Proc Antigc So Scot, (1998)8 12 , 255-271

Re-excavation of an Iron Age wheelhouse and earlier structure at Eilean Maleit, lan Armit* with contribution Brab A StrachaR y b sy& n

ABSTRACT Excavations were tidal the carried isleton out settlement of Eilean Maleit, previously excavatedby Erskine Beveridge in the early part of this century, to test the hypothesis that the site represented a wheelhouse built into an earlier Atlantic roundhouse or broch. It is clear from the re-excavation that the wheelhouse indeed was into earlierset an massive-walled dry stone structure, probably Atlantican roundhouse but almost certainly not a classic broch tower. The denuded condition of this early structure when wheelhousethe builtwas suggests that significanta period of timehavemay elapsed between occupationstructures.the two the of Publication thisof work sponsoredis Historicby .

INTRODUCTION Eilean Malei rocka s i t y tida soute l isleth Vallahn e o t sidth f eo y Strand (illus 1) vasa , t expanse of inter-tida lnort e sandth hn so coas Nortf o t h Uist (NCR 774F :N 8 7388, NMR7 S7 refF N : SE isle e 5) connectes i Th t. shor e substantiaa th y eo db t l stone causeway some 2m widd ean around 30 m long, and is accessible except at unusually high tides. The islet has an elevation of , somwitm ha steepe4 , rocky norther n onlna facyd slightlean y gentler southern approach (illu summi. Close remaine s2) th th Irowheelhous n e o ee t a li ntf Ag so cellulad ean r structures, partly cleared out and excavated by the antiquarian Erskine Beveridge in the early part of the 20th century (Beveridge 1911,207-9). During September 199 re-surveyes site 5th ewa partld dan y re-excavate attempn a n di o t t test the hypothesis that the wheelhouse, rather than being the primary focus of settlement, was a secondary structure, built into an earlier Atlantic roundhouse or broch tower. It was hoped that, as at the nearby Sollas wheelhouse, Beveridge had left unexcavated deposits in situ (cf Campbell 1991).

VALLAY STRAND WHEELHOUSES e areTh a aroun e Vallath d y Strann exceptionalla s i d y rich archaeological landscape with outstanding evidenc f formeeo r settlement patterns, fro latee mth r prehistori post-medievao ct l periods in particular (illus 1, Armit & Dunwell forthcoming). Eilean Maleit is one of three closely

* Historic Scotland, Longmore House, Salisbury Place, Edinburgh EH9 1SH | SOCIET 6 25 ANTIQUARIEF YO SCOTLANDF SO , 1998

5km

ILLU S1 Location map Foshigarry1 : Mhic Ba c;2 Connain Carr;3 y lochdrach Cno;4 cComhdhalacha Eilea;5 n Maleit Sollas;6 Udale th ;7 . (BasedOrdnancee th n o CrownSurvey© p ma copyright) ARMIT: IRON AGE WHEELHOUSE, EILEAN MALEIT 257

bank

roundhouse

10 20 60 80m ILLUS 2 Plan of the islet showing contours at 1 m intervals on summit; the areas marked by hachures, surroundin summite gth principalle ar , y bare rock

spaced wheelhouses excavate Erskiny db e Beveridg Vallae th n ei y Stran dearle areth yn ai parf to this century othere ;th s being Cno Comhdhalacca h (Beveridge 1911,200-6 Garrd )an y lochdrach (Beveridge 1931). While there are several other wheelhouse sites in the Vallay Strand area (illus 1), including Foshigarry (Beveridge 1930), Bac Mhic Connain (Beveridge 1931), Sollas (Campbel lUdae 1991)th ld (Crawforan , Switsud& r 1977)bees ha n t i ,suggeste d that Eilean closo tw es Maleineighbourit d an t distince sar beinn i t g secondary constructions within earlier structures of complex Atlantic roundhouse form (Armit 1992, chap 12): traces of what appear originall havo yt e been intra-mural gallerie celld discernible san s ar e withi structurae nth l clutter represented on the original excavation plans of each (illus 3). This suggested sequence, by which Atlantic roundhouses are remodelled as wheelhouses, forms part of the basis for the wider interpretation of the relationship between Atlantic roundhouse wheelhoused san Westere th n si n Isles particularn I . , along with radiocarbon dating evidence from wheelhouse sites like Cnip in Lewis, Sollas in North Uist and Dun Bharabhat in Lewis (summarized in Armit 1996) this interpretation provides support for the hypothesis that wheelhouse generalle sar y later than Atlantic roundhouse Westere th n si n Isles thatd ;broaan n ,i d terms settlemene th , t pattern with which wheelhouse associatee sar lates di r than that associated with Atlantic roundhouses (this does not coursef o , , mean that certain Atlantic roundhousesr fo , example fully developed broch towers like those at or Loch na Berie in Lewis, need not be contemporary with some wheelhouses). It was decided in 1995, therefore, to select one of the three Vallay Strand wheelhouses for re-excavatio orden i teshypothesio e t r th t s that these wheelhouses were secondary structures 258 SOCIET ANTIQUARIEF YO SCOTLANDF SO , 1998

