Curle, AO, Account of Excavations on Traprain

PROCEEDINGS of the Society of Antiquaries of Scotland

Our full archive of freely accessible articles covering Scottish archaeology and history is available at http://archaeologydataservice.ac.uk/archives/view/psas/volumes.cfm

National Museums Scotland Chambers Street, Edinburgh

www.socantscot.org Charity No SC 010440 ACCOUNT OF EXCAVATIONS ON TRAPRAIN LAW. 139

II. ACCOUN P EXCAVATIONO T TRAPRAIN O SN TH I E W PARISHNLA OF PRESTONKIRK, COUNTY OF HADDINGTON, IN 1914. RY A. O. CURLE, F.S.A. SCOT., Director of the National Museum. Traprain Law, or Dumpender Law, lies \\ mile to the south-south- west of the small town of East Linton in the parish of Prestonkirk, and is conspicuous from the railway as the train passes southward from Drem towards Dunbar. It is situated in an undulating terrain, swelling gradually upwards fro Base mth t Lothian seaboar Lammermuirse th o dt . s summiIt t 0 fee—71 t fee0 abov36 ta leve abovd se e an ls basit e e — commands a wide prospect ranging from the Pentland Hills round by Gullane Hill and North Berwick to Dunbar, while to the southward the Lammermuirs roll backward to the horizon. Of the two names which the hill bears, Traprain Law and Dumpender Law, the latter is the more ancient, but the former may claim an antiquity threr o t oleas a fo e ttw centuries anonymoue th n I . s descriptio f Basno t Lothian publishe Macfarlane th n di e Geographical Collections,1 " Traprain e county Lawth e hill f th thougd o mentiones "i f an ,o e h on tha s da t particular articlcollectioe th n ei n bear dateo sn , yet, fro havins mit g been in Sir Robert Sibbald's possession, one may safely infer that it was writte e seventeentth n ni e namhth centuryr e fo Dumpender s A . e w , firs e forth t m meen i Dumpelde t i t t Kentigern,'S e Lifef o th e n b i r o t 1 referre hereaftero dt , writte Bishoy b n p Josceline probabl 1180n yi . The etymology of the word, kindly suggested to me by Professor Watson, is Dun = a fort, corrupted to Dum, and the Cymric paladyr = a spear shaft, plural peledyr; the meaning might therefore be "the Fort of the Shafts," with possible reference to a palisade which no doubt topped the still existing ramparts. For Traprain the same authority proposes Tref= a homestead, and the Cymric pren, a tree, supplying for the whole the meaning of " the Tree-stead," i.e. the homestead by the tree. Considerable thoug populatioe hth f Dumpendeno r must have been i past times ricd e han relic, th too s sa , found indicate, accordine th o gt standar f wealtdo h then prevailing s houseit , d probablha s y crumbled to dust and its importance vanished before any page of the authentic history of the neighbourhood opens to us. Tradition, or legend, however, incidentally bring interestinn vieo a t t n swi i g connection store f Th yo . t KentigernS , referre aboveo dt , narrate Teneww sho , daughte f Lotho r , 1 Macfarlane's Geographical Collections (Scottish History Society), p. 112. 2 Forbes, Kalendars of Scottish Saints, . 364p . 140 PROCEEDINGS OF THE SOCIETY, FEBRUARY 9, 1915.

Kin f Laudoniago , being witfoune b ho t d child, was, accordine th o gt custor tribehe f m,o cast down fro e heightmth f Dumpendero s a s a , punishment for her supposed crime. From the fate decreed for her she was, however, miraculously preserved, though only to be accused of witchcraft and to be exposed by her father, at the instigation of his heathen subjects landeopee n a sh n e Forthi ,w d boath nean Ho o t. r Culross and gave birth to St Kentigern are matters more intimately connected with the life of the Saint, better known as St Mungo, than with the history of Traprain Law. With yet another Saint, Modwenna, considered by some to be identical

with Medana, the hill is also associated1 . This holy lady (who died in 517 or 518), though carrying on her mission chiefly in Ireland, is said to have founded seven churches in Alba. Her foundations were for the most par hilln o tr rock o s y eminences, which were probabl sitee yth f so existing settlements, and such undoubtedly was Dumpender, the site of f theso e e on foundations. The only referenc ehavI e foun histor n dstatemeni hila e s i th l o yt o t the effect that when the English Fleet was expected in the Forth during May 1547, Patrick, Ear f Bothwello l s appointewa , havo d t e careth f eo e bai r beacoth o l "Dunprenden no r Law."2 The statemen t Statisticald containeol e th n di Account tha name tth e changes wa Trapreno dt e Law Frenco , "frotw e hmth word s traped an reine," owing to Bothwell having carried off Queen Mary to Hailes Castle in the neighbourhood, may be most conclusively disposed of by stating the fact that the name " Trapren" appears in the Great Seal Register as earl s 1451ya . Another piec f informatioo e n whic e authoe th hth f o r foregoing remarkable etymological proposition gives is, however, of more importance to us, as perhaps explaining certain puzzling founda- actuae tionth n slo summipassage hille Th th . f eo t runs: " Several years ago a small plantation of different kinds of trees was made on its summit by way of experiment, and inclosed with a stone dyke or wall feex si t hightreee Th s. succeeded very well while they were sheltered by the wall, but since that time they have not made the slightest progress."s From the earliest time, owing to its commanding position at the edge of a rich champaign and its strong natural defences, the hill must have been periodically under occupation. On plan it is oval, lying with its main axis north-eas south-westd an t somewhas i d an , t pointed towards the former direction (fig. 1). Along the south-east flank a precipitous

1 Forbes, Kalendars of Scottish Saints, pp. 404, 406. 2 Chalmers, Caledonia, vol. iii . 452p . . 3 Stat.-Acct., Whittingham, . 349volp . , ii . 1792. $Q O '00 J00 JM -JOO I""|NM| I I I I

Fig . pla1 . Trapraif no showinw nLa g areas excavated. 142 PROCEEDING E SOCIETYTH F O S , FEBRUAR , 1915Y9 . wal rocf o l k rises fro msteea p rock-strewn slop feet0 heigha o 20 e,t f o t breaking away as it sinks to a lower level towards the south west in hardly less precipitous crags. Along the north-west flank the ground rises directly fro e basmth e very sharpl a heigh o y t f somo t feet0 e5 , •with a broken and rocky surface which in itself constituted a formidable barrier. Above this, on the more southerly half of the north-west exposure, it mounts upwards with- a gradually diminishing gradient to actuae th l summit, whic greaho n lie t ta s distance back fro edgee mth . of the precipice on the opposite side. The fortifications will her describee eb d onlgeneraa n yi l wayt i s a , is the intention of the Excavation Committee at some subsequent date to examine them with the help of the spade more thoroughly than it has so far been possible to do. They practically contain the whole hill, followin e stee e edgth sid e d th gf p eo e on basan slop th n eo t ea the crest of the south-eastern precipice on the other, thus including n area a approximately hal a milf n lengti e y somyardb h0 33 en i s breadth. Starting from the termination of the precipice towards the southern end of the western side, a rampart "with a stone revetment on both faces swings round the broad extremity of the hill, and is thence carried along the edge of the lower escarpment on the north- western flank until it meets an obtruding mass of rock, beyond which the side of the hill becomes steeper and is rough with masses of rock detached and in outcrop. Taking advantage of this chang f surfaceeo rampare th , s deflectei t d abruptl e righth p u to t y e hillth , thence wit a hshar pe lef t i passeth turt o t n s alone th g upper edge of the steeper slope, eventually turning round the north- east end to meet the precipice on the opposite side. • Some distance beyond the rocky outcrop where the first deflection occurs, a terrace breaks acrosa considerabl hile r e flanth fo l th sf o k e distanceo t n O . this terrace at its north-eastern end, and close to the huge quarry which now disfigures the face of the hill, there leads an approach which appear havo st e timformee importann on ea t a d t e accesth o t s fort. The road proceeds up a hollow, and where it debouches on the terrace its outer side is demarcated to right and to left by large stones set on end. The lower edge of this terrace has been defended b ywala s lowelit builto t r s coursea , s (which alone remain), againse th t e bankth edg f .o e Alon e terracegth , proceedin e south-westth o gt a , road may have led, crossing the obtruding rocks by a rough track and enterin rampare th maie gth n i n p tenceint jusga a t y whers eb ha t ei been deflectes ther a hill e indicatione t th e ar bu ;p du f blastinso e th n go line of the track, and as the gap through the rampart is so inconspicuous, s possibli t i e tha e ramparte brea th botth e trac n e i d kth h ar tkan ACCOUN F EXCAVATIONO T TRAPRAIN O S N LAW. 143 secondary. Fig. 2 shows the northern flank with its lines of fortification, s stila r l sovisiblfa e fro ma distance . t certaiA n place e pickesb thert indicationsy dou ma e , often vague and indefinite, of another system, which, to judge from the worn and attenuated appearanc remainse th f eo s possibli , ymora e ancient one. The summit-are hile th l lief o as paralle edge th f th eo o elt south-eastern precipice, wit vero hn y marked alteratio gradienn ni endo tt frod . men Towards the south-western extremity, facing the north, this area is bounded by an escarpment broken up with masses of rock of sufficient importanc treatebe a facto schema to e das in r fortificationeof At .

Fig. 2. Traprain Law from the North, showing lines of fortification. the southern extremity of the escarpment, just where a roadway enters the summit-plateau, traces of an ancient wall or rampart may be observed, sometimes marked by upright stones and sometimes by e debristh , partially covere y turfb d , forming excrescencee th n o s tracee b y surfaced ma meanderin t I . g alone escarp e th edg gth f o e - ment for a considerable distance till the rocky surface gives place to grass, whe t turnni s down ove haunce hile th rinterseco th t l f ho e th t main rampart, coming up from below, jusb at the point where the latter makes its higher deflection to the left, and proceeds direct towards the north-east end of the hill. On the upper side there is no actual contact with the two lines; but from the lower face of the main rampart, the rampart whic presuminm f earliea ho I e b ro gt constructio distinctls ni y visible curvin a shor r fo gt distance farther dow d e thehilth nan l n alon flane gth k north-eastwar a directio n i d n roughly parallee th o t l main defence. At the point where the upper portion of this ancient 144 PROCEEDINGS OF THE SOCIETY, FEBRUARY 9, 1915.

wall swerve o proceet s d dow e haunc e hillt th anothenth ye , f o h r rampart of the primary system may be discerned. Starting here it runs north-eastward toward e extremitth se summit-plateauth f o y , roughly parallel to and at a considerably higher level than the main secondary defence, but fades away before it actually attains the end of hille th . Besides the entrance in the neighbourhood of the quarry, four other gateways give e enceinteaccesth o t s , seemingly arranged in pairs. Approached from the north-east, the first of the first pair occurs about 175 feet beyond the point of deflection of the main rampart, s beeha n d approachean a well-define y b d d track which windp u s the lower escarpment. Through the rampart, with a width of about 10 feet, its course is oblique, pointing eastward, and when it opens on e interio th s flanke e i "westwaro t th t i r o a t rockd y b dy bank, pre- ventin easy gan y acces n thai s t direction s neighbouIt . r fee0 lie13 ts further on and is somewhat wider. Afte intervan a r e f nexsom o feelth 0 e firs f tth e34 t o tpai s meti r , approached through a deep hollow. The approach is faced on the interio a travers y b r f roco e k which completely command t froi s ma heighfee2 1 tf o somt o abov t 0 e e 1 gatewayesolun th th e f o id an , around the ends of which the paths in either direction must have turned approace otheTh e . th carries ri o ht easn a yy d b gradien t along the flank of a bluff, from the crest of which it is overlooked, to the entrance in the rampart, about 12 feet in width. Running from it directly into the interior a roadway can be discerned by the slight hollow which mark e s occasionasurfaceit th s y b s wela ,s a ll occurrence of large stones set on end at one side of its course, leading up the broad haunch, which the hill presents at this point, to a dip in e escarpmenth t demarcatin e summit-plateath g u toward e southth s - west end on the northern side. As this road approaches the dip, a bifurcation may be observed, marked out by occasional stones set on end protruding from the turf, leading in a north-easterly direction towards a fairly extensive plateau that lies just below the hill-top slope. On the surface of the plateau the indications of disturbance and occupation were considerable: nettles luxuriate n densdi e clumpse th , rabbits foun e soidth l d attractivfrean e o burrot e w in; large stones stuck up here and there that were never so placed by Nature, and a stick pressed throug ture hth f showe groune dth loose deepb d o det an . Here, accordingly, it was decided to commence excavation, and for guidance in planning, a datum line was laid down from the beacon on summie th e rampar poina e th south-eas o th n t fee 0 o to e t t21 t th f o t most southerly of the entrances. ACCOUN F EXCAVATIONO T TRAPRAIN O S N5 LAW14 . The areas which were cleared out this year are marked on the plan of the fort (fig. 1), under the letters A, B, C, D, and E, and comprise altogethe a totar l superficial are f abouo a t one-sevent n acrea f .o h Plots A, B, D, and E all lie contiguous at the south end of the plateau, while C lies to the northward of the roadway about 90 feet further down the hill. t presena S oe littlw o t d e periodkno th f w o whico t s r nativhou e forts belong, or of their internal arrangements, that there are practically no existing data available to guide the excavator in his search for foundation f dwellingo s y othean r r o s erections preliminare Th . y work of exploration is consequently fraught with considerable difficulty, until at least the stratification of levels of occupation has been ascertained. This, then, was our first task. Besides the luxuriant growth of nettles which covered the surface of Are , severaA a l large stones protruded like hog-backs througe th h turf ,t circle removan suggestinhu O a .e sitf th o e l s g wa tha t i t of the turf this circle was found to be merely apparent, the form irregular e stoneth , s varying considerabl t generallno n buli d y an ky contiguous. Within it, and abutting on the east side, there was laid bare what appeared to be a segment of an inner circle, indicated in the same way as the larger circle by large stones irregularl yf o varyinplace d an dg o sizedefinitN . e s floowa r exposed, nor were any relics found to throw light on the period of these foundations, if such they may be called. There seemed no doubt, however, that the placing of the large stones was in- f tentionalsomo t e hu d thasord occupie an a ,t ha t e sitet th I d . having been ascertaine f forceo dd dthabe soia t l underlay this level, the stones and earth were removed, and, near the centre of the larger circl d directlan e y beneat e largeth h r e innestoneth f ro s segment, one end of a well-formed rectangular hearth was laid bare, consisting of three n edgeblocko t , e se souths whil a th smal f thiy o o t e la s l paved area t beyonbu ; d these features there wer o placen e d stones, which could be definitely taken as foundations, giving any indication of the form of the superstructure. The evidence was, however, sufficient to prove a second occupation. On clearing off the soil to e paving th e leve th f o la numbe, f pieceo r f Samiao s n ware were found, which at once threw light on the period of this occupation. Digging downward a dept t f a abouo ,h fee3 t inche6 t s froe th m surface, the natural subsoil was reached, overlying which, and immediately beneat f refuso e hearthd th he be consistina , f boneo gd an s discoloured soil occurred, supplying evidence of a third occupation, the earliest of all. As on the level above, here also the evidence of VOL. XLJX. 10 146 PROCEEDINGS OF THE SOCIETY, FEBRUARY 9, 1915. structure was of the vaguest. Several large blocks of stone lay over flooe th r area, but, though their positions were carefully surveyede th , plan produced on paper failed to disclose any structural arrangement. A hearth, however, was revealed as well by the refuse on it as by a long narrow slab forming one side set on edge and firmly sunk in the e a stonrotar e groundon f floo o th ee yrn Th o querarea. y la .n Fig. 3 shows in section the three levels disclosed, the tape measure lying

