EXCAVATION HOWNAT A S M RINGS, ROXBURGHSHIRE3 19 .

VIII.

E EXCAVATIONTH T HOWNAA S M RINGS, ROXBURGHSHIRE, . PIGGOTTM . C 1948Y B ,. F.S.A., F.S.A.ScoT.

CONTENTS. INTRODUCTION : PAGB The Circumstance Excavatiof so n ...... 3 19 . . .19 Th. e 4Selectio . Hownaf . no . m Ring . s The Site before Excavation ...... 196 DESCRIPTION : Summar Resultf yo s ...... 8 19 . The Excavation Described ...... 0 20 . Phase I. The Palisade Enclosure ...... 200 Wale l Th PhasFor . te II ...... 2 20 . Are I ...... a 2 20 . Area II. The Entrance ...... 203 . .20 . 3 ...... Are I aII Phas Rubble e IIITh e. Ramparts .....4 20 . The Inner Rampart ...... 4 20 . The Inner Ditc Berd mhan .....4 20 . 5 20 . . The Secon. d. Rampar . Ditcd an t h Entrance through Second Rampart ....5 20 . The Outer Ramparts ...... 205 Blocking of Phase II Entrance ...... 206 6 20 . . Hut o Storag. d san Tw t . ePi B V I d an PhasA V eI Hut I, Area III .... 206 Storage Pit, Area III . . . . 208 Hut II, Area IV . . . .209 THE FINDS ...... 212 DISCUSSION . . . . . •...... 220 APPENDICES I-IV ...... 223

INTRODUCTION. The Circumstances of Excavation. The excavation t Hownasa m Ring Roxburghshiren si 1 were undertaken, in Jun Juld e1948an f y o behaln Societe o , th f f o Antiquarie f yo Scotlandf so , 1 First describe planned dBordere an th n di Magazine, September 1863, 183—4. "Accounn a f o t Ancient Fortlet near Hownam, Roxburghshire: with illustration," by David Milne Home. VOL. LXXXII. 13 194 PROCEEDING E SOCIETYTH F O S , 1947-48. kiny b d permissio H.Gf n o Duke th . Roxburghef eo tenane th d tan , farmer,- Mr J. Bryce. The excavation lasted a month, and labour was provided by students who were enable receivatteno o dt t d dan e some training through grants froe Scottismth h Universities allottee Scottisth o t d h Field Schoof o l

LOCATION MAP

FOR.

HOWNAM 'RINGS.

HOWNA\\ R.INO OTHERO S . If-ON ACE FOR.TS

Pig. 1.

Archaeology. To these students, to Mr Bryce, and to many other volunteers, notably Pla u Misd t. J sTaylor , Mis . BalfouHr . GilmorG .M M s d rean Paul writee th , greatls i r y indebted t onl writer—foe no y th d an , r Scottish Archaeolog significantls yha y benefited through this voluntary effort under- taken at a time when camping life was extremely difficult, and when the weather was so bad that some 20 per cent, of the digging hours were lost.

The Selection Hownamof Rings. bees ha n t I possibl somn ei e part Britainf so , more especiall southe yth , shoo t w that certain kind hill-forf so t defence equatee b n sca d with different period culturesd san . Thus single ramparts enclosine th ghilltoa n i e par EXCAVATIONS AT HOWNAM RINGS, ROXBURGHSHIRE. 195 south generally foun belon o "AdIrot e e th n"Ag o g t culture , while multiple ramparts are more typical of the "B" culture. A great deal of excavation remains to be done in all parts of the country before the very significant difference workee b n ca sd out therr fo , e were many varied kind defencef so s e invadinth use y db r peacefull go y settling peoples withi e Ironth n Age. Britise parth o Ann f o tn hd i Isle less si s know thesf no e defences than i Southern Scotland, where scientific excavation on hill-forts has been almost non-existent. Carefully planned excavation followed up by extensive field-work by trained archaeologists is needed before the numerous local variations among the hill-forts can be classified, and related by associated find f pottero s d metal-woryan e cultureth o kt f pre post-Romad so an - n times. For whereas in the more Romanised regions of southern Britain there are comparatively few fortified sites of the Dark Ages, in the north generalld an highlane th n yi d part Britaif so n ther knowe mane ar b o nyt Dark Ages sites. Of these, too, almost nothin knowgs i theid nan r character- istic workee b s o havt dt outeye . In "A Survey and Policy of Field Research in the Archaeology of Great Britain" (Council r Britishfo Archaeology, 1948 e followinth ) g sentences occur (p. 105):— "Several forts (Traprain Law, Castle Law, Kaimes Hill—-ail in the Lothians) have been shown by excavation to belong to several periods, each with a distinct system of fortification; and this can sometimes be worked out on the ground without excavation t withouBu . t careful fieldwork, descriptio unexcan a f no - vated fort as ' multivallate' or 'bivallate' is simply misleading. Again, the excavation of Kaimes Hill proved that hut-circles within the enclosure were later than the latest system of ramparts, so that references based on superficial survey e exposurth o t s f suco e h structures withi na for t hav a vere y limited value." Again, on the following page:— . "The very chronological framewor native th r ekfo cultures nort Hadrian'f ho s Wall remains to be built. Hence the prime need is an index-series of type- fossils, based mainly upon stratified pottery. Thin onl e obtaineca sb y y b d further excavation." field-wors a r Asfa concerneds ki fortunats i t i , epresene thath t a t t time Roxburghshir s beinei g studied intensivel e Royath y ylb Commission no Ancient Monuments, and as a result of this work it is possible to distinguish the following types of defended settlements l:— (1) Hilltops enclose palisadesy db , either singl doubler eo . (2) Forts defende simpla y db e sheer-faced wall. (3) Forts with multiple ramparts and ditches. Apparently later tha nsmale thesth e l eenclosurear s know "homes na - Steee 1 Se Archceological e r inTh News Letter,(Decembe8 . No r 1948) . 10-11pp , . 196 PROCEEDINGS OF THE SOCIETY, 1947-48.

steads, d variouan " s other presumably post-Roman structures whico d h not concern us here. Examination on the ground of some of these forts, including Hownam Rings, reveale apparenn da t sequenc whicn ei h multiple ramparts seeme havdo t e been adde defencedo t s wit hsingla e wall. Hownam Rings was clearly a fort which had had a comparatively long life, during which tim e defensivth e e syste d beemha n altere d redesignedan d t I . also included a small "homestead" partly built over the rampart. For thes remover e fa reason chosent s face no wa th alsd s td r t si thao an ,fo wa t ti from the Roman road of Dere Street, running northwards to Newstead; for along this road would have been carried many Roman wares which may have been traded wit e nativhth e people thesd an , e Roman objecte ar s invaluable for the evidence of date which they afford. generallo tw e Oth fy practised method f excavationo s , total stripping of the site, or test excavations by carefully placed cuttings, the second method was that selected for Hownam. In the course of the excavation it was hoped that some at least of the following questions might be answered:— (1) What was the sequence of hill-fort construction in the Border Country ? (2) To what cultures and dates did the different defences belong? (3) Wer e multiplth e e dltched-and-ramparted forte sth maden i s a , south "Be Iroy b , n"Ag settlers gradually moving northwards like their "Ae IroAg n" predecessors t accompanieno f i , Or ? d by settlement, was it only the fashion that spread as the new weapon slingf s o possibl d san y chariot swald madol l e defenceeth s inadequate ?' (4) How were these cultures affected by the Roman conquest of Southern Scotland? Answers were obtained to some of these questions, and will be discussed in the later part of this report.