within complex Atlantic roundhouses. Eilean Maleit was chosen primarily because the lack of apparent soil depth withi interioe structurne th th f ro e suggested that ther beed eha n o littln r eo backfillin Beveridge'f go s trenches thauntouched y an ,tan d primary deposits might, thereforee b , accessibl limitee th n ei d time available. Both Garry lochdrac Cnod han Comhdhalacca h appear to have a much greater internal depth of deposits which, on the basis of Beveridge's accounts of his excavations (1931, 1911), suggests that thebeed yha n backfille considerabla o dt e extent, perhaps to stabilize the piers.

BEVERIDGE'S EXCAVATIONS Beveridge's 800-word accoun excavations hi f o t t Eileaa s n Maleit (1911, 207-9 restrictes i ) d largely to a general locational description, a brief list of finds (see below) and a detailed verbal description of the complex morphology of the exposed structural remains. The latter, however, adds little to his plan of the excavated structure (illus 3). The report, unfortunately, includes no description of the nature of the excavated deposits; no mention of their depth, consistency or variability thers i r indication ea ;no limite th excavationf f nso o . Significantly, therattempo n s ei t disentanglo t e wha complee th t e appear b modere xo th t phasin e o excavatee t s ney th f go d remains as presented in the published plan, other than to note that the annexes were 'evidently to e considereb f secondaro s da y origin' (ibid, 209). Callander's recors discussionhi f do s with Beveridge about this site in 1914 appears to confirm that the possible composite character of the excavated structure was not considered even then (cf Callander's notes in Beveridge 1931, 33). The published plan appears to show a wheelhouse approximately 7 m in internal diameter, divide stony db e piers into nine bay varyinf so g shap sized evidencey an . Nonpieran e d th f seha o

i^!____*f tj> if atfab

ILLUS 3A Beveridge's plan of Eilean Maleit

260 | SOCIETY OF ANTIQUARIES OF SCOTLAND, 1998

for bridging lintels linking to them to the outer wall, such as are normally found at around waist to shoulder height in better-preserved wheelhouses (cf Armit 1996, chap 8). The two westernmost piers, however, abutte innee dth r wheelhouse wall. Beveridge appear havo st e been uncertain abou outee limite th th t f rso wall, indicatint gi only with a dotted line around the east arc. Elsewhere, the periphery of the wheelhouse was adjoine masa y cellulaf db so r structures. Despit sprawline eth apparentld gan y irregular forf mo these cells, however possibls i t i , discero et n traceintra-muran a f so l passage withi originan na l double-walled arrangement wit originan ha l overall thicknes f somso m5 .e4. Thi especialls si y soute are e cleawesth d f th hao an n tri arcs. Similar arrangements, equally disturbe latey db r cells, can be discerned at the neighbouring sites of Cnoc a Comhdhalach and Garry lochdrach where the circumstance excavatiof so recordind nan g were similar (illus 3). The Royal Commissio e Ancienth d Historican o an nt l Monument f Scotlano s d (RCAHMS) plan (1928, 88-9, fig. 142) differs in a few respects from that of Beveridge, particularl outedetaie e th th dispositio e f rn yo th li cell d san certaif no n internal piers. Although somewhad idi t t modif neae yth t circularit interioe th f y o shows ra Beveridge'n i t s planno s i t i , clear whether the site was formally re-surveyed or whether Beveridge's plan was simply modified clarified RCAHMe an th y db S surveyors RCAHMe Th . S plasubsequentls nwa y reproducey db Scott (1948 discussios hi n )i wheelhousef no related san d structures.

RE-SURVEY

WHEELHOUSE e re-surve Th e sit n 199th ei f yo 5 showed that Beveridge's original plad beenha n somewhat idealized althougmaie th nf o feature l hal s were represented. Judging fro accurace mth certaif yo n measurements in the original plan, such as the distance between the inner ends of the extreme east and west piers, it appears that Beveridge took a number of key measurements by tape to form a basic framework, and then sketched in other features by eye. Thus the disposition and relationships of interior features are generally fairly well represented (allowing for the distortion in overall shape discussed below) whil outee eth r walcellulad lan r structures away fro centre mth e are much less faithfully rendered. The most notable divergence, however, between Beveridge's plan and the 1995 re-survey was the shape of the interior of the building. Rather than the near-circle shown in the original plan, the interior of the structure was markedly D-shaped internally. This interior shape would have made the structure quite unsuited to the wheelhouse architectural form which depends on a complex channelling of weight stresses through the wall and stone piers, and which would have been most stable within a symmetrical circular interior (Armit, in prep). It is unlikely, therefore, that the builders of the wheelhouse would have opted for such an irregular shape if constructing the structure de novo, and more likely that the irregular interior was constrained by the form of primare th y buildin siteirregularite e th .Th n g o interior e th f yo , combined wit relativels hit y small size, might also suggest thawheelhouse th t t Eileaea n Malei rathes wa t r less grand than some others in the Western Isles, such as the nearby example at Sollas (Campbell 1991). In other details, however, there is less divergence between the 1995 plan and that of Beveridge smale Th . l wheelhous maximua d eha m internal diamete , divided f somm o r 3 e7. s a , Beveridge indicated, by nine piers into nine bays. The central area defined by the inner ends of these piers is, most unusually for a wheelhouse, markedly oval with maximum dimensions of around 3.4 m ENE/WSW by 2.7 m. The inner end of each of the piers appears to have been ARMIT: IROWHEELHOUSEE NAG , EILEAN MALEI | T26 1

