Proc Antiqc So Scot, 129 (1999), 281 -302

An Atlantic at Durcha, Andrew well* with contributions by D Alexander, F Hunter & N Murray

ABSTRACT Recent excavations have confirmed the presence of an within a large mound at Durcha. presenceThe stonea of fortress Durchaat beenhad recorded 18ththe in century, but following stone robbing in the 19th century no clear traces of the structure survived to allow confirmation of its identity. The difficulties of the identification and classification of such structures in Sutherland, where little excavation has been carried out, are highlighted. The project was funded by Historic .

INTRODUCTION This report details the results of survey and excavations conducted in 1992 and 1996 at the reputed broc a sit f eo Durchat ha , near Rosehall (NGR 501C ; :N 72 0239o n W ; S NMR 0 5 C SN besidD O e m sit e All5 eTh illu14 liet. t Doirsa 1) s ' a' Chatha tributara , Stratn i r Allf yo h Mo t Oykel (illu st circle 1) field .Hu san d system presene eassoute d sar th an t 023ho t 7 t (NGR50 , C N : illu als; s1 004) 4 o NGR50 . C PrioN :investigatio o t r allegee nth d representes sitewa a y db large, grassed mound, roughly oval and irregularly pitted, with maximum dimensions of 33.5 m by 20 m; the modern dwelling house and outbuildings of Durcha immediately surrounded the site previouo (illuN . 2) s s archaeological excavatio knows i n havo t n e been carriee dth outd an , mound itself had not been surveyed in detail. Documentary sources provid modesea t amoun informatiof o t n abou earliessitee e tth Th . t known source is a reference to 'Dun Cor' in Pococke's account of his tour of Sutherland in 1760 (Kemp 1887, 115): 'at a rivulet, came to Dun Cor, another such stone fortress, it was thirty feet [c diamete n 9.1i ] 5m thick'] r m conditiowalle withine 8 th fee1. .Th x d c ssi [ t an , thif no s structure appears to have deteriorated significantly during the 19th century. The Sutherland itinerary of Pococke's tour was republished separately in 1888 by Kemp with added background material. Kemp's (1888, 48) explanatory note on Pococke's visit to Durcha (quoted here in full) states that:

broche th , Doir-a-Chatha Durchar o , , nort Rosehallf ho , below Cnoc-a-Choire onlw mounya no , d of stones, but the older inhabitants of the remember when many feet of it was standing, and whence hundred loadf so stonf so e were take builo nt dyke dth e whic enclosew hno Invernaule sth d Wood.

* Centr Fielr efo d Archaeology, Universit Edinburghf yo Higd ,Ol h School, 12 Infirmary Street, EH1 1LT 282 I SOCIETY OF ANTIQUARIES OF SCOTLAND, 1999

ILLUS 1 Location maps, showing Durcha, and the broch and dun sites (as classified in NMRS) of south- east Sutherland. Numbered broch site those sar e considere statistican di l analyses (see illu& 3 s1 14): 1, Dail Langwell; 2, Sallachadh; 3, Alltbreac; 4, East Kinnauld; 5, Coich Burn; 6, Caisteal na Coille; 7, Carrol; 8, Leadoch; 9, Backies; 10, Cam Liath, Golspie; 11, Feranach; 12, Altanduin; 13, Suisgill; 14, Kintradwell; 15, Gailiable; 16, Balvalaich; 17, Kilphedir; 18, Cam Bran. Other site: 19, Tor a' Chorcain, Langwell. (Based on Ordnance Survey maps © Crown copyright) DUNWELL: DURCHA, SUTHERLAND 283

ILLU S 2 site Vieeth f frowo m north-east, showin house gth e during redevelopment partly obscured behin grassee dth d mound

The first edition of the Ordnance Survey, surveyed in 1874 (Sutherland CII), indicates the site as dashea d circular feature annotate 'Pictiss da h Tower (Remains of)', suggesting tha fore tth f m o structure longeth o n s rewa clearly visible 190n I visite.Royas sit e e 9th th ewa y dlb Commission and recorded as 'a structureless heap of ruins, from which most of the large stones have been removed apparentls ha t I . y bee greaf no ruie tth n size d measurean , s 106' [32. across] 3m , buts a , no wall-faces are visible, the actual dimensions of the structure are not obtainable' (RCAHMS 52)mouno e 1911n ; Th . 20 ,d appear havo st e become grassed over sinc 190e eth 9 visit, with little stons ofit e conten exposedw tno .

SUTHERLAND BROCKS/ATLANTIC ROUNDHOUSES origi developmenThe nand count 'brochsold the Sutherlanythe of of t in ' lesthe ds formof sone well understood aspects of the Iron Age societies of northern Scotland. Relevant to this issue has bee lace consensunf th k o recenn si t debate about wha broca t h actuall som: yis e have adopted new terminologies based upon architectural characteristic alloo st w unitarya , nested classification range oth f f massiveo e roundhouses distributed across Atlantic Scotlan Armig d(e t 1990a), whereas other Swansog s(e n 1984) regar tere dth m 'broch potentialls a ' y includin granga f eo structure possiblf so y different date functionsd san . This report hereafter adopts Armit's (1990a, 59-60) classification of these structures as Atlantic roundhouses. His terminology proposes that drystone roundhouses with evidence for intramural galleries or cells are classed as complex Atlantic roundhouses. Only complex Atlantic roundhouses with upper galleries are classified as broch towers. | SOCIET 4 28 ANTIQUARIEF YO SCOTLANDF SO , 1999

