is well preserved, closely stratified, and datable by the use of parallels, as, for examples, nos. 4, 5 and 14: s usefula thes e s ar emina ' t coins descriptioA . d discussionan f o n individual pieces follows (Fig. .7)

fulle1A r treatmen questioe th f o t f survivano l wil Birrene foune th b l n di a report, PSAS., Ixxii (1937-38) 318-20. pp , . 32 PROCEEDING E SOCIETYTH F O S , 1949-50. (b) Flavian Vessels. secone 1.th f Founsouto e d th d Flavia t hden a n barrack-building. Small delicat t witepo h narrow base sofn i , t orange-pink fabric. Ther evidenco n s ei r efo the date of the type; the fabric is common in Flavian deposits. 2. clae Foundth ysecone n flooo th , f ,3 o rwit d . FlaviahNo n barrack-building. Rim fragment, doubtless from a carinated bowl, in sandy orange-buff fabric; there is no reeding on the rim.

1 10

16 18

Pig. 7.

EXCAVATION E ROMATH T NA S FOR F NEWSTEADO T 3 3 .

Cf. Chesterholm, no. 14, A.D. 79-125. presene th n I tknowledgr statou f eo type eth e canno datee b t d more closely than, from time th Agricolf eo a unti earle lth y year Hadrianf so . 3. Found, with No. 2, on the clay floor of the second Flavian barrack-building. Fragment of a platter with incurved sides, in pinkish-buff fabric; the base has been supplied in the drawing. Cf. Cartridge, , A.D191119 . . 79-125no , d Ohesterholm,an , , A.D18 . . 79-125no . This type does not appear on Hadrian's Wall, and so had presumably gone out of use before A.D. 122.1 4. Found lying secone hith blacf o d p kFlavia to silt n o , n road barracke soutth f ho - building. greatee Th r parmortariua f o t m broken into several pieces. Mor thif eo s vessel survives than of any other from the present excavation. It has a flat rim, thickened on the underside wher t joinei bodye slightlsth d an , y hookeextremitye th t da beae th rudimentarys ;di . The fabric is smooth and light brown throughout. Both the original surface and the fractures are covered by a fine network of cracks, as though after it was broken the vessel had been soake waten di r long enougbakee th r dhfo clabeed softeno yha t n d attackean , frosty db . grie opaques i Th t , whit gred ean colourn y i sparsed an , e conditio vien th i ; f e wo th f no fragment sparsite t necessaril sth grie no th s ti f yo y evidenc heavf eo y wear faiA r. amounf to grit survives on the upper surface of the rim. Afte mortariue th r brokes m throwwa d nan n awa t yseemi havo st e remained undisturbed surfac e oroade nth th f ,eo wher t becamei e covered firs watey b t r which softene claye dth , then by black silt laid down by the water, and finally by the levels of the first Antonine fort. It would thus appear to have been broken at the very end of the second Flavian occupation. It is therefore important to establish, as precisely as evidence will allow, the date when mortaria of this type went out of use. The mortarium belongs to one of a small number of closely related forms which have these three feature commonn si broaa ) (a :d witflam ri thrudimentar a y bead, (& yellowish-whita ) e or light brown fabric withouwhic m uses a ri s he wa d a sli a th t pn i gricoatingt ) (c se t d an , rubbing Wroxetesurfaces hi n I . r typ . Bushe-FoeP serie. J r sM x illustrates three varieties of the type, numbering them 10, 14 and 18, and says that the "type hardly appears to last into the second century." This judgment has been generally accepted. It was based on evidence from three sites, one of which was Newstead, where Curie had found several mortaria of this earltypen i l yal , contexts accepo T . t Bushe-Fox's dating without further inquiry, e seconth f o dobtaio t d Flavia t i en indication n a e e d theus date th n th an f o n t eo f no occupation at Newstead, would, however, involve a petitio principii. Published examples of the type from dated deposits include: Brecon, no. 94, C5, Trajanic; , Brecon,C2396 . , no earlies t occupation; Caerhun, , earlies16 . no t level; Caerleon, 1926, nos. 102 and 104, initial structure; Caerleon, 1929, no. 227, initial structure; Caerleon, 1939, no. 69, Flavian layer; Caistor'-next-Norwich, no. B.2, A.D. 70—110; Caistor'-next-Norwich, . R.3no , lowest clay layer; Caistor-next-Norwich, . R.5no . A.D. 110-160; Camulodunum, . 19 no , Neronian5A ; Corbridge, , 1911circa13 d , nosA.Dan .2 .1 100; Fendoch,, 1 . no Agricolan; and Malton, fig. 15, no. 7, and fig. 16, no. 7, pre-fort occupation. In addition, a mortarium of this type was found by Dr K. A. Steer in the earliest level t Ebchestera havo tw e d recentlan , y bee nfirs e levee founth tth f Flaviao ln di n occupation t Corbridgea . Most stratified examples of the type are from first-century deposits. One of the examples from Brecon was found lying on a hard clay floor which contained, low down, a bronze coin of Trajan, minted probably between A.D. 104 and 110, lost while still in almost mint condition. The mortarium fragment could not have been deposited before A.D. 104 at the earliest. Professor Wheeler describe a "distinctivel f e pieco th ss a e y first-century type which rarely survived into the first half of the second century." One of the examples from Caistor came from a pit dated by the other pottery it contained to approximately A.D. 110- 160. Professor Atkinson commente sar tham ri t e "th beath w flattened ef lo dan o p to d