The Site before Excavation. Hownam Rings lies just withi e 1000-foonth t contou northere th n o r n slopeCheviotse th f so mile5 , s sout Morebattlf ho hald milean a f e easf o t the village of Hownam, where the Kale Water is joined by the Capehope Heatherhopd an e Burns (Grid Referenc e36/791194): . The appearanc e countrth f o e y to-da s thayi f rollino t g treeless hills with rounded profile, covered with poor grass suitable only for sheep grazing. Hill-forts and homesteads are thickly scattered, usually, as shown in fig. 1, between the 800- and 1100-foot contour lines. Round about Hownam greatee Ringth d rsan heigh Hownaf lane o t th fairl s di w myLa well drained HOWNAM RINGS ROXBURGHSHIRE MULTIPLE RUBBLE R.AMPAR.TS t. DITCHES EXCAVATIONS 1948 (PHASE III)

'HOMESTEAD IV B) PHASE IV),V " ,

;\\R.OBBED FORT '\\ (PHASE II)

.-•' .--'DITCHES (PHASE up

PALISADES. ••' (PHASE'S IA&IB)--

0 5 10 IS 20 25 30 5S 40 45 SO 55 60 £5 70 METK.ES C.M.P 'i-S

Fig. 2. Plan showing excavated areas. [Based on a survey by the Royal Commission on Ancient Monuments. 8 19 PROCEEDING E SOCIETYTH F O S , 1947-48.

by the numerous burns which empty into the northward flowing Kale or Bowmont Waters and ultimately into the Tweed. Geologicall Sandstond lowee Re th d n surfaceOl rye o eth geolog vers yi y varied. The lower-lying parts are often sandy, and there is a considerable amoun f alluviao t l soil forme deposity b d s washed down fro e highemth r land. soute Thuth easd n han so t fore sideHownaf th o t f so m Rings the subsoi loams li sandyr yo ,nort e whil westh rocd s i n h induratean er t o kti o d clay n additionI . e largth , e numbe f erraticso r , some probably brought from far afield in the course of glaciation, include grey amethyst, rock crysta quartzd lan , etc. muc w impossibls i ho ht I y woodlansa o et d would have bee thesn no e hill latn si e prehistori probablcs i timest i t bu ,e that ther considerabls ewa y more than to-day, and it is certain that the large number of trees used in the two palisade stockades must have been cut down locally. One can reasonably imagine that a large number of prehistoric sites may have been hidden from woodse vieth y wb . Before excavation began, ground observation showe e followinth d g features. Remain wala rounn f ra o sl de hill th par .f o t Several rubble ramparts were obvious on the west, the most rocky part of the hill, and had either never been completed beed ha n r eas southe ploughed ,o th tan n o ;t dou circular hollow numerouf so s huts were thickly clustered insid defendee eth d area, and several on the north had distinctly been built over or into the decayed ramparts. Lastly, on the east and partly outside the fort was a small "homestead" enclosure containing signs of several huts. This enclosure fro s positiomit n might have been expecte havo t d e beee th n latest feature in the history of the site.1

DESCRIPTION. Summary of the Results Obtained. The sequence discovered as a result of excavation was as follows:— Phase soute /.Th s enclosehe hil th wa sidl f o e d b ya palisad f o e 2 wooden posts which had ,t verafteno ya rlon g life, been replacea y b d similar palisade following nearl same yth e entranclineo N .foun s ewa o dt this enclosure, whic onls hywa traced alon curve boundars gth it f ee o th n yo south side. This earliest phase was not dated. sheer-facePhaseA . II d wall, some 10-12 fee tbase e wid beed th t e,ha a n lonw hille g Ho th thibuilexiste d .f o sha defenca t p s rounda to e s ei d th On the south side of the hill and outside the apparent limit of the fort, a row of some 28 standing- stones1 , known locall "Shearers,e l thath al s ye ta b remain y fiel"a ma f sdo dyke, probably considerably later than the fort. The word "Phase" rather than "Period" is used here advisedly, since the whole occupation of 2 this fort appears to have belonged to one archseological period (the Iron Age, pre- and post-Roman) thoug fouo ht r phases within that period. EXCAVATIONS AT HOWNAM RINGS, ROXBURGHSHIRE. 199 200 PROCEEDINGS OF THE SOCIETY, 1947-48.

not at all clear, but it was evidently considered obsolete in the late first centur t thaa y r tA.D timFo . e (Phase III defencee )th s were modifiedd an , a datable quern was included in the blocking of the original entrance. Phase III. This started with the old wall being reduced in height, and large somth f eeo facing-block s wer einne e usekera th severaf s do o ra bt l rubble ramparts; the reduced wall being incorporated in the innermost of these. As was the case in the south of England, this new idea of defence deptn i moss hwa t probabl offensivyresponsa w ne a eo e t weapon sucs ha sline th chariotr go . Phase IV. The fourth phase began when the defensive life of the fort endn waa t ,s a probabl mid-secone th y yb d century A.D. therr fo , evidencs ei e to suggest that the third phase was of short duration. The hilltop continued to be occupied,"however, and the hut excavated on the north-west was apparently being lived in in the late third century A.D. The other hut on easbelongine d th tan "homesteade th go t " date e coulb accuratelyt o ds d no , though it, too, was built over the obsolete ramparts of Phase III. There was no very noteworthy change in the pottery from all these havphasese reasow o en d supposan o ,n t e that durin centuriee gth hile sth l was inhabite profouny d an ther s ewa d cultural change suc mighs ha t have been expecte f largdi e number settlerf so r invaderso moved sha fron di m thire th df anothe o centuryd en later rd e region e evean ,th d th , y nAn b . influence of Rome on the material culture of the native people was hardly perceptible. The Excavation Described. Phase I. The Palisade Enclosure. Evidence for the presence of this earliest stage at the site was found accidentall Aren yi I onlI a n cuttinI y (fig. , whic3) gF . s designehwa d to strip what apparently was one of the entrances into the wall fort of Phase II, two bedding-trenches, originally holding timber uprights packed round with stones, were discovered and appeared to be contemporary. But as they were subsequently shown to converge (cuttings E and D) and finally apparens joi cuttinn nwa i t i , tgC thaline eon t must have bee nreplacea - s bee e otherha mennth t I f mentioneo .t d that these palisade trenches could not have been co-existent with the wall of Phase II since they ran across the entrance belonging to that phase. They must, then, be either earlie later ro rknoe thaw wt n fro PhasBu m . AreeII aI (describe d below) that Phas I immediatelII e y succeede addition i d de An Phasth n . II e innermost rubble rampar f Phaso t I overlieeII palisade sth e post-holen si chronologicae Th . cuttingE d an l positiosD therefors ni e established. n cuttinI g Gr, unfinishe e clos th f excavations o et a t dleasa s t wa t i , possible to establish the absence of the palisade trenches in the length uncovered s therefori t i d ean , likely that they bega t thia turo nt sn ni

H INNERC T I D .

HOWNAM RINGS

. CUTTINGI AI R EA

j CMPtS

Fig. 4. Pla sectiod nan Cuttinf no g P, Are. aII 202 PROCEEDINGS OF THE SOCIETY, 1947-48. point unless they had been cut away by the Phase III ditch—on the whole less probable. Ho r northwardwfa e enclosurth s e extende t knownno s e widtdi th ; h of both the palisade trenches averaged a little over 2 feet. Stones had been packed roun e uprightsdth , whic d beehha n abou inche6 t n diametei s r (Pis. XXXVII, XXXVIII); and in the case of the outer line, at least, these uprights had been staggered, probably to allow wattle, hurdling or brush- interwovenwooe b o dt depthe Th . s ground belool e wdth surfac thesf eo e uprights varied fro inchet alway6 m feet,-bu1 no 2 s o st swa possibl t i t e b o et certain when the bottom had been reached, as the old surface of the rock weathered ha d considerabl thin yo shile sidth l 5).f d (figseo an 1 4 .

HOWNAM RINGS. AK.EA . CUTTING11

PALISADE TtENCHES

Fig . Sectio5 . Cuttinf no Are, gD . aII

Phase II. The Wall Fort. Thidiscoveres swa thren di e cuttings Arean i ,I (thI , sI e entranced an ) III (figs. 6, 4, and 8), and in addition was visible for the greater part of the circuihilltope th f o t . founs wa consiso t d t I f largo t e stones face eithen o d r sid largey eb r blocks, and there was no evidence of coursing inside, though the facing may have been more carefully constructed. This wall showed no signs of having had wooden tie-beams in the Gallic wall manner, though the possibility of this must not altogether be dismissed in view of the short length of walling removed.2 XXXIX).—IAreaL P d I (fig an .6 n this cuttin wale founs gth wa l d fee0 width1 n i t e largb e th ,o t e facing-block outee th n rs o side having 1 For similar palisade trenche P.S.A.S.,e sse vol. xxxiv. (1900) . 117-20,pp fortr ,t fo Orchill sa d an , Kempy, Gask. Forts with apparently similar walls are known from Wales, and in several instances a rampart-walk on th2 e inner side can still be seen. (Tre'r Ceiri, Arch. Canib. (1926), p. 257, and fig. 20; Caer-y-Twr, Holyhead, ibid. (1934), p. 158; Caer Drewyn, ibid. (1922), p. 117.) EXCAVATION HOWNAT A S M RINGS, ROXBURGHSHIRE3 20 .

been thrown dow incorporated nan innee th n rdi sloping rubble rampart of the succeeding phase. One fragment of pottery was found in the make-up of the wall. Area II (fig. 4 and PI. XL, a). The Entrance.—This, unfortunately, beed ha n very much robbeease th tsingl a n side do t ebu , very heavy

HOWNAM RINGS .AREA J SOUTH-EAST END.