Previous investigations of Sutherland Atlantic roundhouse sites have been limited in comparison to those of the Western Isles (eg Armit 1990b), the (eg Hedges 1987; Ballin Smith 1994) and even neighbouring (eg Fairhurst 1984). Much of what is known of Sutherland examples has been derived from surface survey. The Royal Commission's Inventory (RCAHMS 1911, xviii-xxiii monumentf )o Sutherlann si d presented both site descriptiona d san synthesis of the visible structural characteristics of the 67 sites recorded at that time as brochs. Although additional discoveries were cite listn i d s produce Grahay b d m (1949, 94-6d an ) (1968, 175-9) a mor d ean , stringent classification base n visiblo d e architectural characteristics was proposed by MacKie (1965), the Royal Commission Inventory remains the primary published sourc f informationo e e NationaTh . l Monuments Recor f Scotlanddo y b , 1996, had recorded 84 sites in Sutherland as 'brochs', with a further 14 roundhouses described as . Fairhurst (1984, 183) regarded the Sutherland 'brochs' as representing a secondary spread from an /Caithness core area, and drew a distinction between the distribution of sites along the coastal edge, often attended by extra-mural settlements, and the mostly isolated structures occupyin more gth e rugged inland areas. These latter sites suggestede h , , were th e home f 'intrudino s g broch lords t withise ' nlandscapa t circleshu f o e. Young (1964, 173-5, 184-9 d previouslha ) y considere e inlanth d d 'brochs f Sutherlano ' a developmen e b o t d t secondar coastae th o yt basie l siteth sn sthao t those inland wer smallef eo r size, more strategically locate r defencdfo e and, although skilfully engineered, built with less carefully chosen materials. These hypotheses, while of interest, had no dated excavated material to support them. Recorded excavations of Sutherland Atlantic roundhouses are extremely scarce. Four of the better-preserved examples were either excavated in a rudimentary fashion or cleared out in latee th r 19th century Liatm t Carroa ,Ca h d (Joasan l s 10)1873& 7 , Kintradwels ; illuno , 1 s l (Joass 1865,1873; PSAS1873, 53-6; Smith 1871,188-9; illus 14)o 1,n , and Backies (Stuart 1870, 291, 305; illus 1, no 9). Cam Liath is the only site to have been subject to modern excavation, but this at a site already examined in the 19th century (Love 1991 and references therein). A vitrified dun at Langwell (equally a complex Atlantic roundhouse in Armit's usage) was examined extensively in the 1970s (Nisbet 1996). A number of sites have been cleared of rubble to expose Borgwalline Th g , gDunrobi(e n Wood Sallachadh)d an , , althoug mosn hi t case recordo sn s exist structurey oan f r artefactso s recovered during these operations r indeedcircumstancee o , th f o s clearance datd th f an eo e work. surfacn O e evidenc recordee eth d Atlantic roundhouse site f Sutherlanso dividee b n dca d into three broad group certaintterme n si th f so theif yo r identification: 1 Approximately 40 sites, some surviving to several metres high, with diagnostic architectural features; this group includes the excavated or otherwise cleared sites. 2 Approximatel grasse5 y3 r rubbldo e mounds, often structureless, wit situn i h o littln r o e masonry visible, their form resulting principally from stone robbing and structural collapse. The identification of these sites as Atlantic roundhouses commonly stands upon factors such as fragments of exposed masonry, their general size and morphology, and antiquarian accounts. Durcha falls into this category. 3 Approximately 25 further sites have been claimed. Either they have been destroyed, or that clai mbases i d solel place-namn yo e evidenc fictitiousr eo . The distribution of Atlantic roundhouse sites in south-east Sutherland (illus 1) indicates that relatively low-lying locations adjacen majoo t t r watercourses were preferred e locatioTh . f no DUNWELL: DURCHA, SUTHERLAND 285

Durcha beside the higher reaches of a minor inland tributary of the appears somewhat atypical in this regard.

FIELD WOR 199199D KN I 62AN In 1992 a major refurbishment of the dwelling house at Durcha was proposed following a change of ownership. This included the demolition of a lean-to structure on the northern side of the building replacemens it d an , largea y tb r extension which would project int southere oth n sidf eo the mound Centre Th . Fielr efo d Archaeology (CFA commissiones )wa Historiy db c Scotlano dt undertak excavation ea 199y 2Ma within i footprine nth proposee th f o t d extension, immediately in advance of its construction (illus 3, Trench 5). A detailed topographic survey of the mound was conducte t thida s time (illu. s3) Following the completion of this fieldwork a programme of post-excavation analyses and reporting was commissioned by Historic Scotland. The report resulting from this work was not considere publishable b o dt fieldwore th s ea k require curatoriae th y db l authoritie recovered sha d insufficient information abou e characteth t e site th n ordeI .f o ro supplemen t r t this dataa , programme of further exploratory excavations was proposed by CFA in consultation with Historic Scotland, with the objective of characterizing the principal structures within the mound. This work comprised the excavation of four further trenches (illus 3, Trenches 1-4), and was conducted in October 1996, sponsored by Historic Scotland.

FIELDWORK RESULTS scop199e e 199d th Th 2f an e o 6 excavation fundamentalls swa y different 199e Th .2 investigation was designe recoro dt fule dth l stratified sequenc remainf eo s withi developed e aree b nth o at y B . contrast 199e th , 6 excavations wer t conducteeno responsn di threata o ewert d an , e limiteo dt revealin naturprincipae e gth th f eo l structures present withi moune nth d throug removae hth f lo rubbl demonstrabld ean y modern deposits, without disturbin situn gi pre-modern archaeological remains. This information from the 1996 work is important as it will help to determine how best proteco t managd tan structuree eth . resulte Th s fro excavatione mth t presenteno e sar chronologican di l orde investigationf ro . Trenches 2- detailee 4ar d first thess a , e indicat natur e principae eth th f eo l structure encountered. Trench 5 is then described, as the interpretation of the less well preserved remains of the structure recorded herbees eha n enhance subsequene th y db t excavation Trenchen si s 2-4. Trencs i h1 describe remainde lasth s ta s identifie principat relatde herth no o o et e d l structure. More detailed archive reports (Dunwell 1993 primare ; th 1996 d an )y site records have been depositede th n i National Monuments Record of Scotland.

Mound (illus 2 & 3) moune Th d appears from surface trace hav o smait o etw n components higherA . , north-western sector, wit hrelativela y level summi severad an t l indentation scard san s cause stony db e robbing, forms approximately two-thirds of the mound. The south-eastern area is lower and more irregularly mounde hollowedd dan , althoug edge hth e aroun south-ease dth t corne moune th f ro d forms a smooth arc. With hindsight from the 1996 excavations the basic topography of the mound can be related to the form of the principal structure present. The somewhat ambiguous surface traces previously had led the excavator to speculate on the presence of a galleried to oo OS

o •fl > H O > 52

---> r , \f . ' i X>>/"

r...... are f stonao e robbing — — — —Roundhouse wall (extrapolated)

ILLU S3 Topographi cmounde plath f no , showing trench positions, principal excavated feature proposed san d extenroundhouse th f to e DUNWELL: DURCHA, SUTHERLAN7 28 | D

roundhous south-eastere th n ei npost-medievaa ared aan l building occupyin summie ge th th f o t north-west area (see Dunwell 1993).