Fig . Showin3 . relative gth e position e threth f eo s level Aren i s . aA

on the kerb of the hearth noting the second level, and the end of e tapth e touchin e earlies th e floo th gf o r t inhabitatio e siteth .f o n Following out the lowest level to the westward, for the strata above showed [no particular features of occupation at this point, a slightly more definite arrangemen f stoneo t s 'was laid bare, whic probabls hwa y e foundatioth a cla r f turyo o n f dwellin plae gTh n (figindicate. 4) . s ovan a l structure lying wit s maihit n axis north-eas d south-west,an t measuring d 9 feeslightlsom1 an 5 feety 2 eb ty imperfec n fori t m toward e north-eaststh probable du , previoua o yt s remova f stoneso l . Excep t thia t s poin ovae s surroundeth t wa l mora y r lesdb eo s definite outline of stones, largest on the western arc, where one of them was an enormous block studyinn I . plane observee gth on , aren sa a amounting ACCOUNT OF EXCAVATIONS ON TRAPRAIN LAW. 147 to about one-quarter of the interior space devoid of stones on the south, and three triangular beds of stone with their bases resting on the periphery, two opposite each other with a space between occupying the north-west end, while the third occurs to the south of one of these on the eastern side. Just at the apex of this last triangle two stones t righa t tse angle eaco st h other appea mar o rt hearth a site f kth eo . An irregular bed of stones occurs at the south-eastern end of the plan, "where the outline seems to be incomplete. Before leaving the description of the features observed in this area

0 10 f0 EET3 - 111 iniii 11 H ] Fig . Foundation4 . s expose lowesn do t leve f Areo l. aA it may be noted that at more than one place there was evidence in the nature of the soil that the level of the second occupation had been artificially raise filliny db n materiali g . Thus occasional deposits were •encountered of the pan through and beneath which occurred strata of discoloured soil. t shoulI d als recordee ob d that adjoinin hearte secone th gth n hdo level there was found a small oblong compartment formed of stones set on edge, measuring som inche2 1 e widtn si h wit hheigha inches8 f o t , and containing clay, in a plastic condition, of several pounds' weight. On the lowest level a small dome-shaped mass of worked clay about 10 Ibs. in weight was also discovered. Having ascertained from this excavatio t leasa d nt ha thretha e w te U8 PROCEEDING E SOCIETYTH F O S , FEBRUAR , 1915Y9 .

distinct occupations to deal with, and also something of the revealed natur f eacho ee proceedew , d wit e exploratioth h f Areo n a B. To e werw et i also attracte a lin f largy o eb d e outcropping blocks of edgstonen contiguousd o t an e se s , extendin a northerl n gi y direction e weslyind th f Are an o to . g t ThesA a e stone a curve sn o wer -t se e and appeared to bear some relation to the branch road previously noted and marked by similar stones at intervals on its sides. On e wesremova th f thie tur o to th t s f f lin o f lstone o e a singularls y confused mas f stoneo s s exposewa s d revealin o structuren g d an , o point tw y featur r an so e on whiconl t a yh suggested intentional arrangement e stoneTh . s were angular blocks e mos, th lyin r t fo g f thepari d bee s ha ya tn f carttippeo d sprea t an s ou dd about any- groune howTh . d rises here from, sout northo t h , showin n sectiogi n a dept f abou o he south inche8 th 1 d t n f an o ,soi o s n l pa abov e th e of some nortth feee4 n he o excavation tth edg f o e d whilan ; e the settin f stoneo g s bounde e eas th e are t th n do wit a a hdept h from subsoi o surfact l f somo e 3 feete outcroppin th , g rock appeared immediately e westgenerabeloe th e turTh n w.th o f l charactere oth f exposed surface is best seen in the illustration (fig. 5) taken from its. north end. As observed in the lowest level of Area A, there are here also spaces left clear between projectin gf "whic o bed o f stoneso hstw occun o , r hearths s indicatea , alsy observee crosseob y ma d b t I fign si . d 5 . from illustratioe th n tha stonee th t s formin triangulae gth betweed rbe n these- opeo tw n spaces presen e appearancth t f havineo g been carefully laid, in distinction to those on the extreme left of the figure which are lying- much more unevenly. On one of the hearths lies half of the stone of a, rotary quern, while the greater part of another quern-stone stands on edge behind it. The hearth in the foreground had been paved with thi nsandstond slabre f o s e which showed many tracee influencth f so e of heat. On northware clearin th e tur o th t ff Aregof n mor a do aD e regular circular settin f stoneo g s disclosedwa s , wit n interioha r diameter of som feete8 wels a ,s anothe a l r heap t revea whicno s purposed it lh di . Neither over this arer ovee adjoininano th r g therd Aredi eE a come ligho t e samth t e inexplicable layee sout seeth s f stoneo n rhnwa o s a s side of Area B. It is a conjectural explanation of this layer that it. represented either the very complete ruins of a dry-built wall, or the cradling on which a turf rampart had been erected; but the latter hypothesis is weakened by the fact that no swelling remained on the surfac indicato t e e existenc eth sucy an h f eramparto . At the uppermost or north-east side of Area E a number of large ACCOUNT OF EXCAVATIONS ON TRAPRAIN LAW. 149 blocks of rock lay confusedly with their bases about 1 foot 4 inches below the present surface, but there was no indication of any floor or hearth connected with them. Belie smean o wern y eb s plentifun o l this uppe rsectiony levean n i l . After the removal of all the stones that lay on the upper surface, and als f abouoo foota f soil o e tsecon th , d leve f occupatioo l t witme h s nwa Aren i . s HeraA a ecompletela y different condition from wha pred ha t-

Fig . Vie5 . uppermosf wo t leve occupatiof o l Aren no ; remain aB hearthf so s marked thus, +.

vailed on the level above was revealed. Near the centre (as shown on plane th well-constructea ,y figla ) .6 d oblong hearth (fig som) .7 feee6 n i t length by 3 feet 6 inches in breadth, formed of slabs set on edge firmly grounde th n i t , witse hpavina g ove greatee th r r part, beneath which mucs soie th lwa h ope s burned norty wa soutnla d towardd t han I h.an s e soutth h end n froni , f whico t h extende farthea d r are f pavingo a . Some seven feet distant from this hearte easth t n sidho approxid ean - mately parallel lay a row of boulders (see fig. 7) extending contiguously for a distance of nearly 16 feet, beyond which point their line appeared to continue northward, though irregularly e area t e th limiA th f . o to t , som fee6 e1 t distant from this heart north-westerla n hi y directiod nan 0 15 PROCEEDING E SOCIETYTH F O S , FEBRUAR , 1915Y9 .

SECTION N O LINS E A~B 0 3 5 10 0 0 2 10 Illllllllll I TEE]:

Fig. 6. Plan of the second occupation level in Areas B, D, and E. ACCOUN F EXCAVATIONO T TRAPRAIN O S N LAW. 151

lying at an angle to it were the remains, consisting of one corner and some paving whaf ,o t appeare havo dt e bee similana r hearth. Some large stones, as shown on the plan, lay adjacent but did not reveal any arrangement. Somwestware feeth 6 first-mentionee 1 o t t th f do d heart hfea w boulders seemed to have been placed on a curve and may have had some relation to a superstructure of which no trace remained; there was no indication, however, of a floor connected with them. Over Area D the stones revealed no arrangement except at the north corner of the area,

f bouldero e secon Figw . th Heart 7 ro n . o sd hlevean f Areo l . B a where a row, which seemed purposely placed, extended for a distance of some 7 feet, terminating with a large stone only partially uncovered. On varioue Areth aE s boulders scattered ove e surfacth r e discloseo dn definite plan, but towards the south-east side, another oblong hearth was laid same bareth ef o ,typ thoss et a wit et me otheha r part f thiso s level. It measured some 4 feet 3 inches in length by 3 feet in breadth and was paved all over. No post holes were met with at any point, d thougan h her therd cas o discoloureds soin e an eeth n wa l i s wa t i , e extenth o f indicatint o t o s foune b e limit gfloora o th t d f o s . Along the east side of the row of boulders in Area B there lay a bed of clayey soil abou feet2 breadtn i t inche 8 somd r h o an depth n es6 i . This leve f occupatioo l n just n averagdescribea t a ey dla dept f ho 2 15 PROCEEDING E SOCIETYTH F O S , FEBRUAR , 1915Y9 . abou foo1 t t below tha te firs oth f t occupation depta t f abouA ho . 6 t inches lower occurred another level, reveale distinguished Aren dan o , aB d by two hearths constructed on a much smaller scale; also two small areas of paving whic I hshal l hereafter designat s levea e l IB. Thougn i h neither case did a hearth of the one level overlie that of the other, they were situated in such proximity as to render the difference in level obvious. Also pavina ,large e nortth th f eo h o gt heart e seconth f ho d occupation dipped at an angle which, if continued, would have carried it beneath the latter. Immediately to the east of this paving, but at a slightly lower level, was yet another paved area measuring some 8 feet7 fee y ,b twhic h shalI . Among1A l s refee stonea th o t rf whic so h uppee formes th wa y rt i la dston rotara f eo y quern. Beneat t agaihi n s mucsoie th wa l h discoloured, showin stila g l earlier occupation, which appeared to coincide with a more or less circular setting of stones a little further east which was obviously the first. As on these levels no floors other than the occasional blocks of paving were observed, it can be understood that they were extremely difficiilp u m identifyo t tsu o T . e resultsth , evidenc s affordedewa a dept n i f ,som ho fee 4 ef soil o tf o , five different occupations topmose th , r lateso t f whico t h occurred about 1 foot 6 inches below the turf where the soil was deepest. Of these occupations the earliest, the latest, and the intermediate one were recognised in each area excavated and appeared to be general, while e otherth s were very partial e sectionTh . linn o se A—B (fig sho) 6 . w this general group. On variou sseee hile partb th ln o f t thero soblon e ear g depressions, with here and there large stones set on edge in the outline, which e site suco th f huts appeao se h Tw b . depression o t r n alignmeni s t (fig. 8) lay some 90 feet further down the hill from the western limit r distanfa o ft Areno t , e roadparalled fro B ath e lin an f mth . eo o t l uppee measuree Th on r n diametei d e rth feetsom4 d 2 fee an 2 ,y e3 b t lower—some 8 feet distant from it—33 feet 10 inches in length and some 32 feet 8 inches in breadth. An exploratory trench was dug through the centre of both, from the lower end of the lower depression to the opposite end of the upper. In the lower site this disclosed no trace of occupation, the soil beneath the turf being undisturbed s naturait n i ld e conditionuppean th d tha f e di r"o tth r half no o ;f second site differ from this. It was evident, however, that the lower half of the latter and the greater part of the intermediate area were occupied by disturbed ground. Over this area, at a depth of about 1 foot below the present surface, a stratum of grey clay was met with nearly a foot in thickness, not discoloured on the surface, as one would have. expected it to ACCOUNT OF EXCAVATIONS ON TRAPRAIN LAW. 153 be had it been a floor, nor compacted in any way. A very few fragments of coarse native pottery were found in this deposit, also one or two pieces of grey Roman ware. No hearth lay on its surface, and it was difficult to account for on any other ground than that it was e remainth disintegratea f o s d wal f clao l y whic surrounded hha e dth dwelling. Immediately belowe leveth t l a , correspondin intere th o g-t mediat seconr eo d othe e leveth f ro l areas oblonn ,a g heart lais dhwa bare.