: :•'.(>•': RUBBL ESILT-.• '

DITCH INNER R.UBBLE RAMPART FOUT WALL (PHASE III) (PHASE 111) (PHASE 11}

Pig. 6. Plan + section of inner rampart in Area I.

stone inche8 1 fee 2 , y sizen b sprobabli td an , cumbersomo yto e th r efo robber removeo st remainedd ,ha , indicating righe ar supposinn e ti w f i , g stils positionn i wa l t i widta , e entrancef onlho th feeyr 4 fo t , which showed no signs of gate-posts. This entrance may have been a single gap in the wall, closed when necessary with some movable object like a hurdle or brushwood. Area III (fig. 8).—This cutting was designed to show the chronological relationship e inne betwee Aren th i hute , s th rI a A e wal.d nth an l outed an r kerbstone wale th lf so wer e discovered e lowe wels th a ,s ra l stones of its make-up, overlaid by the rubble rampart of Phase III, which in turn had been cut into by the hut of Phase IV. Beneath the 204 PROCEEDING E SOCIETYTH F O S , 1947-48.

wal thin i l s cutting were found several piece f potteryo s lona , g iron object like a pin, as well as animal bones, slag and charcoal (see pp. 212-13 below). Jus t whata t tim againsr eo t whom t possiblthino ss walbuils wa e t lwa ti to find out, but that it came to an end, was dismantled and the hilltop defences remodelled in the late first century A.D. will be shown in the discussio . 206)Phasf (p no I querA e.II thaf no t datfouns ewa d among the material used by the Phase III people for blocking the entrance of this wall fort.

Phase Rubble III.The Ramparts. The Inner Rampart sectiones wa numbea t da pointsf ro , whil three eth e additional outer ramparts were also cut through in Area I (PL XL, 6). As has already been pointed out, the innermost rampart of this period was partly composed of the dismantled wall of Phase II but, like the outer ramparts, its composition varied according to the surface geology of the hilltop. Thus on the west and north-west of the hill, where the soil was rocky, ther mors ewa e stone than eartmake-ups it hn i , whil place n ei th f eo ditches whic beed hha n relativel sofe th tease d n soili th y an tg f easso di o yt south sides, only a slight scarping had been achieved (fig. 7). With the more steeply sloping hillside the scarping produced the same effect as that give thy nb e ditcheothee th t wouli n r forts e o d side unwise th dan b ,f so e to infer that this was either hurried or inefficient work. But the sudden appearance of the different technique of defence construction, reflected both in the digging of ditches and in the clever scarping of the slope to give maximue th m effect wit minimue hth m work, suggest possibilite sth y that this work may have been directed by refugees from the south of Britain, where earthworks of this type are abundant. It is clear from the little apparent change in the pottery of Phases II and III that these rubble ramparts were built by no strangers to the district. scarped Than hile g obtaio dth peoplt ldu o ne wh defence deptn i sy hma have been immediately descended from those who e hav,w e seen from Phase II, depended more upon stone walls than earthworks for the defence of their possessions t seemi t sBu probabl. e that their labours were directed by one or more people who were familiar with the defensive systems of the south. This inne rhav o rampart onle e eth yon bees twa n provided witha kerb, and it was evidently by far the most formidable of them. The Inner Ditch and Berm.—On the south side of the fort the ditch, from which the earth was obtained for building the inner rampart, was found to be som fee 6 presence 1 e tth th outsid d n bera o f an et o m, eit no her t ebu steeper slope suggests that there was some advantage in having the berm flat, thoug tacticas hit l significanc t understoodno s ei ditce wids Th hwa e. YVoe. Soc.Scot.]Ant. [VOL. Lxxxir. HOWNAM RINGS. ROXBURGHSHIRE. SECTION THROUGH DEFENCES ON WEST. (AREA i) SECOND RAMPART EXCAVATION HOWNAT A S M RINGS, ROXBURGHSHIRE5 20 .

and shallow and its filling was informative, though it must be remembered that only one cutting was made across it, and it would therefore be unwise to argu categoricallo eto sectioe th yn I nfro expose. mit cleas wa dr t i (fig ) 4 . largw thafe e a tstones , probably fro e rampartmth falled ha , n inte oth ditch soon after it had been dug, and before silt had formed in it. This suggests tha ramparte th t scarceld sha y been constructed before they were allowed to fall into decay, and perhaps even more can safely be inferred from the upper part of the ditch filling; for this appeared to be deliberate levelling overlaid by ploughsoil, and the possibility that the builders of the rampart havy sma e been instructe leveo dt l them mus borne b t mindn ei . Suc hcoursa eventf eo indee y sma d have been possible politicallye th r Fo . most convincing hypothesis seems to be that these defences were built at the time of the Agricolan Campaign, and it is historically known that relatively peaceful conditions followed the successful advance of the Romans. entrance Th e through this rampart excavatedno s wa t . The Second Rampart Ditch.d an —This rampar sectiones wa t Aren di , aI whil correspondine eth g ditc examines e hwa Th Aren . i d I aundeIV I t rHu rampar exposes ta foun s Aren di consiswa o dat I t almost entirel scrapedf yo - up material, and it was evidently considered impracticable to quarry into native th extentey rocan o kt . This dumped materia largels lwa y gravelly and stony soil, but there was also a large deposit of ash containing animal bones, and this may well have been debris from a midden or hut-floor belongin earlien a o gcontemporarr t o r y phas occupationf eo . The ditch corresponding to this rampart on the east underlay Hut II d agai an e fillin , (fignth 9) . g suggeste deliberate dth e shovellin f soio ln gi rather than natural silting. The hut overlying it may be as early as the second century t datinbu , g evidenc r thi fo s eunsatisfactoryi s e on d an , cannot be certain that the ditch of Phase III was filled in so early. Certainly appreciably lace an th f ko e amoun siltf eitheo n ti r ditch sections exposed suggests thae rampartth t s were dismantle possibld dan y ploughed over withi yearw theife f so na r construction. The filling of this ditch contained a fragment of sandy red close-grained pottery and fragments of carbonised wood. The Entrance through the Second Rampart.—This was clearly visible before excavation, for the kerbing-stones of the rampart were showing through the grass. The entrance was stripped (fig. 3), and found to be perfectly fee1 simpl1 t d wide post-holeo an e N . s gave evidency an f o e timber-work. No sign of wear showed in the entrance, though the rotted rock would undoubtedly have quickly formed hollow ruts or a slightly concave surface had wheeled or foot traffic frequently passed through. This lack of wear again supports the view that this phase of the hill defence was shora t one. The Outer Ramparts.—The two outer ramparts which showed at their 206 PROCEEDING E SOCIETYTH F O S , 1947-48.

greatest height on the west side of the fort, were sectioned in Area I and were foun consiso dt t entirel scraped-uf yo p soil varying from stone sano st d dan gravel accordin nature surface th th o gf t e o e geology shows e i s th A n ni . sectio nramparte (figth ) 7 . s decreas sizn ei e from within outwardse th d an , outermost rampar founhardls e b wa to d t feey 2 t high. The Blocking Fort e oth f Entrance.—The entrance throug e forhth t wall of Phase II was found on examination to have been much robbed and apparently blocke blockine dTh evidentls (fig. gwa 4) . earliee yth r event, and was presumably done by the rampart builders of Phase III, whose remodellin e defenceth f go s alread ha s y been referrewhosd an , edto wor k include constructioe dth entrancw ne a f eno farthe weste th o .rt With very little soil above the rotted rock it would, but for a fortunate discovery, have been impossible to disentangle the likely sequence of events which took e walplacth lt ea for t entrancee loweth t r stonBu rotara . f o e y quern f lato e) (figfirst-centur11 . y datfouns ewa d place uprighn a n di t position agains e westh t e walt th fac lf o eentrance . This important discovery provide blockine th sa dat entrance r th efo f go e and inferencey ,b e th r fo , replacement of the old defensive wall with the new multiple ramparts. Within and around this area later robbing had destroyed the east face of the entrance, and soil had also been removed from the rubble rampart immediately outside. In the disturbed soil immediately outside the fort waldiscovereds lwa inche, w onlfe ysa belo turfe w th iron a , n knif postf eo - Roman type (fig. 14).