1 The type appears on the Antonine Wall; cf. Balmuildy, xlviii. 21; the possibility that this example is Agricolan and not Antonine cannot be ruled out. 3 VOL. LXXXIV. 34 PROCEEDING E SOCIETYTH F O S , 1949-50.

early characteristics which seem not to continue long after A.D. 100." In view of these remarks, of the wealth of evidence for a first-century date for the type, and of the fact that there was first-century occupation at both Brecon and Caistor, it is probable that these piece survived sha s rubbishda . Mortari f thiao s typ absene ear t from Gellygaer, Throp, Haltwhistle Burn,1 and the milecastles of Hadrian's Wall—sites first occupied in the first quarte e seconth f ro d century. Such flat-rimmed mortari appear—o d s aa cf. Poltross Burn, d 6—aran milecastle2 e , usuallIV , 48 locan yi l hard-fire fabricd dre s with cream slipd an , do not have grit on the rim. Sir George Macdonald considered it possible that mortaria of Wroxeter types 10-18, found on the Antonine Wall (of. Mumrills, 91, 3-7), belonged to the Agricolan occupation.2 Evidence discovered since Bushe-Fox wrote in 1912 has amply confirmed this conclusion. Though fragment sometimee sar s foun second-centurn di y deposit siten so s thaalreadd ha t y been occupie firse th t n dcenturyi , vessel thif so s type appeahavo t t e rno survive e us n di into the second century. It is therefore improbable that the mortarium No. 4 was broken l lateal muc t ra thaf hi n A.D. 100 f thiI . s typ f vesseleo , which hardly outlast e firssth t century, was the type in use at Newstead at the close of the second Flavian period, it would support an abandonment of the second Flavian fort circa A.D. 100.