Fig . Pla8 . f surfacno e depressions ,sectiod Arean , aC n through them. It measured 5 feet by 3 feet and was constructed in exactly the same manne others,e th s ra outlined with long narrow edgstonen d o et an se s paved all over. Under the paving the soil was burned red by the action of fire. A few fragments of red pottery of Roman make came from its vicinity. Beneat hearte hth e forceth h d soil extended downwardr sfo about 1 foot, much discoloured in places. It contained few relics and little pottery, though include e latteth n di r -were fragment f Samiao s n war blacd ean k cooking-pots clearance Th e botto. th o et m showee dth original excavatio o havt n e been oval, measuring apparently—fos it r limits were obscure toward e south—somth s feet6 1 fee9 ,y 1 eb wit t h s longesit t axi t righa s t angle e site th o that s sf e o tvisibl th n o e e surfaceuppeth n s a sundeptrO wa .o sid f t k about o h i e feet3 t , 154 PROCEEDING E SOCIETYTH F O S , FEBRUAR , 1915Y9 .

the depth gradually diminishing with the slope of the hillside. Against the sides, which were sloped back at an angle of about 45°, stones had been placed at intervals, no doubt to keep up the soil; somewhat towards the north side was an irregular arrangement of large stones whicn i h were forme recesseo dtw f whic o e sh on (fig , reste9) . d against e north-west e banth th n ko eacd an ,h measured som fee6 e n lengti t h fee4 n breadthi y t b . Thus far, with the exception of the hearths, very little that was structural has come to light in the excavation, and no plan has been obtained whic n reallca h y a satisfactorgiv s u e y ide f whao ae th t dwelling e timth ef o swer e like; but, fortunately, other data were

Fig. y. Arrangement of large stones on bottom of Area C. obtaine dexplaio t whic lona y e absenc o structurey ghwa ng th an f eo . e Olowesnth t leve f Areo l a B, toward e easth s t sidd ia nplac an e e where there was much discoloration of the soil, a number of lumps of burned clay were picked up which bore distinctly on one surface the impressio wattlef no portioe on , whicf no shows hi figalonn i 0 1 . g witha sketch illustratin arrangemene gth e wattleth f o t s actually impressen do . it Piece f thiso s material were discovere lowese th n di t leve Aref o l , aA uppee alsth n oi r leve f thao l te uppe areath weln s i ra , s levela f Areo l a last-mentioneo C,tw thoug e th n hi d case quantite sth y extended onlo yt single pieces. As, however, all the examples found, with one possible exception, appea o havt r e been y supposburnedma e ew , that when affecteo s t hea y no e daub ds graduallth t bwa y resolved eartho t , and f clao yd whicbe accouny e alongsidy thith la hma s r fo te lin f eth e o boulder e seconth n e stratuso d th leve r f f s welfo mArea o o l s , a lA a the same material on the upper level of Area C. It may perhaps also affor n explanatioa d e unusuath f no l accumulatio f soio n l ovee th r ACCOUNT OF EXCAVATIONS ON TRAPRAIN LAW. 155 various levels during a comparatively short period of time, as evidenced by the respective relics to be discussed hereafter. In the Romano- British village at Woodcuts, excavated by General Pitt-Rivers, similar

evidence of wattle and daub construction was found; and more1 recently piece f clao s y impressed wit e markhth f wattleo s s have com ligho et t

INCHE2 S I Olinlinl——— h

Fig. Piec10 . e of burnt clay with impression f wattleo s sketcd an , h showing their arrangement. in the excavations on the site of the Roman town at Wroxeter2—both, as the relics show, of coeval occupation with Traprain. Let us now turn to the relics recovered from the various areas excavated, and see what reasonable deductions we may draw.

. POTTERYI . If we consider in the first place the pottery, we find in the lowest level of all on each area many sherds of native pots, thick hand-made ware, fashioned from material unrefined and mixed with pebbles and grit (fig. 11, Nos. 1-3). In the majority of cases the fragments are those of pots with straight sides thinning away to the lip, which is rounded, though occasionally havin g flaa r bevelle o t d lip s show,a n ni the illustration of sections (fig. 12). The sooty incrustation of the walls of almos l thesal t e vessels shows that they have been cooking-pots. They are all absolutely devoid of decoration either by applied or incised ornament wit single hon e exception, wher sherea crosses di d horizontally 1 Pitt-Rivers, Excavations in Crariborne Chase, Woodcuts, vol. i., pi. lii. 2 Bushe-Fox, Excavations at Wroxeter, Shropshire, in 1912. PI. v. fig. ]. 156 PROCEEDING E SOCIETYTH F O S , FEBRUAR , 19159 Y .

b broaya d impressed band thin I .s respect they differ markedly from e potter e kitcheth th f yo n midden e wes f th Scotland o tn o s d alsan o, from that of the brochs. The curvature of the sherds indicates that the vessels were of fairly large dimensions, but not enough pieces of any pot could be put together to show the original size and form; in

0 INCHE4 S

Fig . Specimen11 . f Nativo s e Pottery fro e varioumth s levels. thickness they average about ^ inch, though some pieces exceed half an inch. In addition to the large coarse cooking-pots there was represented a much finer vessel, presumably a beaker, made of washed or refined clay, smaller in size, with thinner walls not exceeding •£$ inch in thickness, and revealing in section in each example a straight side and a simple rounded lip (fig. 11, No. 4). The difference in character is apparent contrastinn i e sectionsgth , that show fig.y fro4 n b 1 e uppe. mth r level e last-mentionebeinth f o g de secon risee th quality on o t sds A level. , a marked diminution is noticeable in the supply of the native pottery; ACCOUNT OF EXCAVATIONS ON TRAPRAIN LAW. 157 and by the time one reaches the uppermost level of all it has practically ceased. Fig show3 .1 e sectionth s f fragmento s se founth n o d

. FigSection12 . f Rimo s f Nativo s e Pottery from lowest levels. (J.) second level, and fig. 14 that of the only native sherd found on the uppermost level. On Area C a few fragments came from the bed of

Fig. 13. Fig. 14. Section f Rimo s f Nativo s e Pottery: fig fro3 .1 m second level; fig fro4 .1 m uppermost level. (J.) clayey soil which overla secone yth dAren levelo somad B ,an e fragments of one of the more finely made cups alone came from the top. From all three levels .came fragments of Roman pottery: little in 158 PROCEEDINGS OF THE SOCIETY, FEBRUARY 9, 1915. proportio e nativth o t ne ware fro e lowesmth t level a considerabl, e excess ove nativthe r e ware fro seconmthe d a smallelevel and , but r still an appreciable amount from the top, where native ware was virtually unrepresented. Most of the fragments are small, but a careful collation give e followinth s g results f terraO . sigillatae therth e ear remain f twenty-sio s x different vessels f thesO . e twelve came from the lowest level, ten from the second level, and four from the top. Of vessel f otheo s r Roman wares some eighty-four appeae repreb o t r- sented, comin e proportionth n gi f thirty-ono s e fro e lowesmth t levels, thirty froe secondmth d twenty-threan , e fro e latesmth t occupation. This give Roma0 totaa s11 f o nl pot thin si s small area. An examinatio e terrath f n o sigillata fragments give e followinth s g results J:— 1. Fragments o/ Vessels f Undecoratedo Type. 1. The cup, type Dragendorff 27, occurs three times—twice from the lowest level and once from the second. All three pieces might belong vesselo t s possibl e firsy th mad tf o century ed piece beforen On e e. th e which came from the earliest occupation of Area B is particularly suggestive of the pottery of an early date from the brilliance of the glaze and the hardness of the body. fragmene 2On . t fro e uppemth r level appear o belont s a larg o gt e platter, one of the many later developments of type Dragendorff 18, approximately the same as the type No. 10 from the Pudding Pan Rock Collection (Proc. Soc. Antiq., 2nd ser., vol. xxi. p. 279). 3. A small fragment from the lowest level belongs to a vessel which appear o approximatt s o typt e e Dragendorf . Ther18 f s nothingi e , however a first-centursuggeso t t ,i r fo t y origin excepn o t s i tha t i t the whole of moderate thickness, and the form of the lip is not exaggerated. 4. A portion of the rim of a large bowl from the upper occupation level might approximat e Puddinth o n et Roc Pa g k (ibid.). 9 typ. No e It is rather coarse and probably of second-century date.

2. Fragments of Decorated Vessels. 1. From the lowest level. (a) A fragment of the rim of a bowl, type Dragendorff 37. From the narrow rim above, the usual egg and tassel moulding, and general character of the fragment, a first- century origi s stronglni y suggested. indebtem a brother y I 1 m o dJamet r M , s Curie, F.S.A. (Scot.) terra e noter th fo , n s o sigillata. ACCOUN F EXCAVATIONO T TRAPRAIN O S N LAW. 159 smalA ) l1. (6. fragment) No (Fig , 15 . e , th a bow als , f o o37 l desig f whicno h suggest first-centura s y origin. fragmenA ) 2. (ca bowl . ) f (Figo No t ,, als15 .f typo o , wit e37 h medallion decoration, the subject of which is unidentified. 2. From the second level. (dfragmenA ) e uppeth f o r t a globulapar f o t r vase, type Dechelette 67. This is a first-century type, and the dish of which fragmena thi s i s y havma te belonge o that d t period example Th . e seem havo st e bee f ratheno r larger size than those found at Newstead (A Roman Frontier

Fig. 15. Fragments of decorated Samian Ware. (§.) Post, plate xxxix. figs. 7 and 8); it shows no part of the decoration. (e) (Fig. 15, No. 3.) A small fragment of a bowl by Cinnamus, bearin f second-centur o a g portioe s name i th t I f o n. y Lezoux ware. fragmenA ) (/4. bowla . )f o (FigtNo , typ .15 fro, e37 m Lezoux, showin a gfigura gladiato f o ea larg n i re medallion (Dechelette type 117) f second-centuro ; y origin. fragmenA ) 5. (grsecond-centura . )f o (FigtNo , .15 y bowl, typ, e37 with a floral design. a bowl f o , m fragmentyp(h) , A ri coars e 37 rathed th ean f o t r late. (i) (Fig. 15, No. 6.) A fragment of another bowl, type 37, with what appear metope b o t s e decoration e floraTh .l orna- 160 PROCEEDINGS OF THE SOCIETY, FEBRUARY 9, 1915.

beet menno ns identifiedsomewhaa tha s ha t I . t Germanic charactepossibly ma d yan r belonTrajan-Hadriae th o gt n period. 3. From the uppermost level. (j) A fragment of the rim of a large coarse bowl, type 37, showing none of the decoration; of second-century date. fragmen(kA ) e bassmala th f eo f lo t bow globular o l r vaso eto small for identification.

3. Fragments f Unglazedo Roman Pottery. The fragments of unglazed pottery are for the most part small, but all the larger pieces are illustrated in figs. 16 to 18, and sections are give . fign ni 19 . {a) From the Lower Levels. Fig. 16, No. 1; sec. fig. 19, No. 4. Portion of the neck and lip, the latter much everted, of what appears to have been a fairly large vessel; the warcolourd t seemi re burns d ei a havo st an , o t t e been coated with yellowisa e outehth slin rp o surface . greA . 2. y;1 . No secburnishe , FigNo . 16 , fig. 19 . d ware ornamented wit hseriea narrof so w bands wit hmata considerabls i t p surfacli e th e; y everted. a a smale portio sid e f bason d o eTh l an f e o n. 3 . No , Fig16 . globular vessel of a light brown colour, coated on the exterior and partially on the interior with a buff slip. Fig. 17, No. 2; sec. fig. 19, No. 2. Cooking-pot of a brownish tint on the sides, which are rough and ornamented with a lattice pattern, while the darkee ar polishedp d huen i li r an nec,d kan . Fig. 16, No. 5. A rather thin red ware covered on the exterior surface with a black slip and decorated with rows of short impressions produced by a roulette. On the shoulder there is formed a slight overhanging ridge informe m d thisr Thomaa an M I , , y b ds May, characteristis i 1 c of a particular type of bulbous beaker. ornamented with roulette impression f fourth-centuro d an s y date. Thi s difficuli s beint i , tso g accouno t r thifo ts late example being bottoe founth n i dm level. Fig. 18, No. 1. Grey ware decorated with lightly impressed and burnished vertical lines.

I am indebted to Mr May for the following references:—May, Roman Pottery in York Museum,1 plate xi. 7; "Roman Camp at Housesteads," Archceologia JEliuna, xxxv. p. 297, . figRoac55 . h Smith, "Roman Sepulchral Remains foun t Strouda d n Kent,i , " Collectania Antiqua,. 17 . p . i

162 PROCEEDINGS OF THE SOCIETY, FEBRUARY 9, J915. . Gre2 . y No war Fig, 18 .e with lightly impressed lattice ornamentation. Fig. 18, No. 3. Grey ware with a lightly impressed and burnished scroll ornamentation.

Fig. Fragment17 . s o{ Cooking-pots with lattice ornamentation.

(b) From the Second Level. Fig. 16, No. 4; sec. fig, 19, No. 6. Portions of the mouth and shoulder oa globulaf r vesse f ligho l t grey colour f fino , e texture, made from well-washed clay; the lattice ornament on the shoulder is lightly im-

Flg . Fragment18 . unglazef so d Roman Pottery. (^.)

pressed; the side of the pot has been slightly pressed inwards in the formation of the loop-like handle. Fig. 17, No. 1. Portion of the side and the base of a cooking-pot, . simila2 . No o figt r , .17 Fig. 18, No. 6. The base and portion of the side of a small grey vessel whic bees hha n decorated with vertical burnished lines. ACCOUNT OF EXCAVATIONS ON TRAPRAIN LAW. Fig. 18, No. 4 ; sec. fig. 19, No. 10. Portions of the lip of a dark grey cooking-pot. Fig. 18, No. 5 ; sec. fig. 19, No. 12. Portions of a similar vessel, but of a lighter shade of colour. Fig. 19, No. 3 (section only). Small fragment of grey ware. (sectio5 . No Fig, n 19 only). a vesse f .o f o lp Fragmenli e th f o t a brownish-grey colour, and of a hard ware showing many particles of mica in the body. Fig. 19, No. 7 (section only). A small fragment of the mouth of a vesse f ligho l t red-coloured ware. Fig. 19, No. 8 (section only). Portion of the lip of a vessel of a hard ware, gre colourn yi .