Trench 2 (illus 3-6) This trenc investigates hwa asseso dt charactee th s structure d datth f an reo e occupyine gth summit of the north-western part of the mound. A small trench was opened at a point where a cluste f largo r e boulder beed sha n expose resula s da f ston o t e robbing massivpara s f A . o t e drystone wall on a curvilinear alignment was revealed, this trench was extended to reveal its full widt characterd han .

Structure of the wall The wall, as exposed in plan and elevation, proved to be of complex character. It had uppee levee th th f ro l d surfac innes an totaa it , f m lro e 5 widtfac4. ef ho abovstoo m 1 d2. e th tha f o t exposed base of the outer face. Its remains were sealed beneath a spread of small angular stones containing modern artefacts, which probably represent the debris left by stone robbing and the recent use of the mound repositora s a r domestiyfo agriculturad can l midden curvilineae Th . r alignmen walline th f o t g intimated tha formet ti d parsubstantiaa f to l roundhouse. The inner face of the wall was relatively even, and composed of well-fitted sub-angular stones of varying sizes, including both blocky and tabular examples, with small stones packing any gaps between the larger components (illus 5). Some of the boulders forming the face measured up to 0.75 m from face to rear. It was the irregular rear sides of these larger facing stones which had been exposed in the robbing scar. The inner face was exposed over a sector spanning 3 m, with a depth of up to 1.2m revealed: its base was not exposed. The variable size of material used in the wall-face parallels the evidence from Crosskirk (Fairhurst 1984,40). outee Th r wall-fac formes wa e f lesdo s regular, sub-rounded boulders, wit gape hth s between them packed with stacked piles of small angular stones (illus 6). This face was exposed to a height of only 0.4 m, over a sector 1.5 m in length, in the northern arm of the trench. The base of the wall-face was not exposed. wale Th l core material comprise dseriea largelf so y unconsolidated t distincbu , t dump roundef so d boulder cobbled an s e naturs (illuTh f thi. eo 6) ss material varied trenche betweearmo th tw f o se , nth indicatin t formeg no tha seriea core s f th tdo concentri f ewa s o c deposits roughlo Tw . y coursed lengthf so boulder alignment were present within the core, in both cases facing outwards and running roughly concentrically with the line of the wall as a whole. The bases of both were identified, and found to rest on earlier wall core material. These alignments were flimsy and irregular by comparison with the internal and external wall-faces, and represent neither early outer faces subsequently buried within an expanded or buttresse e linindth f walintramurar o g no l l chambers. The e interpretear y s constructionaa d l devices perhaps intende ensuro dt stabilite ewalth e analogoud th an lf yo muruso st duplex style defencese on t A . point within the core immediately to the rear of the inner face a cluster of slabs forming an even surface had the appearance of paving, but no other evidence was identified to suggest that they formed the floor of an intramural chamber or passage. Alternatively, they may be explained as a temporary working platform used during the construction of the wall, possibly associated with the erection of the inner wall-face.

Internal features Within the structure a layer of brown sandy soil was revealed at a depth of 1.2 m beneath the surviving wall-head. This deposit was not excavated, and it may represent either the uppermost in situ occupation deposit withi structure nth r lateeo r materia seales l dumpewa dt I beneat. din hsubstantiaa l bod largf yo e sub-angular stones, probably representing collapse r dumpedo d rubble fro innee mth r wall- face.

External features A secondary face c 0.5 m wide abutted and partly obscured the outer wall-face . Thi6) (illus & featur4 s s partlwa e y comprise a roug f do h boulder alignment partld a lineaan , f yo r 288 | SOCIETY OF ANTIQUARIES OF SCOTLAND, 1999

/Rubble

Revetment

Drystone wall remains

Inner face

Midden Revetment \

Midden

Possible flagstone paving

ILLUS 4 Trench 2: plan and section of roundhouse wall arrangemen eartf o tstonesd han t appeareI . represeno dt t eithe buttressine rth roundhouse th f go e wallr o parstructura f to e abuttin roundhousee gth banA . middef ko n including discrete lumpabutteh as f so d this revetment. None of these external features was excavated, and no datable artefacts were recovered. DUNWELL: DURCHA, SUTHERLAND | 289

ILLUS 5 (top left) View of the elevation of the inner wall- face from south-east

ILLU S6 (top right) roundhouse Vieth f wo e wall from north-east, showing house extension as built

ILLUS 7 (left) View of Trench 3 from west

Trenches 3 & 4 (illus 3&7) In orde confiro rt m tha wale tth l identifie Trenchn di thasubstantias a f to wa 2 l roundhoused an , consequentl determino yt internae eth l diamete thif ro s structure smalo ,tw l trenches were opened on other north-wester e partth f so n summi moune th f o t d acros anticipatee sth d alignmene th f o t | SOCIET 0 29 ANTIQUARIEF YO SCOTLANDF SO , 1999

inner wall-face; this element of the wall was considered likely to be the most readily identifiable from limited investigation.

Trenc h3 Thi locates northere th wa s n do nmounde th sid poina f eo t a , t wher irregulan a e r linear arrangemen f bouldero t visibls swa e running approximately east/wes robbina face f th o e n i t g scar. This boulder alignmen founs wa trepresen o dt innee reae th f th tr ro wall-fac structure th f eo walle (illuTh -. s7) face was identical in construction to that identified in Trench 2, and survived to a level 0.5 m higher than the rubble wall core material backing it, of which only a fragment was exposed. Overlying the wall core and abuttin innee reath e gf sub-rounderth a ro facs ewa d boulder whic furthea f o h d former en alignmen e don t of four boulders visible from surface traces running downslope to the north within a robbing scar. This boulder alignment thus post-dated the robbing of the roundhouse, and is of modern origin. Within the wall circui same th t e sequenc depositf eo Trencn i identifieds s sa wa h2 , comprisin gsanda y soil overlaia y nb deposit of angular stones, c 0.5 m deep, tipping away from the wall-face. Finds from this trench were restricte ranga o dt moderf eo n artefacts including glazed pottery ,horseshoe a glas d san , fro rubble mth e within the wall.

Trench 4 This was opened to identify the western circuit of the wall. The inner face was revealed to be of similar character to that exposed in Trenches 2 and 3; a length of 1 m was exposed to a depth of 0.5 m. Loose, angular rubbl presens ewa t withi wale nth t removedls no circuit bass wa s it m t ewa 6 depta ;bu , 0. f ho reached artefacto N . s were foun thin di s trench.