defined by an upright block, as was the case with some, though not all, of the piers in the wheelhouse at Cnip in Lewis (Armit in prep). Despite the existence of breaks in the inner wall in the north-west (thought by Beveridge to originae beth l entrance NNEd )an , both leading through access passage exterioe e th th o st f o r building (the latter through apparently later cells), the entrance to the wheelhouse appears originally to have been to the WSW between the only two piers which appear to have been bonded int innee oth r wall-fac t theiea r basal levels. This interpretatio foundes ni d principalln yo the tendency at other wheelhouses, such as Kilpheder in South Uist (Lethbridge 1952, fig 2), for onle th y piers bonded intinnee oth r wall-fac those b o et locate entrancee th t da , seemingly reflectin desirga lengtheo et entrance nth e passag prevend ean t direct access fro entrance mth e aroun 'aislee dth ' which connect wheelhouse sth e bays. Thi supportes si t Eileada n Maleie th y b t lac evidencf kcontinuatioo y an r e fo inne e th f rno wall between thes bayso etw , even thouge hth presumed acces beeshas n blocke latea dat r site'stagthe se of use . Eilean Maleit thus adds a further example to the existing list of three wheelhouses with broadly west-facing wheelhouse entrances (Cnip in Lewis, Allasdale in Barra, and Clettraval in North Uist) recently cite s exceptionda generae th o t s l tendenc r wheelhouseyfo faco t s e east (Parker Pearso Sharpien& s forthcoming); indeed Eilean Maleit would join this list even no Beveridge's interpretatio north-wesa nof t entrance existencThe . suceof largha e minorityof west-facers (the list should als extendee ob includo dt t leasea froo tw mt Jarlsho Shetlandn i f ) perhaps suggests that re-evaluatio requires ni significance th f d o tendenc e th f eo y towards east- facing entrance wheelhousesn si . The features within the northern part of the interior are now much less easy to define and plan than in the remainder of the interior: the present ground surface rises steeply from the central interioe parth f to r toward innee sth r wall face around this arc, largely endobscurine th st bu l gal of the piers. Beveridge noted that the two northernmost bays (which he represents as being blocke t theida r inner ends) wer levea t ea l 'about eighteen inches higher' tha remaindee nth f ro the interior (ibid, 208), possibly suggesting that earlier deposit survivy sma e unexcavate thin di s area (there is nothing in the original report to support Scott's (1948) explanation that the northern piers wer underlyine stebuila th n n pi o t g bedrock). With hindsight these northern bays would probably have bee mose nth t promising area re-excavationr sfo t timbu , e constraints dictated that bays with less soil dept chosee hb n (see below). Trace intra-murae th f so l galleries shown (bu t apparentlno t y understood Beveridge'n )o s plan e agaiar , n quite distinct, althoug partn hi s modifie latey b d r construction, particularly aroun south-wese nortth e d d th structure e han th f o t . These would appea indicato rt originan ea l wall width of around 3 m (rather narrower than on Beveridge's plan) containing a gallery approximatel widem y1 .

THE ISLET lan Armi Richart& d Strachan As regard islee resth e tBeveridge'f o sth thert o d littls ei ad o et s description:

Befor surface eth disturbed s ewa , trace comparativelf so y modern erections were visible both upon its centre (opposit causewaye eth towardd an ) eass wesd sit an t t extremities; these being evidently remaine th dwellina f so outbuildingd gan s erecte somy db e crofte utilised cottar ha o ro d rwh olde r material whic informe e readhands y ar hhi la e o yt .W d tha likn ti e manner these modern erections— alss causewaa e oth y itsel havf— e been quarrie servo dt constructinn ei bridge gth e which crosses the outlet of Loch nan Clachan (1911, 207-8 ). | SOCIET 2 26 ANTIQUARIEF YO SCOTLANDF SO , 1998