Fig. 19. Sections of Kirns of unglazed Koman Pottery. (-|.)

(sectio9 . No Fig, n 19 only). . Fragmen a thicke f o t r war f lighteo - colourd re . Fig. 19, No. 11 (section only). A very small fragment of the lip of what appears to have been a small vessel of light buff ware.

(c) From the, Highest Level. Fig. 16, No. 6. One of several fragments of a thick, black, rather coarse ware ornamented around the shoulder with a band of impressed wavy lines. Fig. 16, No. 7. Loop-like handle, apparently part of the same vessel vessee lase th side bees th tf s th ;eha o la n formatioe th presse y b n di n e loopoth f . Fig. 16, No. 8. The greater part of the base of a vessel of buff ware coated with a chocolate-tinted slip ; the base is solid and bung-shaped. Mr Thomas May recognises this fragment as the base of a beaker of a ware resembling Castor ware n exampla , f whic o es foun wa ht a d 164 PROCEEDINGS OF THE SOCIETY, FEBRUARY 9, 1915. Pevenseyl and assigned to the fourth century. Similar bases are met with vasen Foreso w s Ne fro te kiln-wastemth latese th f sto Roman period. verA y. 7 smal. No Fig , l portio.18 vessea f no reddisf o l h colour which bees ha n ornamented with parallel line f roulettso e impressions. 8. . ; Portio secNo Fig18 e shoulde, . th .18 f No tig nvesseo a , f .19 o r l of light red colour decorated with burnished vertical lines. 3 (sectio1 . No n Fig, only) 19 .a vesse f o .f o lp Fragmenli e th f o t hard buff ware. (section4 1 . FigNo , only).19 . Portio flangf no lighf eo t tile-red coloured war whicn eo h remain trace a thi f no s black slip. This appeare b o t s a fragmen a flange f o t d bowl resembling imitation i , f Samiano n ware, form Dragendorff 38. Two bowls of the type and fragments recovered from potters' kiln 187n i s 9t Sandfora d Farm, Littlemoor, near Oxford,

Fig. 20. Inscribed Sherd. (§.) preservee ar e Ashmoleath n di n Museum specimend an ; s were obtained decorated with conventional pattern n whiti s e sli t Pevensey.a p A 2 fourth-century date is assigned to the type.3 (sectio5 1 . No n Fig , only)19 . . Portioe mout th a vesse f f o no h l of heavy grey ware, the moulding of the lip somewhat flat. Fig. 19, No. 16 (section only). Fragment of the lip of a small vessel of red ware; the lip is much everted, and around the neck are two parallel incised lines. Fig. 19, No. 17 (section only). Fragment of the lip of a vessel of coarse black ware e surfacth , e much weathered off. (sectio9 1 . No nFig , only)19 . mortariuma f . o Fragmenp li e th . f o t This fragment came from the second level of Area C and appears to be of the Antonine period. Having an interest greater than that attributable to the pottery itself a portio vesse a e sids i f th e o f gref o nlo y waree insidth ,n eo fig , .20 1 Sussex Archceological Collections, vol. 1 li. . , platNo . exv 2 Ibid., vol. lii., plate ix. Nos. 2 and 5. e identificatioth r 3Fo n of thi e references th specime r e discoverfo th e typd o t sth n an e f yo elsewher indebtem a Thomar eI M o dt s May. ACCOUNT OF EXCAVATIONS ON TJRAPRAIN LAW. 165 of which are incised in Roman characters the letters " I R I" and a stop. It came from the second level, and as other pieces presumed to be parts of this vessel were also found there, as well as from the fact that the letters appear on the interior surface, the assumption is justified that the latter were scratche e hilth l n itselfdo .

II. PERSONAL ORNAMENTS. Under this head come fibulse, pins, clasps d othean , r small objectf so bronze e numbeTh . f suco r h article s remarkabli s e considerine th g

Pig. Portio21 . a Penannula f no r Armlet of Bronze. (•£.) small area excavated; the majority of them came from the lowest level. At the very bottom of this stratum, almost in the subsoil, there 1 a-

Fig. Penannula22 . r Brooche f Bronzeso . (J.)

was found a portion, about one-third of the whole, of a penannular armlet of bronze (fig. 21), with a flattened expanded termination. This is of a common Bronze Age type. Two penannular brooches in the collection, with small fluted knob terminals (fig. 22), came, one (No. 2) from the lowest level, and the other (No fro) .1 m what appeare slightla e b o dyt later occupatioa t na level only some 6 inches higher. The type is a common one on Roman site thin i s s s founcountry wa t Newsteada d t I n identica.a d an ; l covered an specime , 1 . d witNo fig, o nt equalln h 22 a . y beautiful patinas i , 6 16 PROCEEDING E SOCIETYTH F O S , FEBRUAR , 19159 Y . among the relics found in the recently excavated Roman fort at Beniulie, near Glasgow. A fragmen f anotheo t r penannular fibula f smalo , l sizd witan eh a plain knob terminal, came fro e seconmth d level. There are seven bow-shaped fibulae, of which four came from the level of the earliest occupation, one from the stratum immediately superimposed, one from the second level, and one from the top. The four from the earliest occupation (fig. 23, Nos. 1, 3, 4, and 7), include f eac o workes kneo n hha a sprin pi tw en e do fibula gTh . e4) (Nosd an .3 which e types usuath a , s containei n ,i l a semi-cylindrica n di l casing. No. 3 is of a form commonly found on Roman sites of the Antonine period, specimens coming from the fort at Newstead,1 from Camelon,2 d froan m Wroxeter.s somewhai 4 . e shapNo Th f 3t o eslighte r than e usuath tha f l o tkne e peculia th fibula d an ,r characteristic which gives its name to the type is less pronounced; it has also a small ring furthes heade i attacheth t I t .a rt i remarkabl o dt remaine th r f efo o s silver platin ge observeb whicy ma hd upoa smal , it n l rosette still remainin e sidon e n towardo ga e shor fibula th d e foo f tth an so , t r abov ba e sprineth g cover. The third fibula from the same level (No. 7) much resembles, in form and scheme of decoration, a brooch found in the Victoria Cave, Settle.* Along the top of the bow runs a row of lozenge-shaped spaces filled with pale blue enamel, while triangular point f yelloo s w enamel fill e angleth s between, along both sides e receshea A s helth . dha t d a s a bossawantingw no , semicirculaA . r ridg r creseo te bac riseth kt a s of the head and has a groove running down the centre of it. A somewhat similar crest appears on the fibula shown in fig. 24, No. 3. e fourtTh h bow-shaped fibula fro e lowesmth s ti leve) 1 . l No (fig , .23 of a type perhaps best known from the pair of silver-gilt brooches foun t Backworta d Northumberlandn hi numbeA . f exampleo r s were recovere t Newsteada d , som f theeo m richly ornamented with enamel and chasing. This brooch is quite devoid of any enrichment, though its line e gracefular s , and e floriates i th ' w bo d e centre knoth th f n eo bi well executed. It has a spring foi* the pin, and the axial wire of the spring, caught in by a collar, forms a loop at the back of the head. The example from Newstead whic t i mosth nearly resemblee on s i s found in one of the chambers at the rear of the principia, a find-spot believed to be not earlier than the middle of the second century. fibul e fro) 6 Th . ma No (fig leve , jus, 23 . 1A tl abov ee lowes th tha f o tt 1 James Curie, A Roman Frontier Post, plate Ixxxvii. figs. 31 and 32. 2 Proceedings, vol. xxxv . 402p . , fig. .43 3 Bushe-Fox, Excavations at Wroxeter, Shropshire, 1912, p. 25, fig. 9, No. 4. 4 Boyd Dawldiis, Cave Hunting, . 99p , front, fig. .1

168 PROCEEDING E SOCIETYTH F O S , FEBRUAR , 1915Y9 . occupation, was found on a paving of stones towards the north end of Area B, which probably represented a hut site. It measures 2J inches in length; the bow, which has a flat curve, has been enriched with a series of rectangular compartments of enamel only one of which now retains the material, red in colour. There is a boss towards the head bace th ksmala t a bowe d th len f an hoo, do k whic hels maie hha d th n loop of the spring, now awanting. This is a fairly common type of brooch, several example f whico s h were foun t Newstead,a d 1 while eon occurs among the hoard of Romano-British relics found many years ago on Lamberton Moor, Berwickshire, and now in the National Collection.2 This particular example, however, shows features which givplaca t ei e midwa developmenn yi t betwee e specimennth s from thes placeso etw . The Lamberton Moor brooc enamelles hi witdw alonhsimilaa bo e gth r serie f oblono s g compartments furthes i t i ; r furnished wit ha spring , as is this example, the loop of the spring passing across the back of the w bo head beind e th an , gf a hoo o caugh n d ko en fastenep u e t th o t d b a ystud . Whil Trapraie eth n brooch still retains ha hooke sth d an , originall springe th d e stuy,th ha d which hel formee dth develope s ha r d into a purely ornamental boss. When we turn to the Newstead fibulae we find tha e sprinth t gives gha n plachingea o et d pin thad ,an t with sprine th e hoo gs alsth kha o vanished. e NewsteaMosth f o t d brooches of this type were found near the surface; one came from an outer e extendeditcth f ho d fort, thus affording negative evidence tha t couli t d t belonfirse no th t o gt occupation majorite Th . thef yo m were assigned to the Antonine period. Associated with the two knee fibulae (fig. 23, Nos. 3 and 4) in the lowest level of Area A, and found within a few feet of either of them, was a disc fibul complet) 5 . a (figNo , e .23 wit s s beautifullpini hit t I . y enamelled ove s conveit r x surface with triangular space f redo s , white blud an ,e enamel unfortunatelw no , y much tarnished with copper oxide similaA . r brooch was recently found at Wroxeter8 associated with two coins of the Emperor Vespasian; another resemblin t gcami e from Camelond an , e Nationath n i w l iMuseum;no s 4 Newstead yielde t anotherdye , rather more elaborate, e appearsfro e bathson mth d havo t an ; e been found at Corstopitum in 1908.5 Fro e seconmth d leve Aren o l camaD e another fibul) 2 . a (figNo , .23 e lapsth f tim en o e samI oth ef . e1 whic , typ No s fig h, ea occurre.23 d between the two occupations from which these brooches were respectively 1 A Roman Frontier Post, p. 323, plate Ixxxvi. figs. 19 to 22. a Proceedings, vol. xxxix . 367.p ; illustrate RomanA n di Frontier Post, . 320p , fig. 466. s Excavations at Wroxeter, Shropshire, 1912, p. 26, fig. 10, No. P. 4 Proceedings, vol. xxxv. p. 405, plate A, fig. 2. 6 Corstopitum, Report, 1908, p. 100. ACCOUNT OF EXCAVATIONS ON TRAPRAIN LAW. 169 recovered e typ s suffereth , eha a slighd t modification- fore un th ;s m i changed, but in place of the pin working on a spring, as in the earlier example simpla , e bee s hingeha nn employeddpi , and corollarya s ,a e th , loop at the back previously formed from a wire passing axially through sprine coile th th f so g has, springe wit losth e hf th so , becom integran ea l part of the brooch. On the trumpet-shaped head there is on either side a single volute, the last fading expression of the characteristically late-

Fig. 24. Fragments of dragonesque Fibulae from the lowest level, and a bow Fibula fro highese mth t level. (\.) Celtic enrichment which distinguishe e fineth s r brooche f thio s s class foun t Newsteaa d d approximatelan d y dated, from their find-spotso t , e firs th e secontth hal f o df centur r eraou .f yo Trace f tinnino s e gar e bow th seee b stiln .no o t l The last of the bow fibulae (fig. 24, No. 3) is one of the few relics from the upper level of occupation. It appears to be allied in type to fig. 23, s roundei w plaind bo d an e ;Th therNo . e a cavithea7 .th s ei dn yi whic sidese hingeds i containes th n n hha , pi whicO .e th dbossha d an , for mcontinuoua s curve intheade devicth oa f s eithei ,o enamen d ei ren l consisting of two opposing crescents of deep blue, facing inwards between two triangular compartments, now too much decayed for the 170 PROCEEDING E SOCIETYTH F O S , FEBRUAR , 19159 Y . colourecognisede b o t r fore mTh . bears some resemblanc fibula o t e a found at Corstopitum in 1910.1 On the head end of the bow, between cavite th y from whic e boss beeth h ha s n e removecrosspieceth d an d , there rises a thin semicircular crest, with a slight groove along the top. This a appearrudimen e b a loo o f t spo t employe o holt dn i d position the axial wire of the spring in the same manner as the hook referred to above: a somewhat similar feature may be seen on fig. 23, No. 7. Fibulae illustrating this arrangement were found by General Pitt-Kiver s excavationhi n i s t a Rotherles d Bokerlean y y Dyke.2 The lowest level further yielded two fragments of the " S " or dragonesque fibula (fig. 24, Nos. 1 and 2), a characteristically Celtic ornament of Romano-British times, and generally dated to the second century. Each of these fragments is a head or terminal portion. The larger of the two (No. 1) consists of the head, about one-third of the body, and a portion of the pin, which, as in all other specimens of the type, works brooce nece oth nth f ko h round whic s bentbronze i t Th hi . f whic o e h fibule th mads ai s somewhaei t decayed t therindicatioo n bu , s ei n that ther bees enamelliny eha an n remainine th n go g portion however, As . , these brooches were frequently enamelled on the centre of the curving body, the part amissing may have been so enriched. The metal is thin, and the appearance of the head is less pleasantly artistic than in most specimen class the leaf-shaped the ;of s , ear-like projectio somewhais n t narrow e '.'snout,th ; " endin a sligh n gi t a knobrestricte s ha , d curve and does not turn back spirally, and the space in the body through which the pin turns is merely large enough to serve its purpose, and e generalitth s notn i ha s f ,examplesa o y , been treate n artistia s da c feature. It is worthy of remark also that no eye appears upon the head. This fibula somewhat resembles one found at Corstopitum in 1910,3 except thae latteth t s enamelledi r t i probabl; y comee s th lat n i e series. No. 2 is a smaller fragment, consisting merely of the " ear" ane " snout.th d " These feature e mor ar skeepinn i e g wit e typhth e as known to us; the ear. has been enamelled, probably in red, though the colour has gone, and the snout is considerably recurved. An enamelled brooch of this type was found at Newstead,4 and another is in the collection of objects found on Lamberton Moor.5 These e onlth yappeae b recorde o t r d previous find n Scotlandi s , thouga h

1 Corstopitum Report, fig, 1910. 40 . 13 . p , 2 Pitt-Kivers, Excavations in Cranborne Chase, etc., vol. ii., plate xcvii. fig. 4; vol. Hi., plate clxxv. fig. 3.