Trench 5 (illus 3 & 8-11) This trench was opened in 1992. With hindsight from the roundhouse wall alignments identified in Trenches 2-4, it is clear that the pre-modern remains excavated here belonged to the southern side of the same large roundhouse. An area of 9 m by 6 m was examined immediately to the north of the modern house, with a slit trench, measuring 3.5 m by 0.8 m, extending westwards from its north-west angle (illus 8). A section of the concrete floor of the lean-to extension was not removed prior to archaeological excavation in the south-east comer of the trench, prohibiting any work beneat . Excavatiohit adjacenf no t deposits suggested tha significano tn t archaeological deposits were sealed beneat concretee hth . Safety considerations severely constrained excavatioe th n ni southernmost 2.5 m of the trench. Limited earth-moving which occurred immediately beside the trench afte completioe th r excavatiof no monitores nwa archaeologican a y db l representativf eo (former) Regional Council.

Roundhouse wall The primary archaeological features in the trench were the heavily truncated remains of the massive drystone wall of the roundhouse, and two associated laid surfaces or possible pavements (illus roundhouse Th . 8) e wallbeed ha n reduce foundatioo dt n level survived an , d eastere onlth n yi ne parth f o t trench. Its remains consisted of the outer face backed by large cobbles. A span of about 2.1 m was exposed consisted (illuan ) s9 carefullf do y chocked large, rounded boulders gape Th .s between them were tightly packed with coursed angular and laminar pieces of stone, an arrangement very similar to the external wall- face recorded in Trench 2. A large tabular stone set over two boulders (illus 9) appeared to represent the vestiges of a second course. The core material did not have regular limits to north and west, and no internal wall-face was identified.

Midde deposiA n middef o t n overlappe residuae dth l cobbl enorth-ease corth n ei trencte parth f to h (illus 10) & e surfac 8 .Th f thio e s midde s distinctlnwa y mounde d irregulardan , suggestin gbeed thaha n t i t DUNWELL: DURCHA, SUTHERLAND 291

Projected lin f outeeo r wall face

X

Post-hole (modern)

Midden

Wall core

Possible wall face

Wall face

, Paved surface

. Cobbling (extent) QM I 1

2m ~~

ILLU S8 Trenc composit: h5 e plan showing principal excavated features

disturbed. A sherd of coarse, undecorated pottery came from amongst the stones of the cobble core, and a fragmen stonf o t e bracele r armleto t , eight lump irof so n slapieced gan cattlf so e bone were present within middee th n (see Finds, below). 292 SOCIET ANTIQUARIEF YO SCOTLANDF SO , 1999

ILLUS 9 Trench 5: view of cobbling and elevation of wall-face

Midden Coarse sand Grey silty sand Wall core (possible) Wall face Paved surface Cobbled surface Wall core J 2m

ILLU 0 S1 Trenc sectio: h5 n A-B

Sandy layer Sealed beneat walle hth layea cors coarsef ro ewa , humic sand (illu structurelesss 10)wa t I . , abrupn a d tha boundard an underlyine th o yt g subsoil glaciaa , l till. Stratigraphic evidence on-sitn a d ean , palaeoenvironmental assessment conducted by Richard Tipping (pers comm), suggest two possible modes of formation: as a buried ground surface or as an accumulation of amorphous organic matter washed down from above origie layea .Th f npeatf o ro y soil sealed between thi cobble s layeth d eran wall cor unclears ei . DUNWELL: DURCHA, SUTHERLAN3 29 | D

Old ground surface Terrace edge Rough paving

Cobbled surface

1m

ILLUS 11 Trenc sectio: h5 D nC-

Outer paving The western continuation of the roundhouse wall within Trench 5 had been entirely removed by stone robbing. Its former alignment was suggested by the course of a roughly paved surface which ran immediately outsid survivine eth g wall-fac continued ean d acros trence sth gentl a n hi e arc, narrowing from (illum . Thi8 s8) 0. s t necessarilo pavint 1. no 5m s gwa yprimara y external featur roundhouse th f eo e wall or even directly associated with its occupation, but appears to have respected its alignment, and thus to have been laid before the wall was robbed. This pathway was exposed, and was possibly in use, when a modern rubble deposit was dumped over it (illus 10). The wall alignment suggested by the paving agrees closely with that extrapolated on the basis of the positions of the roundhouse wall identified in Trenches 2-5 (illus 3).

Pre-wall cobblin roundhouse Th g primar e th t eno y wals constructiowa l n identified withi trenche nth A . second and evidently earlier area of paving, or compact cobbled surface up to 3 m wide, ran across the trench and beneath the outer face of the roundhouse wall. This cobbled surface was defined to the north by a shallow terrace cut up to 0.2 m deep into the gentle slope to create a level surface beneath the paving; to soute th s originahit t definel extenno s dwa t owin truncatioo gt modery nb n activity associated wite hth standing building. This terrace ran on a slightly different alignment to that proposed for the roundhouse waladjacend an l t paved path, bot whicf ho h seale different a t di t points easter e (illu 11)th & t s8 .A n edgf eo trenche th northere th , ncobblee limith f o t d laye respectes rwa larga y db e sub-angular boulder embedded withi smallee nth r cobbl eroundhouse corth f eo e nort outes wallm it 9 f ho r0. , face (illus 10). Thiy sma represent the residual remains of a wall-face associated with the terrace and cobbling, which was subsequently incorporated withi e roundhousth n e wall; alternatively t couli , d simple th e par b yf o t roundhouse wall core.

Post-medieval features All other excavated remains in Trench 5 proved to be of post-medieval, and largely modern, origin l appeaAl . r likelassociatee b o yt d wit occupatioe hth currene th f no t house principae Th . l remains were a compact yard with a relatively level surface, comprising angular, freshly quarried stone dumped directly over the denuded remains of the roundhouse wall and associated features (illus 10). The decayed remain peayara e f th so dt n stacsurfaceo substantiay a kla d an , l post-hole, possibl washina r yfo g line post, had been cut into it (illus 8). Modern rubble abutted the roundhouse wall-face and extended 294 | SOCIETY OF ANTIQUARIES OF SCOTLAND, 1999

beneath the floor of the former lean-to extension. All these features can be dated by associated artefacts to the 19th or 20th centuries.