There is now, as Beveridge suggests, no trace of these later structures, nor are there traces any longe possibla f ro e wall alon soute gth hislee sidth tf eobserveo Beveridgy db e (ibid, a 208)f o r o , smaller islet wes'close Eileaf th o t o et n Maleit. wit. . h apparent trace formea f so r occupation' i

(ibid, 209) erodinn A . g clum turf po f clingin bare maith e ewese o g th t roc e nf th b o tisle o ky t tma : all that is now left of this latter feature. The only features now visible on the islet aside from the wheelhous adjoinind ean g cell mouns w arelo :a d immediately north-eas wheelhouse th f to e which appear Beveridge'e b o st s spoil-heap ovan a ; l depressio north-wese th o ne t th som m f 0 o t1 e wheelhouse whic represeny hma remaine tth curvinstructurea w f slo o a gd turf-covere;an d bank or wall running approximately NW/SE for a distance of some 12 m on the western periphery of theislet(illus2). Running approximately north/south from the shore, the stone causeway now comprises little more tha nsingla e cours largf eo e flattish slabs, stopping shor islebote f to shore th t hth d ean by a distance of a few metres. It is possible that the causeway is a relatively recent feature constructe enablo dt robbine eth stonf go e fro isletdrownine e mdate th th f th Valla:e eo th f gyo Strand is presently unknown and the islet may very well have been a knoll in a dry machair landscape when the wheelhouse was occupied. Work in progress on the settlement and environmental history of the Vallay Strand may in time shed some light on this question (Armit & Dun well forthcoming).

EXCAVATION lan Armi Alat& n Braby 199The 5 excavations comprise relativeldtwo y small conjoined trenches reopenea one ; d areaof wheelhouse th e interior sectioothee w th ne d ra n ,an throug wheelhouse hth e wall ful(illuA l. s5) stratigraphic excavationrepore th n to bees sha n deposited wit Nationae hth l Monuments Record of Scotland (Armit 1997) and what follows is a summary based on that report.

WHEELHOUSE INTERIOR An irregularly shaped section of the north-eastern wheelhouse interior, measuring approximately 3.8m north/south by 2.8 m in maximum extent, was reopened to assess what deposits, if any, remained within the structure and to determine whether the wheelhouse itself was the primary construction on the site. The trench encompassed one complete bay, a complete pier, and part of a second bay (illus 4-6).

Excavated deposits It was quickly established that, in this sector at least, Beveridge had been disappointingly thorough, and had excavated to the decayed bedrock surface. The only exceptions to this pattern of destruction were around extreme th e peripher interiore th f yo , unde pierse betweed th r an , truncatepiere w walld nth fe san A .t dcu features which had been dug into bedrock were also preserved. The turf over the interior had apparently re-established over Beveridge's trenches, which do not appear to have been backfilled. It represents, therefore, a homogenized mixture of recent aeolian deposits limitee th d dan surface deposits lefBeveridgey b t placen I . s this deposit directly overla decayee yth d surface of the gneiss bedrock. A black peaty deposit covered much of the exposed central interior under the turf to a depth of up to . Thi100smm deposi interpretes i t recens da post-Beveridgee (i t derived an ) d from homogenized trample ARMIT: IRON AGE WHEELHOUSE, EILEAN MALEIT 263

/.^ ^""^ '. ''xV'fe

:^"''

•,\\'*'''' iV>>' % 'A oV'. ' '.'

C7 ^ ^C7 ^,fo4^.^& ^. °>w//

lLj"^X or

ILLU S4 Pla interiof no r showing trench positions, base re-surven do 199n yi 5 inwasd an h from exposed deposits. However smala , l containinh patcas f ho g charcoal fleck somd an s e pottery fragments lay above it, suggesting some doubt over its recent derivation. Although it is possible that peate th y deposit represents residual wheelhouse occupation material expose disturbed dan Beveridgey db , charactee th deposie th f ro t tend suggeso st t thamors i patct h ti eas likelmodere he b redeposite th o yt d nan d (perhaps by a Victorian shovel). wheelhouse Onth f eo e pier fulls swa y expose reopenee th n di d trench (illus 5-7) t comprise.I d several large flat stones, separated by a gap of some 0.6 m from the inner wall, with an upright block at their inner end. This bloc supportes kwa deposia y db f smalo t l angular packing stones darA . k silty deposit some 80m mmaximun i m dept s preservehwa d only unde e pieth r r (which Beveridge appear o havt t e no s disturbed). This deposit provided some evidence that the pier (and thus the wheelhouse) was secondary (illus 7). Other truncated deposits survive limitea o dt d extent withi interioe wheelhousene th th f ro gravelly.A - loam deposi preserves twa d directly unde ture rth f aroun extreme dth e peripher interioe th betweef d yo ran n unded stonean e innee rth th f rso wall-fac wheelhouse th f eo likels ei t (illuI y. thas7) t this deposit formerly extended across the interior but has been truncated by Beveridge's excavations except at the extreme 26 | 4SOCIET ANTIQUARIEF YO SCOTLANDF SO , 1998