(c) Antonine Vessels. 5. Found below the clay fillet on the west side of the dividing-wall. The greater part of a mortarium in hard brick-red fabric, with a blue-grey core and traces of white slip; the grit, which is brown and grey, is worn. A poorly impressed stamp has been identifie Birler M fros y ya b d m dieo typstw e recognise e "Bth f "o DOCILISr dfo a , potte probablo rwh y worke north-wese th n di Englandf to . mortariue Th m could only hav int ey positio e founoth wa s dit n discoveres wherwa t ei d at the time the dividing-wall was being built. The distribution 3 of other mortaria with the stamps of DOCILIS is as follows: Balmuildy; Birdoswald; Campfield, Cumberland; Car- durnoc examples)6 k( ; Carlisl examples)2 e( ; Chester examples)2 s( ; Corbridge; Moresby; York. Three of the six examples from Cardurnock—a mile-fortlet on the coast of Cumberland— came fro subsoie mvicinite th th soute n lth i f yho rampart; thi evidencs i their efo r Hadrianic date. It is not unreasonable to suppose that the other three examples from the same site, which wer t stratifiedno e e alsar o, Hadrianic. Birdoswald, Chester d Moresban s y were occupied in Hadrian's reign. There are thus ten examples of the wares of DOCILIS from probable Hadrianic contexts. Of the remainder one is from Balmuildy, which, like Corbridge, was not occupied during most of Hadrian's reign, though both sites were occupied in the reign of Antoninus Pius, while Carlisle may have had a similar history to Corbridge. Thus there Hadrianin te e ar foud can r Antonine examples—those from Campfielo n f Yoro d e dan kar value in this connection—and the inference that DOCILIS was a potter of the Hadrianic- Antonine period, with the emphasis on Hadrian, seems not unreasonable. As there was no Hadrianic occupation at Newstead the present example almost certainly belongs to the very beginning of the Antonine occupation. Its condition shows that it is not a stray, though it is impossible to judge how long it remained in use after being brought to the site, before being thrown precise awath t ya e momen dividing-wale th t builts lwa . Whil mortariue eth m does not exactly date the dividing-wall, its presence below the fillet is not inconsistent with the wall's being built circa A.D. 140. 6. Foun dividing-walle wes0 th d 1 f wito td han 9 Nos , 8 ,, .belo7 bottomine wth e th f go Antonine road; this roa lais ddwa down immediately afte dividing-wale th r demolisheds lwa . Rim and shoulder fragment of a black-fumed cooking-pot, burnished externally above the zone of decoration; there is a wavy line round the neck.

havI 1 e examine unpublishee dth d residu Haltwhistle th f eo e Burn material. 2 This type , appear,3 . like No t Balmuildy;sa possibls . pii cf .t 2 i xli:& 1 .e that these examples are Agricolau rather than Antonine. 8 Cf. Corbridge, Mortarium stamps, AA.', xxvi . 180 p .Cardurnockd an , , CW.*, xlvii . 119p . . EXCAVATION E ROMATH T NA S FOR F NEWSTEADTO 5 3 .