Parflaggea f to d surface, incorporatin greusea upturned dan d socket stone reveales ,wa d beneath modere th n yard surfac slie tth trencsupplementarn e ei th e h(i y cutting extending fro northe mth - west angl unweatheree f Trenceo Th . h5) d naturflagstonese th f eo contrasn i ,stonee th o t st formin pavee gth d surface withi maie nth ntrenche parth f o t , suggests tha t partt no thef se o yar same th e feature tha d flagstonee an ,th t slie tth trencmorn f si o e ehar recent origin, althougo hn direct dating evidenc recovereds ewa .

Trench 1 (illus 3) Trench 1 was excavated at the south-east corner of the mound, at the point where its edge formed a regular arc with a moderately steep slope up to 2 m high. It had been speculated on the basis of the 1992 fieldwork that this edge may have defined the south-east side of a substantial roundhouse. An arc of voided rubble 1 m wide (not illus), lying on top of the mound and roughly concentric wit edges hinterpretes it wa , 199n di possibla s 2a e gallery withi roundhouse nth e- wallL n A . shaped trenc openes hwa 199n di orden 6i interseco t r notionae th t l sout easd han t walld san galler roundhousee th f yo .

Modern disturbanc surface Th e e topograph moune th f yt o thid a s poin founs largele b wa t o resulde t yth t of modern activity. Immediately beneath the turf a substantial dump of large angular stones with a surface of smaller angular stones, up to 0.7 m deep, represented a continuation of the yard surface exposed in Trench 5 to the west. The putative gallery was found to he the fill of a cut into this surface; modern finds, including pottery foiln , ti glas , d wersan e testament recens it o st t origin.

Early deposits? Beneat modere hth n yard material further deposit featured san s were expose t fullno y t dbu excavated. In the north-south arm of the trench a sandy soil layer containing charcoal, probably midden material revealeds wa east/wes,e sand o th tw n I .y m layerar t s were identified overlyin f largo d e be sub ga - rounded boulders and more angular material. There was no evident structure to this material, although the area exposed was limited, and no datable finds were recovered from deposits sealed beneath the yard. Comparison betwee e leventh l reache excavatioy db n withi trence nth thad han t externa moune th o t ld suggests that up to a further 1m of stratified deposits are preserved below those exposed.

FINDS majorite finde Th th f s yo recovere d durin excavatione gth moderf o e sar prenw originfe - e Th . modern finds recovered frol ,al m Trenc compris, h5 ecoarsa e potsherd stona , e bracele armler o t t fragment, eight lump irof so nsmala slad glan assemblag animaf eo l bone. Summary reportr sfo material whic subjecs hwa specialiso t t t analyse presentee sar d here; fuller account l classeal f so s of material appea unpublishen i e rar d archive reports (Dunwell 1993; 1996).

Pottery (not illus) Derek Alexander A single plain bod thicky m recovere s sherdm ,wa 6 5- , d from betwee stonewale e nth th lf corso e in Trench 5. Slight thickening at one extremity suggests the proximity of the base or possibly the DUNWELL: DURCHA, SUTHERLAND 295

\

20mm ILLU 2 S1 Shal e bracelet fragment rim. A precise date cannot be attributed to the sherd, although the character of the fabric and the context suggest that a date somewhere in the first millennia BC or AD would be appropriate.

Shale or cannel coal bracelet (Him 12) Fraser Hunter A fragment of an armlet or bracelet, most probably of finely laminated shale or cannel coal, was middefoune th n di n overlyin wale gth l cor bee Trencs longn efragmene i m ha n m t Th I .. 0 h5 6 s i t snappe t bota f face s shearehdof on eha ends d dan , off s preserve.It d surfac smoothes ei d dan cross-sectionas roundedha t i d ,an fragmene lTh maximaa . s ha mm t 2 1 l y dimensionb m m 4 1 f so decorative groove (4 mm by 2 mm) running along the outer face. The extrapolated diameters of the original ite(internal) (externalm 11e m m ar 0 m 5 9 d .)an Such artefact t diagnostino e sar c foune periodo t ar Bronzn Norsd o i t an ,e eeAg contexts majorite .Th thesf yo e finds, however, occur in later prehistoric contexts. Few decorated examples of such artefacts are known.

Animal bone Nicola Murray Thirteen fragment animaf so l bone were recovered fro middee mth n deposit ove wale rth l corn ei Trench 5. Only two are identifiable to element and species, an upper right and an upper left cattle molar. | SOCIET 6 29 ANTIQUARIEF YO SCOTLANDF SO , 1999

INTERPRETATION The excavations have demonstrated that the north-western area of the mound is occupied by a substantial roundhouse wit approximatn ha e internal diamete norths it f 9.n ro (illu-1O m . 3) s east circuit (Trench 2) the roundhouse wall was revealed to be c 4.5 m wide. No entrances were identified during the excavation. Nothing of the internal arrangement of the structure was exposed, although undated midden deposits appea o extent r d acros e interioth s r beneath collapsed and disturbed wall stones. internae Th l diamete roundhouse th f o r e agrees very well with Pococke's description ni 1760 (Kemp 1887, 115), though his substantial underestimate of the wall thickness suggests it t leasa d t partlha y collapse visits innetime e hi th f .Th e o y r d b wall-fac e survives better, reflecting a pattern of greater robbing and collapse of the outer wall material. In Trench 2 its inner face should stand at least 2.1 m high, and the exposure of its surviving surface at similar levels in Trenches 3 and 4 indicates comparable preservation around the north and west sides. The quality s preservatiooit f n aroun e southerth d n circui s lesi t s s probabli clear t bu , y less good than elsewhere. Excavation Trencn si indicateh2 wale dth l composee corb o et dumpf do cobblef so d san boulders. There was no evidence for any intramural features such as galleries, stairs or chambers, although such featureundetectee li y sma d beneat exposee hth d surface. Discontinuous lengthf so rough coursing within the core may be explained as constructional features designed to maintain the stabilit wale th lf yboto h durin erectios git possibld nan y also subsequently. In Trench 5 the surviving wall core material was also found to consist of cobbles. Here, however, a deposit of midden containing artefacts of later prehistoric character was identified overlying the cobbles. Three explanations for this are possible: firstly, that it represents a constituen wale th lf o cort e (parallele t Crosskirkda , Caithness: Fairhurst 1984, 41); secondly, tha t representti s material redeposited fro interioe mroundhouse th th f ro e durin robbine gth f go its wall; and, thirdly, that it represents later midden, incorporating residual artefacts, dumped after the robbing of the roundhouse wall. On present evidence there is no secure means of differentiating between these possibilities. outee Th r wall-fac comparativels ewa y poorly preserved. Only foundatio s parit f to n course survived in Trench 5; its exposed surface in Trench 2 was 1.7 m lower than that of the inner face, and probably represented foundation level or a little above it. The distinctive appearance of the outer face is closely paralleled at the partly excavated vitrified dun of Tor a' Chorcain, Langwell (NGR: NC 4104 0084, illus 1, no 19; Nisbet 1996, pi 6), which a range of dating techniques (Gentles 1993, 19) suggest was occupied at the end of the first millennium BC. Limited evidence was identified for features immediately external to the roundhouse wall. A roughly paved walkway of uncertain date ran immediately outside the southern side of the buildin north-ease g th (Trenc o T . secondaryha 5) t , poorly built stone alignment abuttee dth outer face of the roundhouse (Trench 2), and was itself abutted by a midden deposit. Although undated, this feature could represent either the wall of an extra-mural building or part of a buttres r casinso g wall simila Fairhurst'o rt s (1984 Typ) 52 , eexampleI s recorde t Crosskirkda . Whateve functions rit constructios it , secondars nwa roundhouse thao yth t f to e wall.