3 Corstopitum Report, 1910, p. 45, fig. 28. RomanA 4 Frontier Post, . 320p , plate Ixxxv. fig. .7 5 Ibid., . 320p , fig. 46a. ACCOUN F EXCAVATIONO T TRAPRAIN O S N LAW1 17 . considerable number have been reported from England, chiefly from the north.1 A T-shaped fragment of silver, f inch in length, from the lowest level may possibly be a part of a very small bow-shaped fibula. Four complete bronze pins were found, besides fragments of one or o otherstw , measurin1 .. No Fig , f inche25 g3 . lengthn si , came from the lowest level on Area A, and appears to be of a type similar to one found at Newstead, though the latter is believed to have had a small

circular setting of re2 d enamel on the head. The stem has been encircled t intervala s with band f inciseo s d rings steA .m only bronzef o , , incised with similar rings came likewise from the lowest level, as did also the f anotheo uppe d en r r bronz n witpi e ha flattene d knob-hea f smaldo l diameter (fig. 25, No. 4). From the second level on Site A, associated with a number of pieces of second-century Samian ware, there came a bronze shoulder-pin (fig. 25, No. 3), with a ring head bevelled to the inside. A similar pin was found in the Ness broch at Freswick, Caithness, while others of iron, without the bevel, were discovered respectively in the kitchen-midden t Gallanacha e Laws, Th e for Obant ,th a tMonifieth n i , , Forfarshire, buriae Iroe th nAg t Moredunn a i l d an , near Edinburgh.8 On the second level also, but on the adjacent Area B, was found a typ f o e ) know2 e hand-pin . th (fign No s n, a pi .25 a , fro mfancifua l resemblance it presents to a closed hand, the flat lower plate repre- sentin e palmth g , whil e pelletsth e , which for e uppemth r segment, denote the knuckle-joints of the folded fingers. Like the last noted, purels i thin f Celtispi yo c character t thin, no thoug ko d tha e I h w t have anothe f identicao r e lNationa th for n mi l Collection wite On h . four pellets was found at Corstopitum in 1910,* and is figured along- side of a pin of the type of fig. 25, No. 3, found in the same year. The Traprain specime s beeha nn coated with tin d stil,an l retains s surfacoit n a considerable e amoun f thao t t metal s lengtIt . s i h 3rV inches. 7 A lon f gbronzo thi n npi e wire, 4T ¥ inche lengthn i s , taperine on o gt s muci t I h ends . waste it 6 e headshows i . ,th d t No an ,dfiga n n, i .25 feature barele ar s y determinable. , als 7 f bronze. o No Fig, s veri ,25 . y suggestiv a swan-nec f eo k pin, metae th t mucs i l bu h decayee objecth d t dan imperfec t bota t h ends, so that I do not consider that a definite determination is justified. It 1 In an appendix to the Corstopitum Report, 1908, p. 116, Professor Haverfleld gives a list of recorded thao findt p t su date . 2 A Roman Frontier Post, p. 337, plate xcii. fig. 11. 3 Proceedings, vol. xxxviii. 1906, p. 433 et seq., figs. 5, 7, and 8. 4 Corstopitum Report, 1910, p. 47, fig. 34. 172 PROCEEDINGS OF THE SOCIETY, FEBRUARY 9, 1915. much resemble e Happfounn th pi n i a sdy Valley, Woodendean, near Brighton, along with a small bronze boar.1 The swan-necked pin, how-

Bronz. Fig25 . e Pin Claspsd san . (\.) ever, belongs to an early period of the Iron Age, and this particular object was found on the highest level of occupation. A dumb-bell-shaped button of bronze (fig. 25, No. 10) came from the

lowest level on Area C. Two bronze buttons of this type were found Proceedings1 of the Society of Antiquaries, Second Series, vol. xxi. p. 489. ACCOUNT OF EXCAVATIONS ON TRAPRAIN LAW. 173 t Newsteada f theo me on ver, y simila o this;t r 1 anothe s reportei r d from Wroxeter.2 e claspsar , 9 , d probablan Pig, Nos, 8 25 .r , garments.5 fo y e th , triangular portion being presumably passed through a slit in one side oe a garmenmaterialf th o sewd t an n t no , whil a eloo p fasteneo t d e otheth r s passesidwa e d ovee hea th ro complet t d e fasteningth e . Tw f theso e objects , wer8 , d Nos ee lowesan founth .5 n o dt level, camwhil 9 s bee. ha e eNo n 5 fro . enamellede topmth No . e centrath , l panel containing oval spaces of blue on a white ground, while the triangular spaces on the two borders have been filled with scarlet. Various examples of these clasps were found at Newstead. The pin-like object from the second level illustrated in fig. 26, No. 2, must, I think, be an awl for piercing leather. It is of bronze, 2| inches triangulaa in s lengthha d an ,r head folde d e endlon e oveth Th gt . a r flat head woul unsuitable db pina r s th,ea fo e rapid increas breadtn ei h from the stem would give it a tendency to slip out of any garment it had been thrust into, wherea e broath s d head coul e helb d d com- fortably between the finger and thumb if the object was to be used for piercing reliA .f almos co t identical fordimensionsd man t lackinbu , g the rolled termination, was found with others in a fort in Ostergotland, Sweden, dated about 400 A.D., and was regarded as an arrow-point,3 but for such an end the rolled head makes the Traprain specimen unsuitable; it would not have assisted the fastening, it is probably broader tha arrow-shafn na t "would have t doei been t confor d sno an , m to its curve. fro, 1 secone . mth No Fig , d 26 . leve f occupationo l terminaa s i , a r fo l strap, notched at the point so as to be used as a nail-cleaner or, perchance a naila scratchere reversed , b s a -s y useit d ma r san ,o ; cleaner it may have been attached to the girdle by a leather strap, a small portio f whico n h still remaine forketh n i ds head. Toilet instruments of this nature, found either singly or associated on a ring wit n ear-picha tweezersd kan e knowe Continenar th , n no t froe mth later Hallstat or early La Tene periods.4 Objects of bronze with somewhat similarly notched ends, describe s nail-cleanersda , were founy b d General Pitt-Rivers in his excavations at Rotherley, Bokerley, and on Handley Hill and Handley Down;6 and a similar use may conceivably be assigned to a small iron object with a bifurcated termination, found in

1 A Roman Frontier Post, plate Ixxxii. figs. 6 and 8. 2 Buske-Fox, Excavations t Wroxeter,a Shropshire,. 1913,n i 14 . p " Opuscula Oscarl Montelio, VorgeschicJitlichene "Di Burgwalle Sclnveden,n i . 342p , fig. .20 43 Dechelette, Manuel d'Archeologie, " Archeologie Celtique," p. 880. 5 Op. cit,, vol. ii., plate cii. fig. 7: plate ciii. fig. 8; vol. iii., plate clxxiii. fig. 8; vol. iv., plate 258, fig. 10. 4 17 PROCEEDING E SOCIETYTH F O S , FEBRUAR , 1915Y9 . e Bustoth n Crannog d illustratean , Dy db r Munr Ancientn i o Scottish Lake Divellings, p. 227, fig. 240. The dot and circle ornamentation is commo objectn e lateo nseee th b n f rni y o sIro e periodma nAg d an , a pai f tweezern o ro e us s from Newstead, combn o d san of bone from

brochs t continue,i etc.d employee an ,b o dt mediaevan di 1 l times. This igracefua s l object, mad f bronzeo e whic takes a fin hha n e n o patina,

Fig. 26. Strap terminal, Awl, portion of a Clasp, Rings, and Tweezers. (}.) in this last respect differing from the majority of the bronze articles e sitefounth .n o d Fro e sammth ee las th levet s cama l e objecth e t representey b d pierce. a 3 Is t. i No d, figplat26 . f s bronzebeeeo ha nd one-halan , a f o f hinged clasp probably fastened to leather. The design is poor and has been produced by a drill, the longer piercings being made by three overlapping applications. Clasps of this nature are not uncommon on late Roma rula ns ea sites t thebu ,y show more artistic merit than this comparee b ony e ma does dt I witfoune . on ht Kastel a d l Zugmantel

on the German Limes.2 Frontier1 Post, plate xcii. fig. 8. 2 Saalburg Jahrbuch, 1910 (Kastell Zugmantel), Taf. viii. ACCOUN EXCAVATIONF TO N TRAPRAISO N LAW5 17 . paie f tweezerTh ro f bronzo s ) cam6 . e e(fig No fro , e highes.m26 th t level. Such article f vero ye sar commo n occurrenc Romano-Britisn eo h sites, and were probably employed for depilatory purposes. Two pairs were found at Newstead.1 They were likewise frequently found by General Pitt-River s excavations;hi n i s 2 they have also been f oxint da Knap HilCasterled an l y Camp Wiltshire.n i s largn i d e an number n i s

the Grlastonbury Lake Dwelling3 . Another personal reli f bronzco a spira s ei l finger-rin} 5 . g (figNo , .26 consisting of two and a half coils of a thin plano-convex strip of metal, taperin gwherd notches markedli en t e ei on do yt alon e edgegth n I . diameter it measures f-f inch; it came from the lowest level on Area A. Such finger-rings are by no means uncommon on Iron Age sites of this period. One of a like number of coils was found in the fort at Castle Law, Abernethy; * another was found in the Hyndford crannog; an caire Iro e dthira th nAg e nn d i th excavate t a year w o fe ag sa d Black Rock t Gullane.a s rina s f bronzi g o 4 5a differen. f FigeNo o , 26 . t type, which also came fro lowese mth t measureI Aretn levelo . t aB bu , s yj inc interion hi r diamete f inc d breadthn hi an r s ornamentei t i ; t da each edge with a plain half-round moulding. Personal ornament materian si l other than bronze consiste f beadsdo , armlets, and rings in jet and glass, or kindred material. Dealing with e formeth r first have w , e segment f eleveo s n different armlets, mostly from the lowest level, shown, with one exception, in the accompanying illustratio segmene nTh t figillustrate. no t27 . s beedha n split t cami ; e from the upper level. As will be observed, these armlets are plain, with the exception of No , whic2 . s ornamentei h d with incised parallel lines followine th g circumference crosd an , s segmente cutsTh . s vary considerabl thickn yi - ness, from the fragile specimen shown as No. 6, which must have been very light when complete, to the rather coarse-looking piece, No. 11. The complete ring and half of another (fig. 27, Nos. 5 and 8) both came from the upper level; "while the small ring (fig. 27, No. 3), with an interior diamete f |fo r inch, came fro bottome m alsd th portioe di oth s a a ,f no smal e segmentTh l . bea1) ,d. amountin(figNo , .27 o almosgt t one-half of a broad ring of shale (fig. 27, No. 15), came from the lowest level on Area C. Fig. 28, No. 1, is an oblong bead of polished jet, faceted on the upper surface, bevelled slightl e ends pierceth d t an ,y a d witlongio htw -

Roman1A Frontier Post, plate xcii. 8 .d figsan 6 . 2 Pitt-Rivers, op. cit., vol. i., plate xvi. figs. 13, 17, 18, and 19; vol. iii., plate clxxxii. fig. 8. 3 Wiltshire Archceological and Natural History Magazine, vol. xxxvii. p. 63, fig. 11; and vol. xxxviii., plate i. fig. 7. 4 Proceedings, vol. xxxiii. p. 31, fig. 16. 5 Proceedings, vol. xlii. p. 334, fig. 2. 6 17 PROCEEDING E SOCIETYTH F O S , FEBRUAR , 1915Y9 . tudinal perforation ensuro st s remainineit correca n gi t position when worn t I measure. s yf inc lengtn hi f inc y breadthn hb i camd an , e

3 INCHES

FigSegment. 27 . f Armletso Ringd san f Jetso .

from the lowest level. The large pin-head of jet (fig. 28, No. 2), shown to natural size, came from the upper level. The pin on which it was set, judging s frosocket it e siz mf morth s eo ,wa e probabl bonf yo e than of metal verA . y similar pin-hea e excavatios founth dwa n e i d th f no ACCOUNT OF EXCAVATIONS ON TRAPRAIN LAW. 177 Mote of Mark reported by me in last year's Proceedings. Such pins were probably worn in the hair, and were the lineal successors of the large spherica headel- d pin f Bronzo s e timeAg e se Continentfounth n o d . Owing to the size of the heads they were unsuitable for use in a garment. There are in the collection fragments of twenty-one different armlets of glass (fig. 29), some of them unfortunately very small. Of these, eleven are self-coloured, six of them yellow and opaque, and five of them opaque white yelloe Th .w pieces, witexceptioe hon n which came from secone th d level, were bottoe foun e excavationth th t da f m o . Thee yar rather thicker and more solid than the white bracelets, and in section e triangularar e describeb y a e canary-yellowtin th s ma :ta d e Th .