Phase IV A and IV B, Two Huts and a Storage Pit. Within Phase IV, all those features of the site known to be subsequent conveniencr fo e Phaso ar t I eII e included, thoug impossibls i t hi e th n ei present stat knowledgf eo whethey sa o et storagre thesth huto d eetw san pit are more than broadly contemporary. For whereas Hut I could be fairly accurately dated, the evidence from the homestead, Hut II, was far less conclusive storage th d eAn pit. , too, could onl datee yb somo dt e time within the Roman period.

, AreaI t Hu III. , PhaseA V I (PI . XLI) a. , Before excavation it was clear that this hut, which showed as a slight circular hollow, had been partly cut into the ramparts of Phase III, and as a result of excavation a very satisfactory section was obtained, which e previouth showef o e o relatios dth phasetw d nf an o s betwee t hu e nth fore th t (fig. 8) . The Entrance was ill-defined and much robbed, and was on the east. It was not possible to ascertain its exact width, since the shallow amount of EXCAVATION T HOWNAA S M RINGS, ROXBURGHSHIRE7 20 .

HOWNAM RINGS, AK.EA m. HUT OF PHASE ivA

LOOKF .T PLA T HU LEVEA N F O L

PAVING STONES OUT EH FACE OF FORT WALL (PHASE II)

0 I 2 34 56 7 S 3 10 II 12 IS 14 IS 1 3 2. 0 FEET I METHES

C.M.P. '4-S

Pig , Are I . Pla sectio8 .t d an an Hu III f no , showin relatios git Pore th t o nt Wal d lan Rubble Bampart of Phases II and III. 208 PROCEEDINGS OF THE SOCIETY, 1947-48. soil overlying the rock at this part of the hill made it impossible to be sure in every case which stones were earthfast and which had tumbled. The Walls.—Bound the circuit of the hut, which was over 20 feet in diameter lowee th , r stonewale th lf swero e mostl positionn yi , thougw fe ha gapirregularitied san side th e n neas i entrance th r e poin lateo t t r robbing, and the removal of many of the paving-stones was probably undertaken same th poinet o a n time t t A wer. walle eth s standin heigha o gt mor f o t e than 3 feet, and the average was slightly under 2 feet. They were composed of heaped-up stones irregularly laid, and mixed with earth and smaller stones. A small cutting was made on the south-west of the hut to find the width of the walling at a point where little disturbance was evident, and it wa feet5 s foune .b o Timedt , unfortunately t allono wd furthedi , r work on this hut: the purpose of its excavation was to discover its date and therefor terminusea ramparte quernth d a r fo f Phasso e III, rather thao nt obtai fullee nth rt construction detailhu a f so thi d sHa bee . purposer nou , considerable time woul spenexaminatioe n b a do n t i hav d elengtha a f no h of hut-walling, to discover how .the roof was carried and whether there were post-hole t int thicknese sse o th walle th f .so The Floor was found to be covered with fallen wall stones and earth, placen i almos s d wa san t immediately belo e moderwth n turf. Paving- stones remaine patchesn di otherd havy an , sema been lefe robbedth t n O . e entrancoth f e grouneth rocks dwa , whic -beddes hwa d almost vertically appeared an quite b o edt unwor r chippedno therefors i t I . e improbable that paving-stones had ever, been placed upon it. It was interesting to note that no finds of pottery came from this area, and it is suggested that her inhabitante et laith hu d e strath f heather wso o sleepinr rfo . Thergon e was no hearth, and a very slight hole in the rock in the centre of the floor havy ma e hel dcentraa l t postthiBu s . hol sided eha s which were quite unchippe s mori t i e d probabldan ynaturaa l feature, thoug t hmusi e b t remembered tha e centrath t beed l ha uprighn e hutt i presentth f f i ,o t , could well have stood on a paving-stone. If this was the method of roofing, raftere th s could have sprun wallse th gf o ,fro intflae p mth to ot whice hth wet would drain as in many Hebridean houses of to-day. The Finds.—A large quantit nativf yo e potter floor e founs th ywa n , do somd an e fragment f Romao s n ware which have been estimate belono dt g approximatelo t e perioyth d A.D. 250-300. Other finds include da ston e weight partla , y bored spindle whorl, many rubber whetstonesd an s , etc. These finds are described in greater detail below (pp. 215-17). In addition, just below the turf in the middle of the hut was found a penny of George III and a broken piece of clay pipe, lost perhaps by some shepherd resting in the shelter provided by the hollow of the hut-circle. The Storage t (AreaPi III) (not illustrated).—A numbe f depressiono r s considerably smaller than those of the huts, and often only about 3 or 4 feet EXCAVATION HOWNAT A S M RINGS, ROXBURGHSHIRE9 20 . across, coul seee db n her therd ean e withi ramparte nth Hownaf so m Rings. accordingls wa t I y decide thesf do thae e on tdepression s coul excavatee db d to determine whether or not they represented storage pits of the kind known fro hill-forte mth frod san m southe Irofarme th mosd nf Ag o san , t fully recorded-b Bersr yD t Littlua e Woodbury, near Salisbury. resule Th 1t e excavatiooth f onls nwa y partially satisfactory depressioe Th . n chosen showed an , waclearlp I du t s closyHu wheo et surface nth e firss soiwa tl removed. But as the excavation continued, it was apparent that the sides of the pit were impossible to recover exactly, since the pit itself had been t dowcu n into made soil whicr fo , hobviouo n ther s ewa s explanationt A . depta reachedonlf s ho feeypavewa e 3 t th t pi , d e quit flooth f eo r carefully laid with paving-stones similar to those used in Hut I. The walls of the pit were then undercut to see whether these paving-stones continued, since t impossiblno s wa merelt d i e ha tha e yw t com earlien ea t dowo t rhu n no floor. But the paving did not continue, and appeared to have been laid specifically for the pit floor. It is clear, therefore, that this pit was intended for storage purposes mayd an , , like southere somth f eo n examples already referred to, have been lined originally with wicker-work or' leather. The pottery was inconclusively dated on its own merits: it was native in type, and it included one base and one rim. These should be dated by the Roman blue glass bead found with them, though that cannot be dated more closely than to the Roman period generally. The filling of the pit was quite homogeneous mixtura , eartf eo largd smald han ean l stones. Results, though not as conclusive as had been hoped, show that this was probably yma storaga hute d Phasth f an t belonginso f , epi o e IV on o gt well belong to sometime between approximately A.D. 200—400.

, AreaPhaseII t (figV B,I V . I Hu .9) seeThie b nsn frohut ca mai e s mth ,a n fore plath tf n(figo belonge, .2) d to the homestead enclosure which had been built after the rubble ramparts had been levelled or ploughed over, at the eastern end of the hill. The hut itself was built over the filled ditch of the second rampart of Phase III, and its section was informative. Somewhat oval in plan, this hut was a little larger than Hut I, and measured 25 feet across inside its longer axis (E.—W.). The Walls showe s sligha p du t banks entirely grass-grown d risinan , g to a height of only a very little above the hut floor, though owing to the placen i s lane swa slop th almost di f eo fooa t t abov groune eth d surface outside. These walls were show excavation o n o havnt e been madf o e sand stond yan y soil—in facnaturae th t l subsoi thin o l shille sidth .f eo Originally these walls had been faced with blocks of stone, but these had almos l beeal t n robbed away f thre,o excepw e ro contiguou a t s stonen o s 1 Proceedings Prehistorice oth f Society, vol . (1940) .vi ff 0 .3 . p , VOL. LXXXII. 14 210 PROCEEDINGS OF THE SOCIETY, 1947-48.

the north-east, another on the north, and one on the north-west. Indeed this whole hut had been so seriously robbed that it was difficult to recon-

HOWNAM RINGS AUA P'HASEOF ,v. HUT IVB

HOLLOW

/ LIME / OF i DITCH OF'PHASE in

OI234S67S910 20 0 I Z 3 4 S 4 T -I I =SF££T. I METRES.