Features preceding the roundhouse Trencn I , wherh5 l survivineal g deposits were excavated, features precedin constructioe gth f no the roundhouse wall were identified. These comprised a cobbled yard surface which had been DUNWELL: DURCHA, SUTHERLAND | 297

made near-leve e cuttin th a shallo y f b lgo w terrace inte gentloth y sloping sub-surfacee Th . similarity in the alignments of the roundhouse wall (as extrapolated) and the earlier terrace might indicate that the erection of the roundhouse site represented the development of an extant settlement; ther certainls ewa stratigraphio yn c evidenc suggeso et tima tbetwee p o ega tw e nth constructions. The indication of an earlier phase of activity allows the possibility that midden material overlyin cobblee gth incorporates swa cora s dea constituen roundhouse th f o t e wall during construction work on an existing settlement site.

South-east area of the mound The nature of the pre-modern remains present within the south-east part of the mound was not resolve excavatioe th y db Trencf no h 1. Layer boulderf so depositd san middef so n were exposed, althoug t datedhno pattero n ; identifies nwa indicato dt presence eth f structureseo likels i t I .y that a further sequence of deposits up to 1 m deep lies beneath the exposed level. Unfortunately, their chronologica structurad an l l relationshi roundhouse th o pt earlied ean r excavated remains are entirely unknown, and the remains cannot be cited as evidence for a settlement or structure associated wit roundhousee hth t least A .excavatio e th , Trencf no demonstrates ha h1 d that this moune th par f o td doe t simplno s y comprise material disturbed durin e robbine gth th f o g roundhouse.

Post-medieval activity roundhouse Th e wal sufferes lha d heavily fro effecte mth stonf so e robbing documentar;a y source (Kemp 1888 indicate) ,48 s that substantial degradation occurred durin 19te gth h century mas.A s of angular rubble up to 1 m deep within the roundhouse appears to represent the collapsed or demolished uppe rinnes parit f ro t wall-face, probably deposite same th t erobbinde a tim th s ea g of the walls for building stone. Subsequent to the robbing, angular rubble was dumped over the denuded remain roundhouse th f so esouthers walit n o l n side creato t , yarea d surface probably associated wit extane hth t dwelling-house. Ther mentioo n s e i moun e parf th n o f o td being used as a domestic yard in the Royal Commission's Inventory (RCAHMS 1911, 20), and it may be tha e yars createth t dwa d only after this date. Despite this post-medieval disturbancee th , excavatio s demonstratenha d that considerable evidenc roundhouse th r efo associated ean d structures survive sitee th .t sa

GENERAL DISCUSSION The scale of the excavated wall remnants at Durcha indicates the structure within the mound is an Atlantic roundhouse. Its construction is substantially more massive than the walls of even the more elaborat substantiad ean t circlenorthere lhu th f so n mainland, suc Fairhurst's ha s (1971) Typ identifieI eI t Kilphedirda circulae th d an ,r houses, probabl earle th yf y o firs t millennium BC, identified at Cnoc Stanger (Mercer 1985; 1996). The wall width (4.5 m) and the diameter of its internal court (9.1 m) both demonstrate its affinities with the Atlantic roundhouse tradition. These measurement criteria have been used previousl r investigatinyfo regionae gth l character- istics of'brochs' (eg Fojut 1982; MacKie 1965; 1971; Martlew 1983). Both measurements for the Durcha structure fall withi e mosth n t common range recorde e othe th 5 Atlanti 3 t ra d c roundhouse sites in Sutherland for which comparable data can be obtained (ie sites recorded in NMR brochss Sa ; liste Appendix)n di factn I . , Durcha lies very mea e closth no et dimensionr sfo 298 SOCIET ANTIQUARIEF YO SCOTLANDF SO , 1999

Wall widths (m)

5.5-5.9 7.0-7.4 8.5-8.9 10.0-10.4 11.5-11.9 13.0-13.4 Internal diamete) (m r ILLUS 13 Wall widths (above) and internal diameters (below) of selected Atlantic roundhouse site Sutherlann si d

the 13 eastern Sutherland structures (9.18 m internal diameter; 4.48 m wall width) considered by Martlew (1983, 257; table 3), using the data from MacKie's (1965; 1971) previous studies of a sample of 120 sites. The analysis presented here is coarse: the quoted dimensions for several sites are based upon only partly visible wall-faces, and the possibility that buttress walls have been measured, thus overestimating original wall width, mus borne tb mindn ei . It is clear that the width of the wall at such sites varies less than the diameter of the internal court. Of the 36 sites studied, 31 have a wall width within the range 3.5-4.9 m and 26 have an internal diamete (illum 9 s9. 13)rd betweeplottinn an .I m 0 gn8. thes characteristico etw s against each other (illus 14) the clustering of the sites is apparent, and again Durcha falls comfortably withi maie nth n cluster. However, therappareno n s ei t tren thin di s grap substantiato ht e eth claim by the Royal Commission (1911, xx) that 'as might be expected, the thickness of the wall increases with the diameter of the interior'. Assuming the wall width of the Durcha structure to be consistent around its circuit, the walls would form approximately 49.5% of its overall diameter, figura e which also accords well wit meae hth 49.38f no other %fo r eastern Sutherland Atlantic roundhouses (Martlew 1983, 257, Table 3), and also for Atlantic Scotland more generally. presentt A Durche ,th a structur classifiee b n eca d Atlantionln a s ya c roundhouse. However, takee minimab a this n na sca l statemen potentias it f o t l complexity certaio N . n evidencs ewa identified for an intramural gallery or cells, although the exposure of the roundhouse wall was insufficien demonstrato t t e that intramural t featurepresenno e t soma ar ts e point arouns dit circuit. It may be that the remains of a complex Atlantic roundhouse, such as Crosskirk, or even a broch tower, are present within the mound at Durcha. In this regard, an 18th-century description of the parish of Creich (in which Durcha lies) in Macfarlane's Geographical Collections Relating Scotland^o t interestf o : DUNWELL: DURCHA, SUTHERLAND 299