ILLUS 5 The excavated area from the west

periphery. Althoug sectioe th n hi n throug interioe hth r this gravel deposit appear abuo st innee th t r wall- face (it also abuts the wheelhouse pier), elsewhere it clearly underlay it. These stratigraphic relationships, together wit e consistenchth e deposith f yo t itself, sugges deliberata s t thawa t i t e construction deposit, possibl ylaia secondarde flooth r fo r y wheelhouse. innee th Betweepiee rd wall-facth ran f outeo e nd th ren e Beveridge's excavations appea havo rt e left in plac efurthea r serie truncatef so d deposits (illu . Undes7) gravee rth l surface described abovthia s n ewa ashd lenre deep,f m somso m ,0 eoverlyin2 gblaca k peaty silt some deep10 m lattee 0m .Th r clearly underlay innee th r wall-fac wheelhousee th f wheelhouseo e th d ,an e pier, suggesting tha might ti t represen survivinta g floor deposit or abandonment deposit associated with the primary (ie pre-wheelhouse) structure. A few features were identified immediately under the regenerated turf and cut into the underlying bedrock. A small, oval pit, some 0.1 m deep and close to 0.5 m in maximum diameter, lay between the ends of the two piers exposed in the trench, slightly closer to the northern of the two. Its uniform dark brown, peaty fill containe potter1 d2 y sherds (see below), includin decoratee gon d wit applien ha d wavy cordon: this was the densest concentration of pottery found on the site. Immediately adjacent to the north of the pit was patca smalf ho l angular stones fillin truncatega surfacn i d stake-holm m e 0 are3 jus y y ab b ta e m somm 0 e4 few millimetres deep truncatioe Th . n cause Beveridge'y db s excavations, however, make t impossiblsi o et link these feature phasstructuree y th an f eo o st .

Interpretation Despite the disturbance caused by Beveridge's excavations it is clear that the inner wall-face of the wheelhous exposee th d edan pier were secondar seriea o yt earlief so r deposits, presumably associated with an earlier structure on the same site. This hypothesis was confirmed and expanded by the examination of the wall section (see below). ARM1T: IRON AGE WHEELHOUSE, EILEAN MALEIT I 265

stake-hole packing

pier

4m ILLU Sexcavate6 e Plath f no d area 266 SOCIETY OF ANTIQUARIES OF SCOTLAND, 1998 ARMIT: IRON AGE WHEELHOUSE, EILEAN MALEIT 267

Although only a few traces of pre-wheelhouse deposits had survived the destruction caused firstly by the construction of the wheelhouse and latterly by Beveridge's excavation, the wheelhouse pre-existin n walbeeo d t ha ln se g deposits whic beed hha n consolidated wit laiha d gravel surface, apparently at the time when the wheelhouse walls and pier were constructed. The surface was apparently laid after the construction of the piers (it was not present under the exposed pier t befor)bu accumulatioe eth deposity wheelhousee flooe an th th f f no o rn t so i d an , fille intersticee dth s withi unded nan lowese rth t cours walf eo l masonry traco N . wheelhousf eo e floor deposits survived Beveridge's excavations in this sector of the building unless the effectively unstratified cut features can be linked to this phase of occupation.

WALL SECTION sectioA n measuring approximately 3.8 orientated 1y mb man d approximately NE/Ss Wwa excavated outwards fro innee mth r wheelhouse wall-fac orden ei determino rt constructioe eth n asseso t walevidence y d ford th san an l f an m o multiplr efo e phase construction (illu . Thess6) e deposits were apparently undisturbed by Beveridge's excavations, although it is likely that this area of walling had been severely robbed in the last century or earlier.