Cf. Balmuildy, pi. xlv, no, 16, Antonine; Cardurnock, no. 2, secondary level; Turret 506, no. 59, A.D. 130-140; Turret 506, no. 83, A.D. 160-197. This is a common and fairly long-lived type; it had emerged before the accession of Antoninu stils s lwa Pius currend an , t afte death s wave hi r Th y. linfeatura s ei e found more frequently before A.Dtha0 16 .n afterwards. 7. Find-spo. 6 . No s a t Rim and shoulder fragment of a large and unusual jar in coarse sandy, dirty buff fabric, simila thao rt certaif o t n amphorce. 8. Find-spo. 6 . No s a t Rim fragment of large jar in coarse grey fabric with a rough reddish-brown surface. 9. Find-spot as No. 6. Rim fragment of a plain-rim platter in black-fumed ware; traces of a wavy line are visible, lightly scorealreadn a n do y burnished exterior indicatee surfacear e d th an n ,d o drawing vessee distinguishabla Th s .i l belongt same bu , th 3 e. o st clasNo e s stypea th d ,an bas bees eha n restored accordingly. Cf. Balmuildy, pi , .Antonine xlviii19 . no , ; Birdoswald, , A.D83 . . 125-197no Kil-d Ol ; patrick, pi. xxii, Antonine3 ,d nosan 2 . . The fragment is not closely datable but would be at home in a deposit of the early Antonine period. 10. Find-spo. 6 . No s a t Rim fragment of an unusual platter in smooth red fabric with a grey core; it is possible that it may be a lid. The above group of five pieces is made up of vessels in use between the beginning of the Antonine occupation and the demolition of the dividing-wall. Except that the cooking-pot belongs ratheearliee th o rt Antoninre parth f o t e perio dlatere tha grouth e o th n,t p throws little light on the date of the demolition. . Foundmake-ue 11 th above n i th , , f witpo e13 roadhd Nosan .2 1 . Rim and shoulder fragment of a dark grey cooking-pot; the surface is decayed. Cf. Balmuildy, pi. xlv, no. 13, Antonine. This high-rimmed piece is typologically more advanced than, for instance, Nos. 14 and 16. 12. Find-spot as No. 11. Rim of a platter or bowl, in dark grey fumed ware; the surface is decayed, but it is still apparent tha vessee th beets ha ln cross-hatched. Cf. Balmuildy, pi. , Antoninexlvii9 . no , ; Birdoswald, , A.D70 . 125-197; Cardurnock,, 30 secondary level; Corbridge, 1938, fig. 7, no. 18, A.D. 139-197. 13. Find-spot as No. 11. Small fragment, about J inch thick, of a hand-made vessel in dark brown crumbly pitted fabric, probably of native manufacture. The fabric resembles that of vessels from the third-century shrin f Vinotonuso e n Scargilo , l Moor, near Bowes. e occurrencTh 1 f o e the present fragment in a dated context is noteworthy, as it provides a contrast between the products of the local native potters and those of the Romanised provincials during the second century. The group of three pieces represents the kind of rubbish that was lying about on the site at the time the dividing-wall was demolished and the second Antonine road built. This could hardly have been much befor etypology e vien A.Dth i f w0 o .16 could an , d well have been later, especiall thercertaints o ya n s ei y t survivetha piecee no th td rubbishs ha d a . 14. Founstructure th ovee n di th nf e o buil t int firse oth t Antonine rampart. Several piece vessea f so l wit dimensione hth beakera f o cooking-poa m f ;ri so e th d an t densn i e black fumed ware, burnishe Shoulderd neckan e th m n ri ,d o . Cf. Balmuildy, pi , Antonine. 1 xlvi . no , ; Corbridge,, 8 . figd no , an 9 . 19385 1 . , figno , 8 . A.D. 139-163; Slack, 8-12, before A.D. 140; and Turret 76, 6, A.D. 122-140. This is a typical early fumed-ware vessel which emerged as a type under Hadrian and barely lasted into Antonine times. In this it resembles the mortarium No. 5; it also resembles the mortarium in having been broken and thrown away at the time that a structure of the first Antonine fort was erected.

1 Yorkshire Archaeological Journal, xxxvii . 111p . . 36 PROCEEDINGS OF THE SOCIETY, 1949-50.

15. Found in the oven built into the first Antonine rampart; this vessel, of which a fair amoun s preservedi t s probablwa , e oveys th thrownf wa o e n us e tim awae th th et ya discontinued. Several conjoining rim neck and shoulder fragments of a narrow-mouthed jar with cordoned shoulder and a lightly scored decoration in straight lines; hard whitish buff fabric with blue-grey surface. "The forts buil Agricola-contaiy b t n only potsherd Romae th f so n type; those buily b t