T FLOOHU R

PHASE III PITCH

01234^6740 0 1 9 i.i 111 i I.. .1. i i.i f££T I METKES

Pig. 9. Plan and section of Hut II, Area IV. struct its original appearance, nor was it possible to discover any suggestion of its entrance. EXCAVATION HOWNAT A S M RINGS, ROXBURGHSHIRE1 21 .

The Floor founimmediatels e b wa o dt y belowese d th ture wan tn th o f a little over a foot below it on the east. It was sporadically paved, though isolated paving-stones remaining hertherd ean e suggest that othery ma s have been removed at the same time as the facing-blocks. Where the floor was not paved it was difficult to distinguish, though its position could be approximately ascertained from the scatter of pottery upon it, and in addition it presented a more compacted surface to the trowel, as might be expected fro mwell-troddea n floor. Post-hole.—d an t Pi A large central pit, rather largpost-holea r efo s wa , found nea e centrhute partlth rd th f an t ,througeo ycu e filline hth th f go Phase III ditch (fig. 9). This pit, which was 2 feet 6 inches across and a little over 2 feet deep, was filled with fine dark soil and contained a fragment of rimpost-holeA . inche7 , diameten si oved fooan r1 r t deep founs wa , d north-ease onth t floorhu te , sidsunth f eo k int underlyinfilline e oth th f go g d compacteditchha t I . d side d containean s d very fine loose dark soil, with no traces of carbonised wood. The Paved Hollow.—Inside the hut wall on the south-east was found a carefully made stone-lined hollow with its walls formed of up-ended paving- stones. This hollow (PI. XLI, 6) was filled with loose earth and stones rather darke coloun i r r tha bettee nth r drained soil overlyin t floorhu e . gth This hollow was less than a foot deep, and it contained two fragments of pottery and a long polished stone—probably a whetstone. The Roof.—As in the case of the first hut to be excavated, there was again no evidence of the method used for roofing the hut. Possibly the central pit may have been intended to carry the main upright to which rafters coul attached e unfortunats db wa t I . weathed e ba tha e th rt con- ditions made further work impossible, and the proper stripping of a length of wallin looo gt r possibl kfo e post-hole neves wa s r completed lengtA . h of three feet was examined on the north-east, but no post-holes were revealed, and they may well, if present at all, have been more widely spaced. In the make-up of the wall at this point was found a Roman bronze nail- cleane unspecifief o r d date. The Finds.—A large quantit nativf yo e potter e floofouns th ywa rn do Romao n hutd e onfth an , pottery, thoug base hon e showed Roman influence. On the north-eastern sector of the hut floor, just oh the lip of that part of it which overlay the filled ditch and would certainly have been once covered e ramparth y b Phasf o t e III, were some fragment yeUoa f whito d s wan e inlaid glass armle typa f o et well knowe earlth yn ni centuries A.D.d an , almost invariably distributed between the Hadrianic and Antonine Walls.

Although found on the hut floor, these armlet fragments cannot be used1 conclusivel r datinyfo e hutgth , since thehavy yma e been e eitheth n i r

1 See Kilbride Jones in P.S.A.S., vol. Ixxii. (1938), pp. 366-95. The only closely similar armlet to our example came from Traprain Law. 212 PROCEEDINGS OF THE SOCIETY, 1947-48.

topmost filling of the ditch, or derived from the rampart which had been removed when the hut was built. A few pieces of slag and charcoal (see pp. 224—5) were found on the floor. The Date f Occupation.—o e armleIfth t floorthu e belong th e , th o t s second century A.D. is the likeliest date of occupation. But if the armlet is derived from an earlier phase, it cannot be more closely datable than to within, or more probably shortly after, the latter part of the Roman period. Evidenc I insufficients ei t relative Hu th f d o ean d , datean f thit o s hu s remain undecided.

THE FINDS RELATED TO THE PHASES. . PalisadeI Period. (a). Pottery (fig. 10).—Owin e shallownes th e fac o th gt t e soid th an lf so tha palisade th t e trenches were overlai innee th y rd b rampar f o t Phase III, it was impossible to be certain of the correct chrono- logical horizon of the pottery found near the palisade trenches. (1) This piece can definitely be equated with Phase I, since it was fortunatel e postth yf - o packine foun e th n on di f go holes in Cutting D. It is a plain rounded rim, poorly made and fine grained. Roughly finished inside and out (see fig. 10, I, I). (2) From between the two rows of post-holes and therefore almost certainl Phasf yo . eI Thilarga s si e coarse base, evidently belonging to a very large storage jar (fig. 10, I, 2). (b-g) Metal, Glass, Animal Bones, Slag, Charcoal, Stone Implements.— No finds. Walle Th . Fort.II (a) e basPottery.on e d wer e an —Oneonl th e m y ri piece s definitely belonging to this phase. Both were hard and well made. (1) Base, well made, hard and grey with slight outer bulge (fig. 10, II, 1). (2) Unusual squared rim. Hard, close grained exteriord Re . . Marks on the top of the rim show that the pot must have had a lid when in use (fig. 10, II, 2). Both these fragments came from the turf-line under the fort wall in the cutting behind Hut I in Area III. Note that rampart cutting (Area I), No. 2, was also discovered in the fort wall, but t sufficiena t undoubtedle no b to t dept t i r hthifo f yo s phaset I . may have worked down from the overlying rampart of Phase III. EXCAVATION T HOWNAA S M RINGS, ROXBURGHSHIRE3 21 .

(b) Metal.——Pointed iron object from under fort wall in cutting behind Hu I (figt . 12). (c) Glass.—No finds. \ \ I?

ffm,. m '• ''*•••

Fig. 10. Pottery from Hownam Rings, (J.) (d) Animal Bones.—Representing , ponox y (siz t knownno e d pigan ) . All from same horizo e iroth n s na objec t described above. (See Appendix I, p. 223.) (e) Slag.—From same horizon. One piece identified as typical iron slag. (Appendix IV, p. 225.) (f) Charcoal.—From same horizon. Fragments identifie s Birch.a d (Appendi . 224.p , x)II (g) Stone Rubbers, etc.—None found. 214 PROCEEDINGS OF THE SOCIETY, 1947-48.

III. The Multiple Rubble Ramparts. (a) Pottery (fig. 10).—The pottery was almost all in the make-up of the ramparts and could possibly have been derived from earlier periods. nearln i t l caseyal Bu s thi improbables si , sinc e sherdeth s were found well outsid aree eth a enclosed eithe Phasn ri Phasr o eI . eII (1) From Cutting B, Area II. Rim. Roughly made with fair amoun f backingo t . Surface smoothed. Diamete- un r certain (fig. 10, III, 1). (2) Rampart Cutting . Rounde14 . , AreNo , dI a rim, slightly inturned. Light red, rather fine clay with smoothed surfaces (fig , III. .10 ,2) (3) From Cuttin , Are C g. II aInturne d rim. Fine, smooth and hard reddish brown. Associated with coarser frag- ments (fig. 10, III, 3). Also belonging to this phase were a quantity of featureless fragments. They are nearly all fine grained, grey, red or black in colour. Several sherdfinf o ee blacsar k ware, somewhat like eth Iron Age "B" ware of many South English sites. One of these came fro innee filline mth th f rgo ditc Cuttinn hi t , Arebu gF , aII this could not be definitely associated with Period III rather than depts it onlPerio s s inches8 a hy wa 1 V dI . Fragments from the Rampart Cutting (R.C.}, Area I.—These positioe describe4 Th 1 . Ill . f . no 2 ,No No s excludda m ri e eth the other fragments can be seen in section in fig. 7. R.C. 1 and 2. Fine grained, grey and red. R.C. 3. Coarser. . 6 Fin d R.Cean gre4 . y ware. R.C. 5. Fine red. R.C . 7 .Fin e black, like fragmenth e t froe ditcth mn i h Cuttin , AregF . aII R.C. 8 and 9. Very fine black, like South English Iron Age "B" ware. . Thick13 d an , 2 ligh1 , t 11 red d R.C.an 0 Wel1 . l made. thesf Vero w ey fe fragment s wer coarss ea e eithepottere th s ra y associated with Phase I or Phase IV, but it is closely similar to that from Phas. eII (b) Metal.—None. (George III penny as a stray find.) (c) Glass.—None. (d) Animal Bones.—Mostly from rampart cutting e ashth , yAren i , I a material largely comprising the second rampart. The animals EXCAVATION HOWNAT A S M RINGS, ROXBURGHSHIRE5 21 .