Excavated deposits Under a superficial covering of turf and topsoil, up to 150 mm deep, was sealed a series of deposits relating secondare th o t y wheelhouse waldepositd an l s abuttin t whicgi h appeare havo dt e formed from eroded wall materia uppermose l (illuTh . 7) s spreaa f thes o s t f loamyewa do , mixed soil wit hhiga h decayed stone content up to 100 mm deep. This overlay a thicker deposit of similar soil, possibly turf-derived, and again with numerous decayed stone fragmentsm5 0. deep o t .p u Bot, f thesho e deposits butte againsp du e th t secondary wheelhouse external wall-face and tailed off towards the east, suggesting that they were formed of material (possibly a turf superstructure) which had collapsed and/or eroded outwards from the wheelhouse wall possibls i t .I e that this materia beed lha n disturbe removae th y db largef lo r stones during stone-robbing operation lase th t centurn si earlierr yo . The wheelhouse wall was revealed in this section to be faced both inside and out by walls, or rather revetments, buil f mediuo t mo largt e sub-angular stones e inneTh . r wall-face survive fivo t d e courses standing highsomm 8 ,e0. whil outee eth r revetmen mors littld wa t eha e broke d survivinan p nu g coursing (illu . Thihavy s7) sma e bee resule nth f stone-robbino t g having remove accessible dth e substantial stones fro outee mth r revetment wall-core Th . e materia formes seriea wa l y df depositsb so uppermoste th : a o t , depth of some 200 mm, was closely similar in composition to the upper external erosion deposits. Below this wa a quits e different material a red-brow; n ashy silt apparently derived from domestic midded an n containing pottery fragments and numerous charcoal flecks. This was preserved up to a depth of 0.85 m and overlay a darker brown sandy silt which was not fully excavated. Under thes ecentrsecondare e depositsth th f n ei o d yan , wall-cor reveales ewa dsubstantiaa l angular boulder largo to , exposo et etrenche fullth n i y , wit 0.2y hb measurement5m m 4 0. y b m excesn 7 si 0. f so (illus 7). This was significantly larger than any of the stones used in the construction of the secondary wall, and clearly not bedrock: its position under the core and set back from the inner wall-face suggested that, rather than being part of the secondary wall, the latter was built over it. The implication, therefore, is that it formed parmassivela f o t y built stone structure pre-datin wheelhousee gth . Assumin casee e th gth e , thib o st wheelhouse wall seem havo t s e been buil faced an t d slightl fron yi originae mth l face, thus reducine gth internal diamete buildinge th f ro . East from the wheelhouse wall the erosion deposits could be seen to bank up against a further series of deposits which appea foro rt msligha t moun r risdo e externa secondare th o t l y wheelhouse wall (illu. s7) These deposits comprise a series of compact dark sandy silts with a combined depth of at least 200 mm (they 26 | 8SOCIET ANTIQUARIEF YO SCOTLANDF SO , 1998

fulle coulb yt excavated)dno positioe .Th thesf no e deposits shows that they coul t havdno e derived directly from erosion of the secondary wheelhouse wall. They may, therefore, derive from an unexcavated and unrecorded feature to the east, or they may possibly mark the position of an outer wall of the primary structure lattee ;th r interpretatio supportens i positioe th depositse y th db f n o woul d ,an d suggest thae th t primary structure was formed of two concentric walls. The lack of obvious stone content, however, means that this canno demonstratee tb d conclusivel thin yi sitese areth .f ao

Interpretation The wheelhouse appears to have been constructed by clearing out and remodelling an already abandone d significantlan d y reduced stone building e internaTh . l occupation deposits were scoured, heaped onto the residual walls, and faced with angular stones to create a near-vertical inner wall-face on a slightly reduced diameter; while the outer wall-face may have been a less formal revetment (although the conditions of preservation are such that this cannot be demonstrated conclusively). While there is some evidence for an earlier massive-walled structure, and the evidence for a double-walled construction is at least hinted at in the excavated section, there is insufficient evidenc categorizo et e conclusivel earliee yth r building. Limitation timf dictatesd o eha d thaa t rather poorly preserved stretc f walchoses ho wa lr re-excavationnfo . With hindsigh clearea t r picture of the earlier structure would probably have come from the more upstanding parts of the walling where there is surface evidence of intra-mural galleries and cells. While some of the more massive stones from the original structure may have been used to creat wheelhouse eth e piers t appeari , s tha wheelhouse th t e walbuils smallef wa lo t r stones. This might suggest, along with the residual state of at least part of the wall circuit of the earlier structure, tha tmassiv e mosth f o t e stone alreadd sha y been removed fro site mth e e prioth o t r constructio wheelhousee th f no alternativn A . e explanation, however, migh thae earliee b t th t r structure was unfinished. After the abandonment of the wheelhouse, the wall seems to have collapsed (or been pulled) outwards, creatin gmouna materiaf do l abuttin lowee gth r course externae th f so l wall/revetment. It is impossible to tell, from the limited area of re-excavation, when this might have occurred and whether occupation might have continued elsewhere on the site after the collapse of the wheelhouse (perhaps in the maze of cellular structures visible on the surface). Local traditions recorded by Beveridge (1911, 207) suggest that the site was thoroughly robbed of its remaining stone in the 19th century to build the nearby bridge over the outlet from Loc Clachan n probabls i h na t i d an ,e tha larga t e quantit collapsef yo d stonewor removes kwa d from the interior at this point (assuming that the wheelhouse was constructed in the usual fashion by corbelling of the bays, there would have been a substantial amount of collapsed stone in its interior). Stone may also have been dug out of the peripheral areas of the mound at this time, possibly disturbing uppesome th f ero external deposits.