Antoninus Piusconta-in Romano-Celtionlw ne y e thosth f eo c type." This statemeny b t Collingwood i1 s so familiar that we rarely pause to consider its factual basis. It was Curie's excavation t Newsteasa d that first made suc comparisoha n possibl isolatiny eb g Flavian Antonind an e pottery presene Th . t piecAntoninf o s ei e t belongi date d clasa an , f o st so vessel snorte rarth hn ei before Antonin et then timesno t uncommonbu , parallele Th . e sar to the general shape. Cf. BalmuiUy, pi , ..Antonine 1 xliv . no , ; Oorbridge, , A.D193812 . ., 163-197 no fig , 8 . ; Milecastle 48, pi. iii, no. 11, A.D. 160-197; Mumrills, fig. 98, nos. 1 and 6, Antonine; and Old Kilpatrick, pi. xx, nos. 6-8, Antonine. e Belgie shaporigie th th Th f n ei no c south-eas demonstratee b y ma t compariny db g various feature presene th f so t piece wit uppee hth rpedestae parth f to l . urnC Irof e so nAg Cf. Swarling,. C , KentIroe 4 . nAg no , 16. Foune fillinth gn di abov e oven eth e overampare seales th th e ; nwa th y db f o t second Antonine fort. fragmenm Ri cooking-poa f o t blacn i t k fumed ware; therwavo n s erime i y th lin .n eo Cf. Milecastle 9, no. 56, A.D. 160-197; Oorbridge, 1938, fig. 9, no. 2, A.D. 163-197; and Corbridge, 1938, fig. 9, no. 9, A.D. 139-163. This single fragment migh froe b t ma vesse l broken some time befor e oves th e nwa covered. It is from a typical cooking-pot of the second and third quarters of the second century. A close dating is not possible, but, as the parallels reveal, vessels of this type were in use in about A.D. 160. pieces16d o ,an tw take 5 e 1 , Th n together, provid approximatn ea e indicatio whef no e nth firse oveth t n Antonini e rampar coveres secone twa th y ddb Antonine rampart. 17. Found in the outermost ditch of the fourth and last phase. Rim fragment of a cooking-pot in black fumed ware. Cf. Bewcastle, no. 3, pre-Severan; and Corbridge, 1938, fig. 7, no. 27, A.D. 139-197. This vessel is typologically later than either No. 6 or No. 16; it is of a type whose peak fell circa A.D. 180. It does not necessarily carry the occupation to the end of the century, t indicatei t bu s tha lasditce e opes th th tt hn quartewa ni centurye th f ro . 18. Found immediately below the ploughed earth behind the rampart. Fragmen smala f o t l mortarium with flanged rim whitn i , e sandy fabric grite th ;, which was once present bees ha , n worn away. Cf. Balmuildy, pi, Antonine.36 xlii . no , . This vessel would not be out of place in a third-century deposit, but as something very simila founs ri Antoninn do e sites t doe ,i prov t sno ethird-centura y occupatio t Newsteadna .

(d) Conclusion. Conclusions from the eAddence of the coarse pottery may be stated briefly at this point; for the individual pieces have been discussed in detail, whil fule eth l significanc pottere th f eo y will only emerg e contexth n ei f o t all the other evidence. The end of the second Flavian occupation seems to have come hardly, if at all, later than A.D. 100. The dividing-wall and firse e oveth th t n ni Antonin e rampar bote ar t h databl circao et A.D. 140. Pottery alone wilt dat e beginninno l eth e seconth f go d Antonine period,

Roman Britain, R. G. Collingwood, 1932, p. 100. 1 EXCAVATION E ROMATH T NA S FOR F NEWSTEADTO 7 3 .

but it does give a rough indication—in or after A.D. 160. There is a hint that occupatio havy nma e continuesecone th f o d d dcentury en unti e th l . Ther pieco n s eei obligewhice ar e hdatw o d t e later than A.D. 200.

(e) References Abbreviations.d an PSAS. Proceedings of the Society of Antiquaries of Scotland. SAL. Reports of the Research Committee of the Society of Antiquaries of London. AC. Archceologia Cambrensis. 4 Archceologiad an 3 . AA ^Eliana, thir fourtd dan h series. CW? Transactions Cumberlandthe of Westmorlandand Antiquarian and Archceological Society, seriesw ne .