12-

10 -• Q) CO V

• •

4- -+- 2.5 3.5 4 4.5 5 5.5 Wall widt) h(m • Durcha ILLU 4 S1 Scatte r grap walf ho l widths plotted against internal diameterr sfo selected Atlantic roundhouse sites in Sutherland

Therseveran i e ear l place saif so d paroch [sic] circular edifice sston y buildr f eo t withou mortay tan r vera f yo largd an e top e buloped th , an k t n wita walle hth thico s k that ther severale ear l [sic] passages through them and apartments within the walls which are supposed to be idolatrous temples built by the Druids (1748, 202-3). sitee th Creicn i sf o t onle hA yon paris h currently identifie NMRe th y db brochs S a Dais— l Langwell (illus 1, no 1) — are intramural features known to exist. Macfarlane's account suggests that other structures, such as Durcha, may have featured degrees of architectural complexity which have either been destroyed through stone robbing or obscured beneath debris of robbing collapser o d wall material. The excavations at Durcha provided no material evidence from which the Atlantic roundhouse and associated remains could be accurately dated. It can be placed only within a broad chronological framework of Atlantic roundhouses emerging during the later first millennium BC and continuing to be occupied into the first millennium AD. The evolution of massive roundhouse constructio s beenha n extende d eve an bacn d earliek mi int e roth first millennium B OrkneCn i y through important excavation t sitea s s suc Quanterness ha s ( 1979), Bu (Hedges 1987) and Howe (Ballin Smith 1994). By contrast, understanding of the Sutherland structure bees sha n hampere lace botf th k o y dhb detailed investigatio surfacd nan e definitio architecturaf no l characteristics 'vitrifiee .Th dar 'dun ChorcainTo t a ' , Langwell (Nisbet 1996), has provided the only reliable dating evidence from Sutherland, suggesting construction in the fourth century BC or thereabouts and destruction in the first century BC or AD. The similarity of walling techniques between this d Durchsitan e perhaps i a s indicativ locaa f eo l building 300 SOCIETY OF ANTIQUARIES OF SCOTLAND, 1999

tradition, but cannot be used as evidence with which to date Durcha. Moreover, the Langwell structure, with an internal diameter of about 15m, timber-laced walls and a timber gateway, is clearly a variant of the drystone structures under consideration here (cf illus 13). As a consequence uncleas i t i , r whethe Sutherlane rth d structures represen ranga t formf eo s emerging over time s opposea , structuralla o dt chronologicalld yan y more unitary development. This restricts the extent to which they can be understood either as the result of a movement of people (cf Fairhurst developmenta 1984 s a r )o responsn i , migratino et g ideas locaf o , l structural forms includin e stronglgth y buil tt circlelatee Typth hu rf I eI o scenturie whicn i , hsBC Fairhurst (Fairhurst 1971; Fairhurst & Taylor 1974,93) detected features reminiscent of broch architecture. It may be that remains preserved beneath the surviving roundhouses, such as those fragments identified at Durcha, hold the key.

ACKNOWLEDGEMENTS The project was wholly sponsored by Historic Scotland, for whom Olwyn Owen and Richard Hingley provided advice throughout. lan Armit (latterly of Historic Scotland; now of Queen's University, Belfast) was CFA's Project Manager during fieldwork in 1992 and continued to give advice. Many useful comments on drafts of the report were received from lan Ralston. Bob Gourlay, JohDorothd an n w WoodyLo , Archaeologist Highlanr sfo d Regio subsequentld nan y Highland Council, provided helpful advice. Richard Tipping kindly provided useful comments upon the stratified sequence of soils in Trench 5. I would especially like to thank Mr and Mrs Mouat, the long-suffering owners of Durcha, for all their assistance and hospitality. The illustrations which accompany this paper were produced by George Mudie (1, 3,4, 8) and Martin Wilson (10-12). While thanks are due to the above, responsibility for the final form of this paper rests wit authoe h th CFAd ran .

APPENDIX

SITES CONSIDERE 4 ILLUDN 1 I & 3 S1 Notes 1 Sites where recorded wall widths varied substantially are not considered here; but slightly varying wall widths have been averaged. 2 Where wall widt calculates hwa d above scarcement level (*), basal wall width will exceed this. 3 Site italicn si s were considere MacKiy db e (1965).

Name Nat Grid Ref RCAHMS Wall Internal Wall: diam (1911)ref width diameter ratio (%) Inshlampie NC 7159 4657 178 4.5 6.5 69 Armadale Burn NC 7994 6266 190 4.5 7.0 64 Caistealna Coille NC 7957 1337 25 4.0 6.4 63 Backies, Golspie 834C N 5 0261 272 5.0 8.2 61 Liath,Cam Golspie NC 8704 0137 270 5.5 9.2 60 Kintradwell NC 9293 0807 467 5.5 9.5 58 Dun Chealmy NC 7199 5140 179 4.0 7.0 57 MhaighDun NC 5523 5303 527 4.9 8.9 55 Clachtoll NC 0366 2784 7 4.2 8.0 53 Coich Burn NC 7880 1087 23 4.2 8.0 53 DUNWELL: DURCHA, SUTHERLAND 301