represented include sheep or goat, ox, possibly red deer, and a pony of 12-13 hands. (Appendi . 223.p , xI ) (e) Slag.—Fragments from Cutting B, Are . aII (Appendi . 225.p , )xIV (f) Charcoal.—Fragments identified as Hazel and Birch from Cuttings D and E in Area II and rampart cutting, Area I. (g) Stone Implements, etc.—From Cuttin , Are B g, camII a a estra y Neolithi e detailcth axe r whicf so fo , Appendi e hse x III . 224,p , . 5 . No fig, d an 11 . (h) Quern (fig. 11).—In the blocking of the entrance to the Fort, Phase II, on the south side of the hill, was found the lower stone of a rotary quern. Thid beeha sn carefully placeuprighn a n i d t ' position near the inner side of the entrance on the west. This quernstone belong typa o st e with gently sloping grinding surface —not uncommo t frequentlno t nbu y described from dated sites. Similar examples were found to belong to the first half of the first century A.D. at Maiden Castle, and it is likely that a slightly later date, say in the second half of the first century A.D., should be ascribed to it in the north.

Storageand II, Pit.B. Huts and I and Period A IV (a) Pottery (fig. 10).—This includes pottere th huto yd tw sfroan e mth storage pit. Hut I was dated to the latter third century A.D. by the Roman pottery associated with the native fragments found on the hut floor. Hut II is more difficult to place chronologically. One imitation Roman pottery base was found, no other Roman ware. The pottery certainly might, from its similarity to that from Hut I, be approximately contemporary with it, but one other pot (fig. 10, bottom right) was made of ware which was quite unlike anything else from any phase of the site. The possibility cannot be quite rule t thadou t this somewhait s n huti s , wa placet t angulai s da r enclosure, may belong to a post-Roman occupation of the hill. It is impossible in the present state of our knowledge to tell, but the presence oknifa f f Dareo k Ages type e robbefounth n di d entranc f Phaseo n i I eI Cutting F, Area II, shows that people were living in the neighbourhood at this time. For the discussion of this knife see p. 219 below.

IV A. Hut I, Area III. Native Pottery.—Several pounds weigh f roughlo t y made thick pottery came fro mwholee th thimors t i floorn t shu ,i O e . gritt coarsd yan e than tha , thougtII frot mhHu mucsamee th s hi . Ther fragmente ear severaf so l pots with inturned rims and flat bases. 216 PROCEEDINGS OF THE SOCIETY, 1947-48.

(1) Large pot with inturned rim, wavy sides and roughly smoothed surfaces s . especialli .1 Thi t , RestoreA po s V I y , fign i d10 . interesting, sincclearls i t ei y simila unpublishen a o t r t disdpo - Hog. A . t Ingraga H . covereA mr M Hil y Northumberlanddn b li .

Pig. 11. 1. Quern from blocking of Wall Port entrance in Area II (J); 2. Quern from I (J). Stont 3 ; Hu e polishe I (i)rI . Ston 4 t fro; mHu e weigh I t(J) t fro; mHu 5. Ston froe emax Cuttin Are, gB I (i)aI .

(2) Similar onlym Ri .. Light brown, smoothed more carefulle th n yo outside. There are several rims of this kind, as well as flat bases, much abrade fragmentaryd dan d som an f thes, belono y e ema g to the pot, No. 1, as above (fig. 10, IV A, 2). Roman Pottery.—The remark thin so s pottery were kindlyr madM y eb John Grillam, to whom I am most grateful. Mr Eric Birley also examined e Romath n war independentld ean y e samcamth eo et conclusionr M s a s Gillam. None of the pottery is illustrated owing to its very fragmentary nature Gillam'r M . s report run followss sa : fragmen"(3m Ri ) Samiaf o t n ware, beaded t fro angle bu , mth t seemei s represeno t t Dragendorf For rathe1 m 3 . Hard37 rr o tha , 8 n1 glassy orange glaze. Probably E. or C. Gaulish. (4) Small piece of black fumed ware cooking-pot; wall fragment with EXCAVATION T HOWNAA S M RINGS, ROXBURGHSHIRE7 21 .

cross hatching piece Th hars ei . d finishede markth a t f o sbu , thum finger bo r drawn across presumably horizontally must serve instead of wheel marks to distinguish the way up of the piece. crose Th s hatchin thes gi n founobtuse b o dt e angled treate Th .- ment of the unburnished part of the side is difficult to describe, but it reminds me more of the Constantian than of the Antonine cooking-pots. (5) Minute portion of rim, burnished inside and out, of a black fumed ware cooking-pot. It has a widely outsplayed rim. (Cf. Birdoswald, 19 C.) There were eight other featureless pieces of Roman pottery. Only two vessels are useful for dating: (a) The black fumed cooking-pot represented by nine or ten fragments. On fabric this is either Had./Ant. or Constantian. It is not pre-Hadrian. I regard it as Constantian for reasons given. (6) The Samian scrap. (Late second century?.) " A t seemsi s improbable thaoccupies twa thi t morr shu dfo e than fifty years or so, and as it is probable that all the pottery should belong together, the most likely date for its occupation would be approximately A.D. 250-300. cooking-pote Th above4 . ,No , certainly canno consideree tb latea s da r stray since ther mane ear y fragment . it Almos f o s t equally unlikely woule b t di tha Samiae tth n sher straa s di y sinc vers i t eyi little abraded t SamiaBu . n sherds of this kind could have lasted till the mid-third century. It is for this reason tha approximate th t e dates A.D. 250-30 suggestee 0ar e th r dfo occupation of Hut I—and the occupation may have extended a little into the fourth century. (b-e) Metal, Glass, Animal Bones, Slag.—No finds. (f) Charcoal.—From floor of Hut I. All Hazel (see p. 224). . (g) Stone Implements.—There were a number of stone rubbers and whetstones from this hut, and in addition, a weight made of ston d witan eh hour-glass borin s foungwa d d wha(Noan , t .4) appears to be an unfinished spindle-whorl. This is carefully made, flat and circular, but the boring has never been completed (fig. 12). The "weight," if such it can be called, could hardly have been heavy enough to be effective for any outdoor purpose, such as weighing dowt ovene thatcherna a mors i dt i e roof d probablan , e looa s mthawa weight t i t fragmenA . rotarf o t y quer, n 11 (fig. hute flooe founth s th f . o rNowa n dThio ) typa 2 . s si e founn di e brochsth d examplean , e knowar s n from Lamaness, Sanday. Thes r removefa e e shoulb n dati dt no ed from this Hownam example, know e o belonlatt nth e o t gthir r earlo d y fourth century A.D. 218 PROCEEDINGS OF THE SOCIETY, 1947-48.

The Storage Pit, Area III. The only finds from this pit were the blue glass bead (fig. 13) and six or seven fragments of pottery, well made, hard and grey. Amongst these were one large base and a wide rounded rim, not illustrated.