Balmuildy The Roman Fort at Balmuildy, S. N. Miller, 1922, p. 76. Bewcastle CW?, xxxviii. p. 219. Birdoswald CW.*, xxx . 187p . . Brecon The Roman Fort near Brecon, R. E. M. Wheeler, 1928, p. 213. Cardurnock CW.2, xlviii. p. 108. Caerhun AC., Ixxxix. p. 37. Caerleon, 1926 AC., Ixxxiv . 280p . . Caerleon, 1929 AC., Ixxxvii . 300p . . Caerleon, 1939 AC., xcv . 139.p . Caistor-next-Norwich Norfolkd Norwichan Archceological Society, xxvi. p. 197. Camulodunum SAL., xiv. p. 202. Chesterholm . AA.*,222p . .xv Corbridge, 1911 AA.3, viii. p. 168. Corbridge, 1938 AA.*, xv. p. 266. Fendoeh PSAS. Ixxiii. p. 146. Malton The Defences Romane th f o Fort t Malton,a . CorderP , 1929. Milecastle 9 A A.*, vii. p. 162. Milecastl8 e4 CW.Z, xi. p. 447. Mumrills PSAS., Ixiii. p. 523. Old Kilpatrick The Roman Kilpatrick,d FortOl t a 1928. MillerN . ,S , p. 41. Slack Yorkshire Archceological Journal, . xxvi61 . p , Swarling SAL., v. Turre6 7 t AA.*, vii . 151.p . Turre6 50 t CW?, xiii. p. 351. Wroxeter SAL., i. p. 76.

NOTFRAGMENA N O E GLASA F O TS ARMLET PROM NEWSTEAD. By W. BULMEK. Mr William Bulmer has kindly furnished the following note upon a fragment of glass armlet found in the earlier Flavian level of the barrack-blocks in Section I. 88 PROCEEDINGS OF THE SOCIETY, 1949-50.

Portio glasf no s armlet 2| inches long f D-shapeo , d section (| inc Jh x inch) and 2| inches internal diameter. The armlet has been made by folding a colourless glass rod of circular section round a mandrel, as is shown by the resulting cross-section of the armlet and surface of its inner side. The glass is of good quality, but is interspersed with sheets of exceedingly minute bubbles, evidently due to imperfect fusion. The sides and outside of the ring are "flashed" with a coating, about 1 mm. thick, of transparent glass of deep sapphire blue, evidently coloured by cobalt as it show greenise traco th sn f eo h blue derived from coppe r iroo f r o n s saltsi t I . unusually fine quality and colour, and its use as a "flashing" oil the body of commoner glass implie scarca s svaluabld thawa ean t i t e articltime th f eo t ea manufacture. Into this blue glass are inlaid two threads of white opaque glass about \\ mm. wide and \ mm. thick; they are not quite concentric with the armlet but the deviatio vers ni y slight. e armle Th s beeha t n groune sided outsidth an s n d afterwardo d an e s fire polished; afte polishine th r completes remainingo wa tw e dth g patche opaquf so e yellow glass were adde barbotine.n de Ther doubteo n seem e b , o sparticularlt n yi view of the shape of the shorter patch, that these formed, or were intended to form, an elongated trumpet spiral, the most usual decoration of these armlets. Ther evidenco n s ei e tha threae tth yellof do w glass, which should have joinee dth two patches and formed the elongated spiral, has ever adhered to the surface of the armlet. Probably it was allowed to get too cool in the course of manufacture, and onl thyat e ends, where ther morewas e meta retaito l heat adhesionthe has , n taken place. Glass armlet Romaf so n date confinee seeb mo t d almost entirel northero yt n England and southern Scotland (a list of examples, including others from New- stead, is given in PSAS., Ixxii, 366-395), although one specimen wandered as far sout Wroxeters ha . They usually show considerable artistic ability, particularly in the arrangement of their patterns and in the choice of contrasting colours, as might be expected from their Romano-Celtic origin t whaBu mors t.i e notabl hige th hs e i technica l skill show thein i r manufacture.