Fig. 28. Faceted Bead and Pin-head of Jet. (-J-.) white bracelets, for the most part plano-convex in section, came from l threal ee ornamente levelsth f O . d bracelet e mosth s t common type shows a series of attenuated scrolls pointing along the mesial line, of a different colour fro e groundmth d producean , spoty db f coloureo s d glass let into the mass of metal and elongated in the formation of the , fro 2 e uppe . mth armlet No r n levelI . e bodth , y coloue palesth s i rt possible tint of green, in which is incorporated a scroll of thin opaque als, 1 . o whitefro No e uppe mn th I . r level e grounth , s opaqudi e white and the scroll brown. In No. 4, from the second level, a bluish-tinted body bears pale yellow streaks. The small fragment shown in No. 6 a rathe f io s r intense blue with bright yellow s als i lines od froan , m the second level. No. 9, from the lowest level, has a greenish body decorated with yellow streaks segmenA . f greenish-tinteo t d glass, from the second level, is ornamented with white opaque streaks, and a small piec whita f eo e opaque armlet show e remainsth f bluso e linen so sidese th addition I foregoin.e th o nt g ther , fro eis secone mth d level, 2 1 VOL. XLIX. 178 PROCEEDING E SOCIETYTH F O S , FEBRUAR , 19159 Y . a portion of an armlet (No. 10) of greenish glass with a cord ornament formed of intertwisted blue and white strands running along the mesial ridge of its convex surface, and along each side; and a segment of very unusual occurrence (No. 8) of schmelze glass of a general purplish tint

Fig. 29. Segments of Glass Armlets. found on the lowest level. Portions of armlets such as the foregoing, with the exception of the last, freqiiently come to light in excavations of Roma nativd nan e site Scotlandn i s . Birrenswark yielde dnumbera , in colour opaque yellow whited an als,d an o, white with scroll pattern. Newstead produced specimens; pieces were found in the crannogs of Hynd-

ford, and Dowalton, in one of the Archerfield caves near North Berwick,

2 3

1 1 Proceedings, vol. xxxiii 383. p . . 2 Munro, Ancient Scottish Lake Dwellings, p. 48. 3 - Proceedings, vol. xlili . 254p . . ACCOUNT OF EXCAVATIONS ON TRAPBAIN LAW. 179 and also in the Borness cave. I have, however, found no record of the findin pieca f f schmelzo go e e glass armlet, though certain large beads made somewhat in this fashion have been noted. Portions of two such beads found associated inter alia with a Roman onyx intaglio are among the relics from the cairn at Cairnhill, Monquhitter, in Aberdeenshire,1 now in the National Collection. Another reli f glaso c s from Traprain t obviouslno , r personafo y l adornment, likewise finds an analogy among the objects from Mon- quhitter. This is a small imperforate ball of clear greenish glass (fig. 30), measuring if inch in diameter and ornamented with discs of red and white opaque vitreous paste, six in number. It was found e highesth t a Monquhittee t th level n I . r cairn there were founo dtw

small ball f greeo s n glass2 abou e siz f th smalto e l marbles, measuring in diameter | inch and f inch respectively, ornamented with rows of

Fig. Imperforat80 , e Ball of Glass. (\.) spirals inlaid in -white. From the Buston crannog, from which also came a fragmen f Samiao t n -ware, ther s recoverewa e a round" d objecf o t the size of a small marble, made of vitreous paste, variegated with blue whited imperforatd an an , Mote th f Marken o i e d ";, an 3nea r Dalbeattie, ther founs ewa disa d f glaso c s inlaid with spot f opaquo s e glass.

Very few beads have been found thus4 far. A small bead of opaque yellow glass, discoid in shape, ^4- inch in diameter, came from the lowest level. Anothe f brigho r t green opaque glass, also flattenen o d both ends and measuring J-% inch in diameter, was found on the upper level. A tiny glass bead of lavender blue colour, only ^ inch in diameter e secon s th foun wa n ,o d da smal level d lan ; doublr o e segmented bea f bludo e glass, ^ inc e lengthn hi th pickes n o wa ,p du surface e lowesTh . t level yielde n additioi d a nfragmen larga f o et bea f amberdo . III. FRAGMENTS OP G-JDASS VESSELS. A numbe f additionao r l object f glas o se portionar s glasf so s vessels, presumably Roman. They are small fragments, too small to afford

1 Proceedings, vol. xxxvi . 679p . , fig. 6 . 2 Proceedings, vol. xxxvi. p. 678, figs. 4 and 5. 3 Ancient Scottish Lake Dwellings, . 233p . 4 Proceedings, vol. xlviii. p. 156, fig. 17, No. 13. 180 PROCEEDING E SOCIETYTH F O S , FEBRUAR , 1915Y9 . y indicatioan e dimensionth f o n r e shapevesselo sth f o whico t ss h they have belonged, except in regard to a number of pieces of rather thick transparent greenish glass which have clearly belonge largo t d e squar r cylindricaeo l bottles with reeded handles, suc s supplha y fairly plentiful remains on Roman sites. The other fragments show that there were a number of delicately tinted fragile vessels in use during e variouth s period f occupationo s n shadei , f greeno s , yellow, amber, and blue. One or two pieces are ornamented with zones of finely scratche amber-tintee d on lines d an , d fragment show opaqun a s e spot of darker colour formin ga swellin e glass th smalA n . go l triangular fragment which came from the lowest level is an example of Roman painted glass banA .f pal do e blue runs alon e uppegth r edge, while beneat n shori t i ht parallel lines extend groua s f narropo w bandf o s alternating light orange and dull red. One other piece of glass which came fro lowese mth t leve vers i l y puzzlingsmala s i lt I segmen. f o t the side and bottom of a vase, or bottle, the sides of which have curved rapidly inwards fro e bottom me glas th s bode a th s f si Th y o . bright as crystal, and it is "flashed" or stained on the outer surface wit hbrilliana t ruby colour e absolutTh . e colourlessnes e bodth yf o s causes one to regard the antiquity of this relic with great suspicion^ as much as does its ruby-tinted exterior. Especially is this the case since authoritativels i t i y stated coloure 1th thaf o stransparene ta on s wa d re t whic e Romanth h s were never abl obtaino e t inventio e Th . f rubno y glass such as this, is attributed to Johann Kunckel at Potsdam in 1679.2 How, if this fragment is really modern, it should have found its way not onl Trapraip u y y halwa fn Law t als,bu o down through somr 3-o 2 e feet of soil, is a mystery.

IV. HARNESS MOUNTINGS. s assuredli t I y evidenc e extenth f o whict o et h wheeled vehicles amone werus native n egi th e populatio e occupae perioth th f t no a d - tio f thio n s site thae shoulw t d have lowesth ed n foundan o t l al , second levels leso n , s than five terret ring f bronzo s r reinsefo , thref eo which are illustrated (fig. 31, Nos. 1 and 2, fig. 32, No. 1). It is probable that they were use paie n pairsi d rOn fro . e lowesmth t level appear to have been identical. The others, fig. 31, Nos. 1 and 2, from the second level, and fig. 32, No. 1, from the lowest level, are all slightly different, though the general fashion of the whole five is the same. Fig. 31, No. 2 doubla s ha extr n ea 1 flang. s aapexe ha No figt bosd th , a e t an ,.sa 31 the base, where the terret was presumably inserted into the leather on 1 Edward Dillon, Glass,. 52 . p 2 Ibid., . 290p . ACCOUN F EXCAVATIONO T TRAPRAIN O S N LAW. 181 s smalli pade 1 th . ;, No figmeasurin , .32 g onl inc1 y interion y hi Ib f r diameter. Such objects are not infrequently discovered on Celtic sites. The hoard of late Celtic harness mountings from Middlebie in Annandale, preserve e Nationath n i d l Collection, contain f theo s mo fivetw , ornamented with three spherical bosses each, in the same manner as those from Traprain, whil a ethir s foudha r truncated f bosseso o tw , whic neae har r togethe e apeth xt a rthoug t touchinghno fign , i s .31 a , No. 2. The remaining pair are plain. One example was found at Newstead,1 but in place of bosses it is ornamented with lip ornaments,

. TerretFig31 . f Bronzeso . (§.) terrete a srice th alsth h e n oshoarar i f Celtido c horse trappings found at Stanwick in Yorkshire, and now in the British Museum. A coi f Trajao n n e seconfounth n o dd level adjacen e terreth o t tt giveshow1 . fore th s No m fign second-centura n, i .31 y attribution. , show6 . n objeca sNo , Fig32 t . forme f threo d e fixed ring alignn si - ment, also a horse trapping and apparently a simpler and less ornate variety of two pieces of the Middlebie hoard, in which the rings were ova crossed lan d each wit hsmala l ornamental bar. Several broken rings appear to have belonged to other articles of the same kind. a quatrefoi , 2 . No f bronz , o l Fig32 . e wit e centr hbosa th d n si ean a square loop at the back for the attachment of a strap, is evidently a variety of an object found at Newstead and elsewhere, in which each

leaf of the quatrefoil bears a circular bos2 s in the centre. A slight ridge appears in place of these bosses on each leaf of the Traprain example. 1 A Roman Frontier Post, pi. Ixxv. fig. 2. 2 Ibid., . 302p , pi. Ixxv. fig. 3 . 2 18 PROCEEDING E SOCIETYTH F O S , FEBRUAR , 1915Y9 .

Its find-spot on the lowest level does not indicate that it belongs to a later period tha Newsteae nth d specimen. The small six-petalled rosette of bronze shown in fig. 32, No. 4 is a

Fig. 32. Harness Mountings of Bronze. ({.)

beautiful piec f workmanshipo e d caman , e froe leve, justh m1A lt above the lowest. It is perforated in the centre, while deep grooves, such as might be intended to hold a thread or light cord, separate the petals. Another object of bronze (fig. 32, No. 3) from the lowest level is probably also a harness mounting; it is an oblong plate with a rect- angula e rbac an th loo o k t moder e a psuggestiv de e n th saddle f o e . ACCOUN F EXCAVATIONO T TRAPRAIN O S N 3 LAW18 . The large hemispherical stud, fig. 32, No. 5, with a point projecting from the centre, was perhaps also used to ornament harness. It likewise came from the lowest level and from the same area as most of the other harness mountings. . WEAPONSV . The only weapons which o fars , , have been foun e severaar d l iron spear-heads, a blade of the same metal which appears to be that of a dagger (fig , Nos.33 portioa . e blad1-5)a d Bronzth f an f ,o e o n Ag e dagger. The spears are of two types—broad leaf-shaped, with closed sockets, as shown in Nos. 1, 2, and 4; arid narrow, with split sockets, as No. 3. The former came from the lowest and second levels and represent the native Celtic type; while the latter, found on and adjacen e highesth o t t t level, thougd preservationba n i h , seem more to have resembled some of the Roman spear-heads from Newstead, or the later Saxon type. No. 1, measuring 8 inches in length, has a well- defined midrib running the length of the blade, seemingly a Bronze Age tradition a featur d an , e which frequently characterise speare th sd an - lance-head a Tene.L f o s1 Onl yfragmena suce on h f spear-heao t s dwa found at Newstead,2 a surface-find and badly rusted. A corresponding e examplriveth f rs foun o Churn wa ed dbe mane , th yn i year o ag s near Cricklade, Wilts, with Roman coins dating from Agrippa to Carausius. , measurin4 . No 3 | incheg7 lengthn i s , show a spear-heas d of simila re last without d flath for an bu ,t o mt e midrib th t . This si doubtless also a native weapon and may be compared with a spear- head found at Hod Hill, Dorsetshire, along with Roman coins, the latest of whicdenariua s hwa f Trajan.so 4 Lik Trapraie eth n spear-headse ,th latte alsd ha ro a close e typth edf o wit sockete h On spli. t socketf o , whic wero htw e found secon A s illustratei , . 3 d. spear-heaNo n di d with a midri bs found (Nopreservationd t wa imperfec ba ) bu .,2 n i d an t . From the second level, and adjacent to the spear-head No. 4, came a long, flat, narrow blade (No 10) .5 | inche lengthn si , which seem havo st e been eithe a dagger a lance-head a r projectio o e butrd th en tt A .n e nec th thi o kr to whicnfo s appeari partanga t e hi b f wouls o to a s,t d have connected a socket with the blade. This fact weighs against the suggestion tha objece tth bees ha tn lance-heada othee , th but n r o , hand, though as a dagger it would have been a very serviceable weapon, it does t appeano confor o t rrecognisey an o mt d double-edgeds typei t I . . 1 Munro, Lake Divellings of Europe, p. 285. Dechelette, Manuel d'Archeologie, " Archeologie Celtique ou Protohistorique," p. 1146.