Period IV B. Hut II, Area IV. (a) Pottery.—Unlike e homogeneouth s collectio f potsherdno s frot mHu fine dw , tha I sherde muce tth ar hI sI morfrot mHu e mixedd an , on archaeological grounds they canno regardee tb closea s da d group. For not only was the floor of this hut built over the filled-up ditch of Phas rampare e th III t thaf bu ,o t t phase must also have been levelled before its construction. Thus all the finds from this area must belon quits i g t i eithee d Phasr Phaso an rt o , I eIV eII impossible with so little topsoil to fix the correct date of any individua bees lha nfind t I mentione. d before thae nativth t e pottery from this floor was mostly very similar to that from Hut I, though perhap whole th n eso rather finer f thiI .s pottert no s yi derived from Phase III t woul,i d seem probablee tha th life f eo tth hut overlapped, at least in part, with that of Hut I. (1) One pot (fig. 10, IV B, 1) was significantly different from the rest. Not only was it much finer made, and sandier, with no gritty backing, but its walls were built up in the coil metho showns da . This pot, like Georgeth I pennII e y and clay pipe from the floor of Hut I, might represent later squatters on the site, but it may represent the only underived pottery—broke facn ni t durin e occupatiogth n post-Romane b t i y e siteoth Ma f . ? Only similar dis- coveries or more excavation will prove whether this small, somewhat angular steading, whose positio ignoreo ns e sth rampart e fortth ,f o sreall belony e fiftth o yt ma h o gt sixth centuries A.D. (2) Incurved rim (fig. 10, IV B, 2), probably the diameter was about 6 inches. This may represent, with other fragments, same potth f eso profilrestoree th ) 1 s e, t a (No A dpo V .I from Hut I. There were four fragments of this pot and the t quitanglno s ei e definite baso eTw .fragment y ma s belong to this pot. These, like many other fragments from Hut II, are more like the ware from the ramparts of Phase III, and may wel derivee lb d from that period. (3) Small base about 2 inches across. Roughly made of light EXCAVATIONS AT HOWNAM RINGS, ROXBURGHSHIRE. 219

brown ware with black surface—still remainin patchen gi s on the inside. This is clearly a native copy of a Roman jar. (4) From pit in hut floor. Dark, close-grained rim with slight internal ledge. From Paved Hollow in Hut Floor.—One fragment grey ware, and one very hard-fired sandy, orange ware with pounded stone backing. Several fragments closely resembling these sherds were unded foun Phasan e rn th di I rampart eII Cuttinn i Are, gC . aII Roman Pottery.—None. Native imitation, see No. 3 above.

Fig . Above12 . , Roman bronze nail-cleaner frot IIm;Hu below, iron object from under Fort Wal Aren i l a III; unfinished whor . lI (Alfrot mlHu £. )

(b) Metal.—Roman nail-cleaner of bronze (fig. 12). Uncertain date (cf. that from Rotherley (Pitt-Rivers, Excav. Crariborne Chase, vol, .ii . 128 p commend )an Wheelery tb , Lydney Report, . 84)p . This swa fountherefors i d d an buil t e earliehu t e int wale th of rth o ltha . nit (e) Glass.—Fragments of amber-coloured, translucent armlet with white wavy inlay (fig. 13). Found ove ditce rth h fillin e positiogth f no this find is archseologically ambiguous. It belongs to a well-known type belonging to the early centuries of the Roman era, and it is further discusse . 211p n .do (d) Animal Bones.—None discovered. (e) Slag (Appendix IV). (f) Charcoal, representing birch and hazel.—From floor of hut (p. 224). (g) Stone Implements, etc.—Several stones showing signe f weaus so r o r as polishers. One of these (fig. 11, No. 3) was found in the paved t floorhu hollo e . th wn i Iron Knife from Robbing of Gateway of Phase II.—The iron knife (fig. 14) from the disturbed soil outside the wall fort entrance in cutting F is of a distinctive type with a shoulder at the junction of tane th g wit bladee hth . Knive thif so t ssee typno mo ed charac - teristi Norte th f cho British Early Iron Age t the howeve,bu o yd r appear consistently in post-Roman contexts in Scotland, notably at Dunadd (Proc. Soc. Ant. Scot., vol. Ixiv (1929-30), p. 118) and 220 PROCEEDINGS OF THE SOCIETY, 1947-48.

Buston Crannog (Munro, Anc. Scot. Lake-Dwellings (1882), pp.. 222-23); in Northern England at for instance Grassington, W. R. Yorks (Yorks. Arch. Journ., vol. xxxiii (1937) n . i 170)p , d an ; Ireland in a whole series of sites ranging in date from Garranes (late fifth-early sixth century A.D. o )Cahercommaut n (early ninth century A.D.) and Carraig Aille (eighth-late ninth century A.D.) (Proc. . Acad.,I . R vol. xlvii (1942); ibid., li isout e (1949)th hn I . of Englan t leasda versio a ttype th ef no survive s int latee oth r Middle Ages (London Mus. Medieval Catalogue (1940), PI. xi. The e Hownadatth f eo m knif likels betweee i eli o y t sixte nd th han ninth centuries A.D.

Pig. Fragmen13 . f glaso t s armlet from Pig. Iro14 . n knife from disturbed area .blud an e , glasII t s Hu bead from near fort entrance in Area II. (J.) storage pit, Area III. (^.)

DISCUSSION. The Hownam excavations have begu throo nt wa littl e th ligh n o t onls i baresBorderse e t yi th th proble t f o tIroe Bu e beginningth . nmf Ag o , and, as in the case of almost all excavations which take place in new districts, e problemth t setsi s , though perhap fundamentalo s t no s more ar , e than it answers. Thus we know that at Hownam the initial phase was marked e buildinbyth f quitgo ea smal l stockade enclosure—a protection surely only valuable against cattle robbers or wild animals. The palisade trenches into which the uprights of the stockade were bedded are detectable on other Roxburghshire forts, for instance at Hay hope Knowe at the head of the Bowmont- valley, and we can suppose on many other forts as well,

though as there ar1 e very few, if any, superficial signs, they may not always br photographyeai e grounvisibly th t b infrequentl n no r o eo d d an , y may have been obliterated by subsequent defences or occupation. At Howna guesn ca se m thaw t this stag approximatels ewa y durin secone gth d or first century B.C. But who were the people responsible for the con- struction of these stockaded enclosures? Were they related to the people who apparently at the same time were also using timber on a big scale for e buildin th e Galli th wha n weri f y cgo , wa fortsteso thesf I ? e people connected with the Iron Age "A" people of England? Certainly the

1 Excavated in June-July 1949. EXCAVATIONS AT HOWNAM RINGS, ROXBURGHSHIRE. 221

pottery fro firse mth t phas t Hownamea subsequenl al n i s ,a t phases, suggests no influence other than the local derivatives of the Late Bronze Age. But of their types f houseso , , their social system, theie rw modt ye f life eo s a , kno e supposww n nothingca e r thano ;t their wooden stockades were intended for defence in warfare. But this need not imply that the people were necessarily peace-loving stock-farmers, for it is not impossible that their war-time defences were built more solidly, maybe eve f stoneno n i , places of greater natural strength. We can guess that the large amount of wood needee upright th e stockadr th dfo f so localls ewa y obtainedf o t bu , charcoae th l specimens identified from Howna l wermal e birc hazelr ho d an , neither of these trees would flourish in thick forests, and this suggests fairly open vegetation. Many Iron Age forts in England and Wales have had a palisade enclosure as their earliest phase thid frequentls san i , y " foun"A belono e dt th o gt e Welscultureth n hO .border t Ffrida s d Faldwyn (Montgomery),d Ol 1 Oswestry (Salop) agaid an , t Eddisburna y (Cheshire) founthis e b swa o dt

casee tha o soute th S th t. h Scottish example t withouno e sar t precedent; 2 but how far, if at all, it will be found that they were inspired from the south, or whether they may have had an independent origin, remains to be seen as a resul futurf o t e excavations. e walOth f l forts (Phas t Hownama I eI knoe w ) wvera y little more, fo rt leasa the e ty ar more often visiblgrounde th t withou n eBu o . t exca- vation it is naturally impossible to know whether or not they belong to the class of Gallic wall forts connected with the Abernethy culture, in which timber framewor uses kwa d (and which when ignited becomes "vitrified"), o anothet r r class with sheer-faced walls built without timber lacing, t Hownama s a Gallie vitrified Th c.an d Aberneth e fortth f so y culture have pase inth t been suppose belono "BdIroe t e th n"Ag o gt cultur e complex, but it has recently been suggested that they are more probably connected wit e eart d timber-framehth an h de "AIroe th fortn soutnAg i "f o s h England—a type of fort which gradually spread northwards to Scotland via the Welsh marches. But a number of south Scottish and highland zone

forts with stone walls canno3 t be ascribed to that culture, and in many cases, in Wales and Scotland in particular, it was most probably the local descend- ants of the Late Bronze Age people who built these strongholds. At all

events, the fact that this type of fort generally precedes the constructio4 n of multiple ramparts (or multiple walls) has been observed on other Roxburgh-

shire forts 5 and borne out by the results from Hownam. Other examples Arch.1 Camb., vol. xcvii. (1942) . 1-57pp , .