FINDS A handful of the finds from Beveridge's excavations at Eilean Maleit are illustrated in the excavation report (1911) and some, including most of those illustrated, survive in the Beveridge Collectio Nationae th n i l Museu mScotlanf o 54-8T d G (ca s , 253t no 595-617)d an , . Somf eo decoratee th dPiggot M potter copieC drawingd e 54-8T s drawth tan s yf (G Mr so )wa y e nb s ar Nationae helth n di l Monuments Record f Scotland,o likelys i t I . , however, thasurvivine th t g ARMIT: IRO WHEELHOUSEE NAG , EILEAN MALEIT 269

ILLUS 8 Pottery from Eilean Maleit photographed by Erskine Beveridge and originally published in North Uist (1911) finds represent only a very small proportion of the material originally excavated from the site. Beveridge summarize materiae dth followss la :

Hammer-stones were scarce at Eilean Maleit, but one deserves special notice. This is composed of hornblende, and, besides showing marks of considerable use upon its ends, has a symmetrical groove middle th t eacf a eo h side evidentlwhers wa t ei y hafted. Many fragment f ancieno s t pottery were found, severa thelof m ornamente varioudby s patterns piecone , e being perforatea by d nearim rits small hole, clearly for the purpose of suspension. A lump of pumice is to be noted; as also a thin disc of schist, an inch and a quarter in diameter, which shows distinct evidence of its edges having been chippe shapo dt e (1911,209).

In the absence of any contextual information there is little that can be usefully added about this material other tha highligho nt broae tth d characteristic ceramie th f so c assemblage. The range of hand-made fabrics is typical of the Hebridean Iron Age but the quantity and conditio survivine th f no g sherds doe permit sno assessmeny tan range th formf f eo o t s originally present in the assemblage. One of the surviving sherds was perforated (GT 55, illus 8, upper middle). The decorative motifs comprise principally applied wavy cordons: GT 54 has two such cordons aroune on , necks dit , belo wshora t everte lowee don rimrd dowan , n (illu , extrems8 e left). The form and the pattern and distribution of decoration closely parallel vessels from Phase 2 at Cnip (MacSween, in prep) and probably dates to the early centuries AD, although similar double cordons are found on much later, probably sixth-century AD flaring rim vessels in Phase 3 | SOCIET 0 27 ANTIQUARIEF YO SCOTLANDF SO , 1998

t Eileaa n Olabhat, als Norton i h Uist (Armit, Campbel Dunwelll& , forthcoming). Another sherd illu, botto, 58 s9 T (G m right applie n pars a f )ha o t d circl closels i banr eo d dyan parallelea y db sherd fro Allasdale mth e wheelhous Barrn ei a (Young 80)o 1953n , . ,96 The other sherd of note from Beveridge's collection is GT 56, where a series of incised infilled triangle visibls si e abov applien ea d cordon which appear maro st widese kth te parth f o t vessel (the sher shows di n upside uppee dowth n rno righ f Beveridge'o t s photograph, illu. s8) This typ f motio e particularls i f y commo Perioe th n ni d Bl assemblag e nearbth t ea y Sollas wheelhouse, which earldatee th yo st centuries AD (Campbell 1991 vesselg ,e 232). Finds from the 1995 excavations were restricted to a small pottery assemblage (53 sherds), one piece of ferrous slag from the wheelhouse wall core, and one piece of pumice from the topsoil. The pottery comprised generally small, abraded and undiagnostic sherds but also included two body sherds with fine applied wavy cordons; one from the wheelhouse wall core and another from the truncated pit in the interior. These sherds would suggest a date range similar to the material from Beveridge's excavations. A notable concentration of pottery (21 sherds, ie almost half of the total assemblage founs )wa d withi truncatee interioe nstructureth e th th n i f t r o deposi da pi n i , t reminiscent of the votive pits in the floor of the wheelhouse at Sollas (Campbell, ibid). No bone or other organic material was recovered. Further details of the pottery from 1995 excavations can be found in the archive report (Armit 1997). decoratee Th d sherds appea indicato rt e that ther soms ewa e occupatio site eth f durinno g the later centuries BC and/or early centuries AD. The pottery characteristics cannot, however, be use o dat dt e chronologica eth l e sitlimitth e f sequenco s e even approximately e apparenth : t absence of later pottery, for example, may simply indicate that less obviously interesting or attractive materiat collecteno s r wa retaineld o need t meadan dno n that occupatiot no d ndi extend into the immediately pre-Norse period.