2 A Roman Frontier Post, pi. xxxvii. fig. 10. 3 Proceedings of the Society of Antiquaries, Second Series, iii. p. 67.

Roach Smith, Collectanea Antiqua, vol. vi., pi. .xi 4 184 PROCEEDINGS OF THE SOCIETY, FEBRUARY 9, 1915.

4 INCHES

Fig. Spear-liead33 . Daggea d san r Blad Ironf eo . ACCOUN EXCAVATIONF TO TRAPRAIN SO N LAW. 185 The portio a a Bronznarro f f dagge e no o e poin wth d Ag e s en i tr double-edged blade, measuring 4| inches in length. Like the piece of a bronze penannular armlet, it came from the lowest level in Area B. Fro uppee mth rsecone parth f o td level cam iroe eth n relic illustrated whic, 2 . s evidentlhi No , tigs a 34 . ya swor d chap a recognise f o e d Celtic form; others similar but in bronze were found at Newstead.

The curved object of iron, perforated in the centre (fig. 1 34, No. 1), may also be a sword mounting, and possibly may have formed an upper guard on which a pommel rested; the perforation is too small for any other part of the tang than the end. Fig. 34, Nos. 3 and 4, represent

Fig. 34. Object of uncertain use, Sword Chape of Iron, and pieces of Bronze Binding. (§.) portions of bronze binding, probably for sheaths, of "which a number of pieces were found in different areas. Similar binding was found at Newstead. . TOOL VI IMPLEMENTSD SAN . e Olowesnth t leve s foundwa l e iro"th n axe-hammer, 6 . , No fig, 35 . also, like the spear-heads mentioned above, of a La Tene type, and

differing considerably from the Roman axes found at Newstead, whic2 h are larger and heavier and have the butt less pronouncedly fashioned as a hammer. Both the cutting edge and the hammer end have been subjecte muco dt h usejudgo t , theiy eb r appearance obliqun A . e perforation from the side of the butt to the socket shows that a nail has been used to fasten the head to the shaft. The second level yielde dstila l more assure a TendL e typ f tooeo n i l 1 A Roman Frontier Post, pi. xxxv. figs. 16-18. . DecheletteCf , Manuel d'ArcMologie, "Areheologie Celtiqu Protohistorlque,u o e . 1358"p , 2 fig. 594. 6 18 PROCEEDING E SOCIETYTH F O S , FEBRUAR , 19159 Y . e e shapsmalth th f elo axe, whic3 . ,n plac i hNo a sockefig , f eo .35 s i t furnished wito correspondinhtw gface wingon e onlyn o s , suggesting, lattee th n ri t respecbu t differing from Bronze th formtyp,e e Th eAg . Tena belongL e e th thir o st d period, and, thoug hspecimeknoI o n f wo n

i i i 4 INCHES Fig. Tool 35 Implementd . an s f Ironso . previously found in Scotland, examples do not appear to be so uncommon e Continent.onth 1 Fig. 35, No. 4 shows an article from the second level, probably a pair of pincers use o holt d d some small delicate e procesobjecth n i f t so D^chelette (op. cit., p. 1359) illustrates several varieties, of which fig. 595, 2, from Gurina. (Carinthia)1 forn i , m most closely resembles that from Traprain, thoug mucsizs n i hi t ei h larger. ACCOUNT OF EXCAVATIONS ON TRAPKAIN LAW. 187 manufacture. When found, one leg showed a recent fracture, and a careful e vicinitsearcth n i h y shortly afterwards secure e pointedth d end showillustratiothe nin completinnas absolutelynot leggam the I . certain that it is the missing portion, as the metal has been so much corroded at the point of fracture that it is difficult to fit the two parts together satisfactorily, but I incline to the belief that it is. From the sharpness of this point and the slightness of the implement it was at first conjectured tha objecpaie a f compassestth s o r wa tfurthe n o t bu r, consideration the conclusion was reached that such an ornamental head would not be suitable in compasses, and no analogy for such a head could be found moreo N . , however, hav eI foun y similaan d r pai f pincerso r . The article shown as fig. 35. No. 1 is one leg of a pair of shears of a well-known type. 3 Fig. 35, No. 2 shows an iron tool 5f T inches in length, square in sectio curvind n an a point o gt , possibl yburnishea r s usesuci y s db ha silversmiths at the present day. Fig. 35, No. 5 appears also to be a tool, 6J inche extremn i s e length -section i , n squar witd ean ha rathe r heavy s perfectli d en tange ye objec blunTh th .lookd f i t an t s coula s d only have been largusea s a de punchfinn similao n ca d I . r tool recorded. There were found the remains of such another, rather shorter and much corroded. Three knives (fig hav) .36 e been foun l wital d h e bladesamth f eo s shape — broad and symmetrically pointed. One of them, No. 1, is supplied with a handle of bronze divided into two longitudinal quasi- cylindrical sections; a double sharp-edged moulding separates the two sections, and a cord moulding forms a border on either edge. In the base of the handle two sockets show that a finial has been affixed with pins. The blade is distinctly spatulate in form and at once suggests a flaying knife; originally it may have had a pointed termination similar to the two other specimens. In the rust and corrosion which cover it are impressions of the straw in which it must have lain when it became rusted and which probably covered the floor of a dwelling. It was e seconfounth n o dd leve f Areo l immediatelaB y adjacen large th eo t t rectangular hearth. The two other examples, Nos. 2 and 3, have handles of iron, continuations of the blade, and neither gives any indication, such as rivet holes would afford, that they were covered by sheathings of bon r wooo e facilitato dt e gripl threth eAl . e knives hav a edistinc t family resemblance. Their respective lengths are : No. 1, 6TV inches; No. 3 2, 5-pg- inches , 5f3 f. f inchesals2 No ;. o No cam. e fro e seconmth d level; t recordedno s type wa Th efind-spoe 3 . doe . t th appeae no No s b f o to rt common, the usual knife of the period having a curving edge, a blade narro risind e backwpoina th an knifo A gn t . o f tapproximatel o e y 188 PROCEEDINGS OF THE SOCIETY, FEBRUARY 9, 1915.

similar form, though larger e Gallo-Romafouns th wa ,n di n cemeteries of Vermand of a date not earlier than the third century. The shape

of the blade 1 is the same, the handle, however, is of bone and circular in section ornamenteds i t bu , fashioa n i , n resemblin e bronzgth e handlf eo e Trapraith n knife, with incised lines running longitudinally down

4 INCHES I I I I Fig. 36. Knives of Iron—No. 1 with a Bronze Handle. centree th . Knivee samth f eo s for e illustratemar catalogue th n di f eo the Niessen Collection of Roman Antiquities at Cologne.2 Another relic (fig. 37, No. 1) is a lancet of bronze. It is a delicate object a razoro ligh r to a fin, fo t t i stils e;coveres i ha edgl d an ed with a beautiful green patina. The tang shows that originally it has been furnished with a handle. It is from the lowest level. e objecTh t whic illustrate2 . h No fig , s bee.37 ha s n formed froma strip of iron twisted spirally to form a socket, and so as to leave a point projecting. A similar article was found at Newstead and is here 1 Theophil Eck, Les deux cimetieres Gallo-Romains de Vermand, . 196p pi, .. 15 xii. .No 2 Niessen, Beschreibung Romischer Altertiiiner, vol. i. p. 229, vol. ii. Taf. cxxx. ACCOUNT OF EXCAVATIONS ON TRAPRAIN LAW. 189 illustrated (fig. 37, No. 3) for the first time for comparison. Though 110 other examples appear to have been recorded from Scotland, they are not uncommo latn no e Celtic sites Generad an , l Pitt-Rivers suggesteo wh , d that they were ox-goads, record e findinth s f severao g t Woodcutsa l , Rotherley d Woodyates,an , o lesn 1s tha x havinsi n g been fount a d e last-mentioneth s discoveredwa e recene placeth On n i dt. excavatio f . Casterleo CunningtonnH . B s y Mr Camp .d an , r Wilts,'M y b 2 Noe thear r y unknow e Continent th s illustratei n e o n On . d among

Fig. 37. Lancet of Bronze and Ox-goads (?) of Iron. (^.)

e relicth s from Kastel Faimingen e Germath n o 3n s i Limes d an , merely describe a ferrule s a d , -while Lindenschmit4 illustrates similar objects from the Rhine and elsewhere, designating them as arrow- points. e sharpeninar 8 d , NosFigan 39 g. 7 .stone f sandstoneo s , eac| 3 h inche s neatli lengthn i se yOn . bevelle t bota d e otherh th ends d , an , whic perforateds hi roundes i , subovaa o dt l form portioA . f anotheno r sharpening ston bevelled e en wit s alse hth dwa o founde s welth a ,s a l large r heaviea hal f o f r whetstone, oblon section gi perforated nan t da e endon .

1 Excavations in Cranborne Chase, vol. i., pi. xxix. fig. 10; vol. ii., pi. cv. fig. 10; vol. iii., pi. clxxxiii. figs. 17-19. 2 Wiltshire Archaiological and Natural History Magazine, vol. xxxviii., pi. ix. fig. 10. 3 Der Obergermanisch-Raetische Limes, ffr. 66c, Kastell Faimingen, Taf . fig. .ix .9 4 Die Alterthiimer unserer heidnischen Vorzeit, vol. iv., Heft viii., Taf. 4.

ACCOUNT OF EXCAVATIONS ON TRAPRAIN LAW. 191

VII. WHORL DiscD STONEF SAN sO . Connected with the industry of the women of the place are whorls to weight the spindle in the spinning of thread ; of these there are fourteen, severa f whico l e illustratehar . fign dNosi 38 . f bake.o 12-1e dar 4 clay and came from the lowest level. No. 11 is one of two of identical size forme f whic o f lead o de hon , came fro othee e seconth mth d r frodan m uppee th fashiones i r4 level. No .d fro base a Samiam th f eo n ware vase, and was found on the second level. No. 6 shows a fractured specimen, the perforatio f whicho n , started from both sides neves ha , r been completed. t camI e fro secone mth d level dis A f sandstone.o c perfeca , t circld ean highly polished . Sligh9 . s showi , No t , fign ridgeni 38 . s nea centre rth e indicate s beewherha n t i ea polisheruse s a d . Similar thin highly- polished discs were found by Sir Francis Tress Barry in the Skirsa broch Roae th dd e Nationa brocan th n t i Keissha w l Collectionno , arie dar . Fig. 38, Nos. 7 and 8 are small circular discs of sandstone 1 inch and 1\ inches-in diameter, probably r somplayinfo en sucgme h gams a e

draughtsT , and resembling in size the numerous discs of coloured glass which were found at Newstead. Fig. 38, No. 10 is a thinner disc of similar sizd probablan e e samy th use r e dfo purpose e discTh ,. , fig38 . No. 5, with a diameter of If inches, is of nearly double the diameter and thicknes e foregoingth f so s purpos it t ; obviousno s i e , thougy ma t hi have been a whorl in the initiatory stage of manufacture.

VIII. OBJECTS OF FLINT. Three scrapere f ar whico ) f flint3 o o s htw d , , (figan Nos 39 1 . . illustrated, were found severad an , l large flakes "which, fro e naturmth e of the chipping on their edges, had evidently been used as strike-a-lights, a purpos r whicefo demann hi s flin wa protohistorin tdi c times.

. MOULDIX CRUCIBLESD SAN . A numbe f mouldo r r castinfo s g object f bronzo s e were recovered, fashioned of baked clay and of stone. The former were all imperfect. They seem for the most part to have been for pins; fig. 39, Nos. 4 and 6 have been for pins of the hand type; and fig. 39, No. 2, which is very incomplete, 5 seem. No . havo , s3 t Fig lik n . 39 .pi e e No beea , fig r 25 .n fo has been for some circular ornament, possibly a flat ring. In all there •were found twelve pieces of such moulds. With one exception (fig. 39, No. 6), which came from the upper level, they came either from the lowest or second level. The other moulds, all cut out of sandstone, are six in number,

ACCOUNT OF EXCAVATIONS ON TRAPRAIN LAW. 198 there being in one case two on one block, on the upper and under surfac e stoneth e f o srespectively e camon et froNo e lowesm.th t froo e uppe leveltw mth thre d t an rhu ,e fro e seconmth d level. Three r castinfo e e mouldar goth f ) narro5 d s (figan w , Nos, 4 bar .40 , 2 .r o s

4 INCHES

Fig. 40. Moulds of Stone and rough Casting of Bronze.

ingot f bronzeo s , measuring respectivel f inches3 y , 3£ inches d 3^an , inche n lengthi s . Similar moulds have been founa numbe n o d f o r sites. Several wer ee Society'founth n i d s excavatio t a Dunadd,n

othere Brocth f n Harrasi o h Orkneyn yi e Lochleth n i , e crannon g(a 1 object from the last named, illustrated as a " hone," '2 being obviously a mould) d othean , r sources n objec A illustrate) 3 . . t No (fig , - d.40 im mediatel a roug s yi , beloh2 . castinwNo f bronzgo e which presenta s resemblanc closo s e e matri e n moulsizd shapi eth th an e f o t eo dx 1 Proceedings, vol. xxxix . 292p . . 2 Munro, Ancient Scottish Lake Dwellings, . 105p , fig. .54 VOL. XLIX. 13 4 19 PROCEEDING E SOCIETYTH F O S , FEBRUAR , 1915Y9 .

show nwarrano t abov s a t ea i conjecturt e tha t i actuallt y camt eou of it. The fact that the ingot came from the lowest level and the mould from the second does not necessarily destroy such a presump- tion, for there is nothing improbable in such a mould of stone surviving over a number of years.

0 A- INCHES

Mouldo Tw Figsam. n o s 41 . e Stone.