2 For both these sites, see Varley and Jackson, Prehistoric Cheshire (Chester, 1940). 3 Piggott, British Prehistory (Home University Library), p. 177. interestinn A 4 g Welsh exampl a for f to e whose history appear o havt s e been partly similao t r Hownam Rings, is at Dinorwig, Llanddeiniolen, Caernarvonshire (Arch. Camb. (1947), pp. 231-48). Multiple rampart herd sha e been adde earlien a o dt r stone wall fort. s Details of these will be published in the near future by the Royal Commission on Ancient Monuments. 222 PROCEEDINGS OF THE SOCIETY, 1947-48.

in Roxburghshire include Shaw Craigs, The Castles, Swindon, and Woden Law. These walls generally enclose quite a small area and seldom loop down rulbeloa s hill e ecrese A wfeeth the 0 th .f abou 1 e o t ty ar o thick t 8 t , faced with larg ecore stoneth e d fillesan d with rubble. One, possibly analogous example only, Hownam Law, encloses a very large area, and must be regarded as a tribal oppidum of some importance. This phase at Hownam ended only shortly before the arrival of the Romans supposee analogyb n o y , ma so , d t i ,tha othee tth r forts mentioned as sharing similar characteristics are approximately the same date. Un- fortunately, this second phas t Hownaea m produced onl ylittla e pottery hardld an othey yan r objects tha o e culturas , th t l affinitie people th f o se building these walled forts cannot be established. One point, however, is worth noticef yo . Such potter founs wa ds ya fro m both PhasPhasd an eI e II (and, to anticipate, Phase III as well) was so noticeably similar in character that it is almost certain that throughout the period of occupation the underlying peasant culture remained unchanged o new-comern d an , s arrived with different pottery or objects denoting an intrusive material culture. Instead, little development took place eved e shapean ,nth f o s the pots hardly altered during two hundred years or more. Bearing this fact in mind, it is hardly believable that a sudden complete change in building technique took place thad an ,t stonework suddenly replacee dth use of timber. Such a theory would be less acceptable than that the two building techniques wercertaia o et n extent contemporary thad an ,t stone- wor uses defencr kwa dfo e while timber stockades were use undefender dfo d village compounds. The last phase, represente e buildinth y db f multipl go e rampartn i s wale placth lf eo whic dismantleds hwa , begasecone th n i d firse halth tf o f century A.D. A large number of Roxburghshire forts show a similar addition of earthworks, and in the three forts already mentioned, Shaw Craigs, Woden Castlese Th Lad w, an Swindon , these ramparts replace singlda e wall juss ta at Hownam, and the presence at Woden Law of what are almost certainly Agricolan siege-works bears out the Hownam evidence for a date in the late first century for these multiple ramparts. But the excavations at Hownam show that once again the cultural heritage of this peasant community remained argun unchangedca e w wit d han , some conviction that the innovations in methods of warfare, reflected in the buildin f defencego t resuldepthn i no s d t di ,fro ma movemen f largo t e number peoplew ne f so int districte oth . Rather than thi t seemsi s likely that onl leadershie yth techniquew p ne changed e th fightinf d so an , g which we know were in common use farther south were brought to'the north at this timeknoe W w. tha soute th t f Englanho greatls dwa y disturben di the mid-first century by intertribal warfare following the Roman invasion. Aristocratic leader theid san r families were driven from theilandn d row s an EXCAVATIONS AT HOWNAM RINGS, ROXBURGHSHIRE. 223

fled to found new kingdoms with their kinsmen elsewhere. So is it not probable that these multiple ditches mark the arrival of such political refugees fro e southmth ? Being establishe thein landsdi w ne r , however precariously held against others who had moved to the same district, their tenur thesf eo t shorte bu land s , sperhapwa t twentsno y years, before they were involved in the new Roman advance under Agricola. If this hypothesis is correctimaginn ca e w ,e tha multiple th t e ramparts wer witg edu h some urgenc preferabld yan y adde siteo dt s already fortified t followsI . , then, that some of the single-walled forts which, as we have mentioned, were re-designed in this way must have been in use during or shortly before the mid-first century A.D., and the palisade enclosures, if they are earlier and t contemporaryno e befor,b nee e first th eno d t century B.Ce 194Th .9 excavation t Hasa y hope Knowe have thrown ligh thin o ts palisade phase. What happene theso dt e forts afte arrivae Romanse th rth s f i o l t I ? more than probable that those people livin fortifien gi d site thesf s(i e were indeed permanently inhabited) were ordered to deface their ramparts, though they were evidentl forbiddet yno sitee livth o n.t n eo This they did, and evidently lived peacefully until the withdrawal of the Romans. Then, once again, times of trouble are marked by the building of new forts in the Dark Ages, when hill-forts like Ruberslaw, The Dunion, and many others appea havo rt e been constructed.

APPENDI. XI The Animal Bones. Misy B s PLATT, Royal Scottish Museum, Edinburgh. Animal Bones from under Stonethe Wall Fort Cuttingin behind Hut AreaI, II. These are all remains of domestic animals such as one would associate with human habitation wilo N d. specie representede sar excludee on f i , s the rather doubtful evidence of a single tooth resembling the molar of a deerd re . Ox.—This seems to be the most numerous. There are fragments of humeru tibiad mild san ,an k molars shed fro myouna g beast, present. Pony.—From the sparse remains it is impossible to estimate the size of pone th y represented. Parlefa f t o t scapulrecognisede b n aca d an , there is also a narrow lower molar tooth about 3 inches from the left side also. Pig.—Milk molars alone represent this animal.

Animal Bones from under Second Rubble Rampart, AreaI. Sheep or Goat.—Horn cores suggest the latter; and animals of different siz s adulta e e presentar s . Also very slender vertebral fragments 224 PROCEEDINGS OF THE SOCIETY, 1947-48.

suggest that some animal probabld sha y been slaughtere lambss da . Bones from forhind an ed limbs—humerus, radiu ulnad an s , femur, tibi fibulaskuld a ad an l an ,fragmen t occur. Ox.—All the larger fragments of bone belong to this animal. They are osmala f slended an l r stock. Vertebral limd an b b bonri , e fragments are the most common. Deer.—d ?Re One molar toot presenhs onle i th ys i evidenc tt i heres A . e of this wild species, it may be of fortuitous occurrence.

From Inner Rubble Rampart. Cuttings B and G, Area II. Pony.—Two fragments form almos completa t e radius, provin size gth e to be of a small, slender Celtic pony, probably between 12 and 13 h.h. Ther alss ei oproximaa l fragmen cannoa f o t n bone (metatarsal). Ox.—This animal is represented by teeth, molars and pre-molars of adult animals shed an ,d milk molars. Sheep.—One fragmen alonb ri a e f occurso t .

APPENDI. XII Remains of Carbonised Wood. By M. Y. ORE, Royal Botanic Garden, Edinburgh. Phase II.—From under fort wal cuttinn i l Aregn i behin I a t IIIdHu . Birch. Phase III.—Under second rubble rampart in Area I. Hazel and Birch. From rubble rampart in Cutting D, Area II. Hazel and Birch. From ditch under Hut II, Area IV. Birch. Under rubble rampart Cuttinn i , AregE . aII Birch. Phase IV A.—From floor of Hut I, Area III. Hazel. Phase IV B.—From floor of Hut II, Area IV. Birch. , Are . II Hazel.t aIV Hu Fron i t mpi

APPENDIX III. Petrological Report on the Stone Axe. By F. S. WALLIS, D.Sc., Ph.D., Bristol Museum• . (The numbe e seria thas th i f e lo rt indeassigneax e xth compileo dt d by the Sub-Committee of the South-Western Group of Museums and Art Gallerie Petrologicae th n so l Examinatio Stonf no e Axes.) No. 423. Macro.—Fine-grained, pinkish, siliceous rock. Micro.—Usual characteristics of Group VI (Stake Pass). TWO TEXTILES FROM NATIONAL MUSEUM, EDINBURGH. 225

APPENDIX IV. . ThHALLIMONDF . e A Slags. y B , Geological Surve Museumd yan .

Phase II.—Under fort wal cuttinn li g behin, AreI t a dHu III . "This appears to be a typical iron slag." Phase III.—Under rubble rampart in Cutting B, Area II. "A frothed glass such as might result from accidental cause mighr siroo n a glasse nt tb slag.no ys i waste t "i t bu , Phase IV B.—Floor of Hut II, Area IV. "Seem includo st e both glas irond t iron-bearinsan bu , g minerals can of course appear in glassy waste."