DISCUSSION The recent work at Eilean Maleit has considerably expanded the information available from earlier records wheelhouse Th . bees eha n shown definitivel secondara e b o yt y constructiod nan havo t e bee nfairla y poorly constructed building relativ otheo et r wheelhouse vicinitye th n si s It . small sizirregulad ean r shape suggest that, althoug t coulhi d have supporte dsuperstructura e similar to (though smaller than) that of grander wheelhouses, like that at nearby Sollas, it is unlikely ever to have reached the same height and degree of monumentality: nonetheless, even comparably small wheelhouses like that at Cnip in Lewis would have had roofs up to 5 m above their floors and were clearly far from simple, utilitarian buildings (Armit, in prep). impossibls i t I sayo et t presenta , , what such structural variation might have mean termn i t s sociaf o perhapr lo s chronological distinction between these sites finde .Th s assemblage, whilt eno especially copious or distinctive, suggests a broad contemporaneity between Eilean Maleit and other excavated wheelhouses in the last centuries BC and early centuries AD, although the absence f contextuao l information fro e earlmth y excavations prevent y definitivan s e attributiof no particular finds to particular phases of occupation. It is also clear from the re-excavation that the wheelhouse at Eilean Maleit was set into an already ruined (or, less likely, unfinished) structure, probably buil f massivo t e angular stones. Little else can be said, however, about this earlier structure from the limited excavations carried out, althoug evidence hth consistens ei t wit initiae hth l hypothesi massive-wallea s s thawa t i t d Atlantic roundhouse. The apparent presence of intra-mural features at Eilean Maleit and at the two neighbouring wheelhouses, an interpretation strengthened by the re-survey, certainly suggests ARMIT: IRON AGE WHEELHOUSE, EILEAN MALEI | T27 1 attributio thio nt s clas monumentf so relativelbasis e it th f so n O . y smal irregularite th l sizd ean y of shape imposed by the site topography, it was, however, almost certainly not a classic broch tower. highle Th y fragmentary stat thif eo s earlier structure time th et a ,whe wheelhouse nth s ewa built, suggests eithe passage rth considerabla f eo e perio timf do e betwee occupatioe nth e th f no two structures, or perhaps that the earlier structure was never completed. The presence of pre- wheelhouse midden and occupation traces tends to support the first hypothesis, although such extreme denudation of the earlier structure at so early a period is quite unexpected and difficult to parallel.

ACKNOWLEDGEMENTS Permission to excavate was kindly granted by the owner, Mr N Macaulay. Thanks are due to excavation staff, Alan Braby, Nicola Battley and Catriona Leask whose efforts in extreme conditions, both tidal and entomological, are gratefully acknowledged. The contour survey and data-processing was carried out in collaboration with Richard Strachan. Thanks are also due to Andrew Dun weld Noean l l Foju r theifo t r comment n earlieo s r draft f thio s s papere Th . illustrations are by Kevin Hicks.

REFERENCES Armit, 1199 Latere 2Th Prehistory Westerne oth f Isles of Scotland. Bri R Oxfort BA Ser = d( , 221). Armit, 11996 The Archaeology of Sky e and the Western Isles. Edinburgh. Armit, I \991Eilean Maleit, North Uist: an archive report on the 1995 excavations. Unpub archive report, NMRS. Edinburgh. Armit pren i I ,p Anatomy of an IronRoundhouse:e Ag Cnipe th wheelhouse excavations, Lewis. Armit Campbell, I , Du& n,E well forthcominJ A , g 'Excavatio Iron a f nno Age , Early Histori Medievad can l settlement and metal-working site at Eilean Olabhat, North Uist'. Armit Dunwell& I , forthcominJ A , g 'Archaeological field surve Nortn yi h Uist'. Beveridge, E 1911 North Uist. Edinburgh. Beveridge, E 1930 'Excavation of an earth house at Foshigarry and a fort, Dun Thomaidh, in North Uist', Proc Antiqc So Scot, (1930)5 6 , 299-357. Beveridge 193E , 1 'Earth house t Garrsa y lochdrac Mhic Ba c d Connaihan Nortn i h Uist', Proc Antiqc So Scot, 66 (1931), 32-67. Campbell 199E , 1 'Excavatio wheelhousa f no othed ean r irostructuree nag t Sollassa , NortC hJ UistR y b , Atkinsonin 1957', Proc Soc Antiq Scot, 121 (1991), 117-73. Crawford Switsur& A 197 R I , V ,7 'Sand-scapin C14d gUdale an th : , North Uist', Antiquity, (1977)1 5 , 124-36. Lethbridge, T C 1952 'Excavations at Kilpheder, South Uist, and the problem of brochs and wheelhouses', Proc Prehist Soc, 18 (1952), 176-93. MacSween, A in prep The pottery', in Armit, I (in prep). Parker Pearson Sharpies& M , forthcomin,N g Between Land Sea:d an excavations Vulan,n Du t a South Uist. Sheffield. RCAHMS 1928 Royal Commission on the Ancient and Historical Monuments of Scotland The Outer , Smalle th Skyed Isles.an Edinburgh. Scott, L 1948 'Gallo-British colonies; the aisled roundhouse culture in the north', Proc Prehist Soc, 14 (1948), 46-125. Young, A 1952 'An aisled farmhouse at the Allasdale, Isle of Barra', Proc Soc Antiq Scot, 87 (1952), 80-106.

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