Two moulds which appear upon the one stone (fig. 41) are of more interest than the foregoing, as their purpose is less obvious. One is for a harp-shaped block of metal 3| inches in length by If inch at greatest breadth by f inch in depth. It suggests the outline of a large t i fibulaseem t bu s, hardly probable, whe fine nw d clay moulds for pins, tha a nearet r approximatio e a forfibul th f m o o a t n could not have been obtained previous to the application of the graving tool. This mould came from the uppermost level. The second object to be ACCOUNT OF EXCAVATIONS ON TRAPRAIN LAW. 195 cast from, the matrix on the opposite surface of the stone is equally inexplicable. It has been 33V inches in length, f inch in breadth at

Fig . Hala Crucible f 42 .o f . (§.) the one end, expanding to lf% inch at the other, which is oblique, and it curves slightly in the direction of its length. The remaining mould (fig. 40, No. 1), from the second level, was for a disc |£ inch in diameter ! fd ¥ an inc thicknessn hi . A numbe f fragmento r f crucibleo s s were samth e f efouno l al d class, cup-shaped wit n ovoia h d base, formed from fireclaye Th . largest portion of one of these vessels is illustrated in fig. 42.

. MISCELLANEOUX S OBJECTS. Another reli f stoneco n ovaa ,s li fig, pebble.43 , 3ff inche lengtn si h by 3| inches in breadth, which has been hollowed out to serve as a

SINS.

Fig. 43. Stone Lamp. lamp. The hollow is still stained black by the fat, or oil, which was consumed in it. Such lamps are not uncommon on Iron Age sites. 6 19 PROCEEDING E SOCIETYTH F SO , FEBRUAR , 1915Y9 .

An almost identical specimen came fro e Keismth s broc Caithnesn hi s excavated by Sir Francis Tress Barry. Two small balls of stone, £f- inch and 1J inch in diameter, were found, objects presumably used as sling stones; and a pellet of baked clay, ^r inch in diameter, was possibly used for the same purpose. Object f bono s e were rare decorate, 3 .. FigNo , d44 . with vertical rows of incised chevrons, is probably the end of a handle. Fig. 44, paia f f exactlo o r e on y s No i simila 2 . r object f unknowo s n use, each terminatin bluna n gi t poin broked an t n throug perforationa t a h t i ; is possible they may have been employed as dress fasteners. Both came fro lowese mth t level. There were found alsportiono otw cylindricaf so l object f boneo s , f inc e morlengthn hi th d e an complet, ) 5 . e (figNo , .44 | inch in greatest diameter. These may have been worn as beads. Such ornaments have on several occasions been found, both in this countr e Bordersd soutth yan f o h , associated with Bronz e interAg e - ments.1 These fragments came fro e lowesmth t level. Amon e miscellaneouth g s relic e severaar s l object f d iroo s an n of bronze tha te mos musth f e noticedto b t remarkabl e On .e th f eo finds is an object of iron (fig. 44, No. 1), 3| inches in length, resembling a well-executed model of a deer's horn with three points. If it has t beeno n intende represeno dt t suc objecn ha s difficul i t i t conceivo t t e bees whaha n t ti mean represento t t camI . e fro secone mth d leveln O . the same level, and not far from the large rectangular hearth of Area B, s foun wa e handle Eomath dth f o en patella. 6 show . No n fig i n, .44 It is covered with a beautiful patina and has been ornamented with diapea r patter incisen ni d lines. Fig. 44, No. 7 illustrates a curious little object of bronze which closely resembles a human leg and foot. It is tempting to imagine that we have here a foot-amulet such as was common in Europe in Early Iron Age times,2 but I think that such an assumption would be rash. The bronze is much decayed, and owing to this cause the form it has assumed may be in part fortuitous. On one side also there is a small cup-like depression e purposth , f t whicobviouseo no s hi . show4 . No a smals , Fig44 .l portio a mountin f no f thio g n bronze ornamented along each f edgrepousso ew witro e a hdots ; while fig- 44, Nos. 12 to 15 illustrate four out of five fragments of a mounting of bronze ornamented with moulded lines; three of the pieces still retain studs for attachment. The largest piece is. fashioned on a curve both longitudinal and lateral. Fig. 44, No. 9 is another and heavier piece of a

1 Proceedings, vol. xlii 212. p . . Dechelettee 1Se cit.,. 1306. op ,p .

198 PROCEEDINGS OF THE SOCIETY, FEBRUARY 9, 1915.

mounting or finial of some sort which appears to have been in the form osheatha f l thesAl . e fragments came fro e lowesmth t leyel. A rin f bronzego , imperfect, yf inc interion hi r diameter, segmented e uppeonth r surfac d bevellean e d beneath, 11 s showi ,. No fign n, i .44 and came from the level, only occasionally evident, immediately above bottoe th a solidesignate d s mi an d 0 rin1 . f bronz. go Fig No d1A , .44 e from the lowest level. Fig. 45, No. 9 shows a half horse-shoe which came from the highest level. It has no calkin, and may be compared with a portion of a horse- shoe foun y Generab d l Pitt-Rivers along with e thircoind th f dan o s fourth centurie e ditc th f Wo ho n si r Barron, Handley Down.1 A considerable number of iron nails were found, of which the best preserved example e showar s fign ni . 46addition i ; n there were found o objecttw s know s splina f twhic o pins e s illustratehi on , figy , db .44 No. 8. A long nail (fig. 45, No. 20), shouldered, slightly curved, and furnished with a flat transverse head, came from the bottom level; it measure inche6 s lengthn i s . Other object f iroso n includ hooea . 22)k, No , 4|(fig- 45 inche. lengthn si . Similar objects were found at Newstead and at Woodcuts;2 a quadrangular ring of iron (fig. 45, No. 16) resembling the loop of a f iroo buckle n. r 19) ba (figNo ,, a 9;f . 45 inches long, .slightly curved e directioth s lengthn i it f no , formin a gsquar f inc f sectionn eo i h , worked to a blunt rounded point at one end and checked at the other to the depth of half its thickness for a length of nearly half an inch, found adjacent to the long hearth on the second level of Area C; a 9 thin plat , 2f iroTeo 2) ¥ . ninche (figNo , n lengthi s.45 , formin righa g t angle, pointed at one end and probably a portion of hasp; a washer

of iron (fig. 45, No. 15)g,- l inch in diameter, with a square perforation T

in the centre; an objec3 t of iron of indeterminate use (fig. 45, No. 13), slightly imperfect at one end, 2}f inches in length, forming an ogee curv outlinen ei ;blade symmetricall) parknifa th 1 . f f eo to e No , (fig45 . y pointed, 2| inche n lengti s y fb h inc n breadthi h a too; f iroo l n

. 18)No , (fig45 5T.V inche n lengthi s a squar, f inc f -sectionn o i he , and slightly tapering to one end; another tool-like flat object of iron (fig. 45, No. 17), 51 inches in length, f inch in breadth, £ inch in thickness, rounde e end on a smal ;t a d l iron punc) 3 h. (figNo , .45 2iY& inche lengthn i s a hoo; ^, f iroko 8) inc n. lengthn hi (figNo , .45 ; several iron tangs for tools or implements (fig. 45, Nos. 4, 5, 11, and 12), varyin lengtn gi h from 1 T2o ^t | inches bola ; t wit hheaa d hammered on at either end (fig. 45, No. 21), 2-f inches in length; the remains 'of an

1 Pitt-Rivers . cit.,op , vol . piiv . 238, fig. .24 2 Ibid., pi, .86 xxviiivol. p . i . . fig. .20

200 PROCEEDINGS OF THE SOCIETY, FEBRUARY 9, 1915.

iron key, which has had a fixed ring at the end of the shank (fig. 45, No. 7), 4f inches in length. Fig. 45, No. 6 is an indeterminate object of iron, 3T\ inche witd n lengthi s an hstraigha d ,en pointee t on edg t t a dea

Fig. 46. Iron Nails. vera s ye other ha rud th t I e. appearanc possibly ma tooa e d ylb ean in an incomplete state. n additio I e foregointh o t n g object f metao s l ther s founwa e a d portion, amountin o about g t one-half a perforate f o , d bal f bakeo l d clay, flattened at the two opposing poles and measuring in diameter

2^f inche inchesy 3b s . Objects somewhat similar, believee b o t d F T

loom weightsB , were found near Mountblair, Banffshire d Ravenslyan , , ACCOUNT OF EXCAVATION TRAPRAIN O S N LAW1 20 . Forfarshire, many years agoJ and are now in the National Collection; anothed an founs rwa Kastel n di l Zugmante e Germath n o l n Limes.2 The last of the miscellaneous relics that I need mention is, I am given understando t coprolita , e excrementh f eo larga f o te fish, probablya

flattenea s skatei t I d. sphere ,inc 1diametern hi , wit a hsurfac e T T

white and slightl9 y polished, marked with numerous black horizontal stria whicn ei h appear small gritty particles.

XI. COINS. Noe leasth t t importan identifica e coinse th e relicth r th e fo , f ar so t - tion of which I am indebted to Dr George Macdonald. These are •certainly three in number, while a small disc of bronze may be a fourth completely effaced. From the lowest level of all, from among a number of large stones which were probably the foundations of a hut, came a denarius of Hadrian (A.D. 117-138). One of the small hearths which on Are markeB a e levedth inchew l fe IB a ,s belo we secon th tha f o dt occupation, yielde a ddenariu f Antoninuo s s Pius (A.D. 138-161), while fro e seconmth d level cam ea secon d bras f Trajao s n (A.D. 98-117)e Th . small disc referred to above was found on the highest level; it is not a styca—fo thicko rto thas i ; t possiblti bees ha nRoma a t y i n minim. What deductions may we draw from the foregoing ? We have here & prominent hill eminently suited by its defensible natural features, its moderate height, and grass-covered slopes for the occupation of s, large community in early times, when protection was required not only for the people themselves against enemies human and bestial, t alsr theibu fo o r flock d herds e havan sW e. seen evidenco tw f o e •distinct schemes of defence of this site, one, with its almost obliterated ramparts, presumably much earlier tha e othernth ; each enclosinn ga enceinte nearly half a mile in length, but the later seemingly com- prehending a broader superficial area. Our excavation, which relatively e wholth o et e extenenceintth f o t e cover n insignificana s t plot, lies outside what we have presumed to be one of the ramparts of the older enclave and well within the later additional enclosure on the south- western slope. Five level f occupatioo s n thio n s plot have been •disclosed e earliesTh .f theseo t , froe largth m e proportioe th f o n whole relics whics yieldedha evidentls t hi ,wa mose yth t importantt a , least in duration. Two relics of bronze—the dagger-point and the portion openannulaa f r armlet compann i , r threyo wito e htw flin t scrapers— might indicate Bronzth terminue n i generae th e o th Age t qu a sbu ;l facies of the pottery, and of the numerous other relics referable to an Iron i Proceedings, vols. iii. and ix. pp. 68 and 176. 1 Der Obergermanisch-Raetische Limes, Nr. 8, Kastell Zugmantel, Tafel xvii., fig. 45. 202 PROCEEDING E SOCIETYTH F O S , FEBRUAR , 1915Y9 . Age horizon, point to these Bronze Age deposits being merely fortuitous. The native pottery, in the present stage of our knowledge, is not illumina- ting further -than that it is of Iron Age character as found in southern Scotland. Its marked predominanc lowese th n ei - soie t levelth im l d ,an mediately overlyin , e nativleavegit th doubo o t en s s charactea t f o r that occupation terrae th f O sigillata. whicyieldes ha t hi d almos l thtal e fragments hav epossibla e first-^centur yearln a date e d yb an dat,y ema attributed to some of the sherds of other Roman wares the periods of manufactur f whico e e lesar hs accurately known e fibulseTh . , pins, clasps othed ,an r objects from this leve e suc ar ls hav ha e been founn o d Roman or Romano-British sites, the occupation of which has occurred at secone th n i d firse d centuryth an t f coie o d nOn .en fro e mth this level> identified as a denarius of Hadrian, proves at least that the site was not abandoned before A.D. 117e commencemen yeae th ,th f o r reigns hi f to . Thhazary ema housesn opinioe da on vern f i o n, y slight evidence, appea havo rt e been roughly circular, \vith wall f wattlo s daubd ed an an , the presence of nails implies a certain amount of squared woodwork. There is nothing to guide us as to the date of level 1A; it only made itself apparent to a slight extent, and evidently followed quickly on the termination of the primary occupation. From the level above, however, IB, which lay but a few inches beneath the well-defined second level e denariuth , f Antoninuo s s Pius carrie s welu s l int e seconoth d century and determines the earliest possible date as probably not before A.D. 140. The fragments of terra sigillata from the second level, as well as those of the other Roman wares, have a distinct second-century facies, and clearly refer the occupation of that level to the Antonine period between A.D. 140 and 180. A remarkable lack of relics characterises the uppermost level of all, but the third to fourth century date attributed to the few fragments of Roman pottery recovered opens a vist f furtheo a r suggestive dis- coveries. I may add that so far no site of a Roman fort is known in this region nearer than Inveresk, which is some twenty miles distant. The Societ s mucyi h indebte e Righth o t d Hon . BalfourJ . .A , to whom the hill belongs, for permission to excavate, and also for the disposa e Nationath n i l l Museu e relicth f mso found. One word more remain succese saide b Th o .t s s whic s attendehha d our first summer's excavation must in large measure be attributed to close th e personal supervision exercise Jame r A.. M G . . Creey dJ sb E r ,M Baird, and Mr Keith R. Murray, who each gave several weeks of almost daily attendance on the hill, as well as to the enthusiastic, vigilant, and intelligent manner in which Mr Pringle, our foreman, and his two assistants wore carrieth kt entrustedou themo dt .