Gailiable NC 9487 1823 311 4.4 8.5 52 Dun Viden NC 7265 5188 181 4.8 9.5 51 Skail NC 72014732 N/A 4.2 8.3 51 Balvalaich NC 9452 1898 N/A 4.5 9.0 50 Ach' an Duin NC 46016053 156 3.7 7.6 49 An Dun, Loch Ardbhair NC 1688 3323 4 3.4 6.9 49 Carnachaidn Du h NC 7213 5269 180 4.5 9.2 49 Durcha 0237 10 9 C N 52 4.5 9.2 49 Sallachadh NC 5491 0922 392 4.6 9.4 49 Baile Mhargaite 697C N 3 6097 184 4.1 8.5 47 Cam Liath, Farr 894C N 0 5233 187 4.5 9.8 46 Carrol NC 8462 0646 27 4.3 9.4 46 Altanduin NC 8097 2605 313 4.0 9.0 44 Dail Langwell NC41161121 49 4.2 9.5 44 Dun Creagach NC 6046 3558 175 4.1 9.4 44 East Kinnauld NC 7438 0159 477 4.1 9.5 43 The Borg 899C N 3 5095 186 4.0 9.4 43 Grummore NC 6107 3669 174 3.8 9.1 42 Alltbreac NC5911 1035 395 3.9 10.0 39 Coill'Acha' Chuil NC 6585 3815 176 3.7 9.3 39 Leadoch 855C N 0 0520 28 4.0 10.2 39 Kilphedir NC 9943 1891 307 3.8 9.9 38 Suisgill NC 8875 2530 308 4.5 12.0 38 Carn Bran NC 9420 1220 468 3.6 9.8 37 Feranach NC 8441 2730 314 3.4 11.0 31 Dun Dornadilla* NC 4572 4501 155 2.4* 8.2 29

REFERENCES Armit, 11990a 'Brochs and beyond in the Western Isles', in Armit 1990b, 41-70. Armit (edI , ) 1990b Beyond Brochs.e th Edinburgh. Ballin Smith, (edB ) 1994 Howe: four millennia of Orkney . Edinburgh (=Soc Antiq Scot Monogr Ser,9). Dunwell 199A , 3 Durcha, y ,b Sutherland, Highland Region. Archaeological Investigations, Archive Report. Edinburgh (= [unpub] CFA Rep, 105). Dunwell, A 1996 Excavations at Durcha, by Lairg, Sutherland, October 1996: Data Structure Report. Edinburgh (= [unpub] CFA Rep, 299). Fairhurst, H 1971 'Kilphedir and hut circle sites in northern Scotland', Scott Archaeol Forum, 3 (1971), 1-10. Fairhurst 198H , 4 Excavations t Crosskirka Broch, Caithness. Edinburg Antic So q= h ( Sco t Monogr. 3) , Fairhurst, H & Taylor, D B 1974 'A hut circle settlement at Kilphedir, Sutherland', Proc Soc Antiq Scot, 103, (1970-1), 65-99. Fojut, N 1982 'Is Mousa a broch?', Proc Soc Antiq Scot, 111 (1981), 220-8. Gentles, D 1993 'Vitrified forts', Curr Archaeol, 133 (1993), 18-20. Graham, A 1949 'Some observations on the brochs', Proc Soc Antiq Scot, 81 (1946-7), 48-98. Hamilton 196C R 8 J , Excavations t Clickhimin,a . Edinburgh. Hedges 198W ,J 7 Bu, GurnessBrochse th d an of Orkney, vols3 Bri. R Oxfort BA Ser = d( , 163-5). Joass, J M 1865 'Two days' diggings in Sutherlandshire', Proc Soc Antiq Scot, 5 (1862-4), 242-7. Joass, J M 1873 The brochs or "Pictish Towers" of Cinn-Trolla, Carn-Liath, and Craig-Carril, in Sutherland, with note othen so r northern brochs', Archaeol Scot, 5(1) (1873), 95-130. Kemp (edW , ) 1887 Pococke Tours,R Scotlandn i 1747, 1757, 1760. Edinburgh. 2 30 SOCIET ANTIQUARIEF YO SCOTLANDF SO , 1999

Kemp, W 1888 The Tour ofDr Richard Pococke through Sutherland and Caithness in 1760. Edinburgh. Love, P 1991 'Recent excavations at Cam Liath broch, Golspie, Sutherland', Archaeol J, 15 (1988-9), 157-69. Macfarlane 174W , 8 (Mitchell (ed,A ) 1906) Geographical Collections relating Scotlando t . EdinburghI . MacKie 196W E 5, 'The origi developmend nan broce th wheelhousd f han o t e building culturee th f o s Scottish Iron Age', Proc Prehist Soc, 31 (1965), 93-146. MacKie 197W 1E , 'English migrant Scottisd san h brochs', Glasgow Archaeol (1971)1 , J , 39-71. Martlew, R 1983 'The typological study of the structures of the Scottish brochs', Proc Soc Antiq Scot, 112 (1982), 254-76. Mercer J 198 R , 5 'Second millennium Bnorthern Ci n Scotland' Sprattn i ,Burgess & D , (eds)C , , Upland Settlement Britain:n i seconde th millennium scandafter. Oxford (=BAR Brit Ser, 143). Mercer, R J 1996 'Excavation of a succession of prehistoric roundhouses at Cnoc Stanger, Reay, Caithness, Highlands, 1981-82', Proc Antiqc So Scot, 126(1996), 157-89. Nisbet 199H , 6 'Excavatio vitrifie a t Langwellf a n o n ddu , Strath Oykel, Sutherland', Glasgow Archaeol], 19(1994-5), 51-73. Ordnance Survey 1874 'Sutherland', Sheet CII (1st edn; 1:10,560). PSAS 1873 'Donations to the Museum', Proc Soc Antiq Scot, 9 (1870-2), 52-9. RCAHMS 1911 Royal Commission on the Ancient and Historical Monuments and Constructions of Scotland Second Report and Inventory of Monuments and Constructions in the County of Sutherland. Edinburgh. Renfrew, C 1979 Investigations in Orkney. (=Res Rep Comm Soc Antiq London, 38). Smith, J A 1871 'Notice of the remains of rein-deer, Cervus Tarandus, found in Ross-shire, Sutherland, and Caithness; with notes of its occurrence throughout Scotland', Proc Soc Antiq Scot, 8 (1868-70), 186-222. Stuart, J 1870 'Report to the committee of the Society of Antiquaries of Scotland, appointed to arrange for the application of a fund left by the late Mr A Henry Rhind, for excavating early remains', Proc Soc Antiq Scot, 7 (1866-8), 289-307. Swanson, C B 1984 The problem of the problem of the brochs', Scott Archaeol Rev, 3(1) (1984), 19-21. Young, A 1964 'Brochs and duns', Proc Soc Antiq Scot, 95 (1961-2), 171-98.

This paper is published with the aid of a grant from Historic Scotland