Breachacha , Coll: Excavation Field san d Survey, 1965-8 by D. J. Turner and J. G. Dunbar

In 1965 Breachacha Castle, Isle of Coll, was purchased by Major N V MacLean Bristol, a descendant of the family of MacLean of Coll. Major Bristol decided to restore the castle for residential use, but suggested that the castle should first be resurveyed and, where appropriate, that excavation shoul carriee db d out. Restoration work was factn i , , re-phase alloo dt w excava- tio continuo nt e over four seasons unde directioe th r Turner M extende o f no Tw . d visits were also made during this perio Royae th stafy db f o lf Commissio Anciene th n Historicad no an t l Monuments of Scotland, who completed a survey of the standing buildings under the direction of Mr Dunbar. The excavations were financed in part by grants from the Society of Antiquaries of Scotlandgenerosite parth n i y d b t an ,Majof yo r Bristo others d pleasura an ls i t I . recoro et d excavator'e th s gratitud thir efo s support. Volunteer labour onlemployes ywa d throughoutd an , the excavator is grateful also for the enthusiastic assistance of the many volunteers who took part Commissionere .Th s have agreed tha survee tth y report prepare Commissioe th r dfo n should be freely compilatio e useth n di drawingthie f no th s l paper al photograph d d san an , Crowe ar s n Copyright except figs 5-9, 12-14 and pi 17.

INTRODUCTION e castlTh e , fig(p13 si 1-4) stand srock w upolo yna promontor heae th f Loc do t ya h Breachacha, formerly called Loch-na-Chaisteal, about 3 miles from the SW tip of the island of Coll site eTh . enjoy s certain natural defensive advantagesd an E ,s beinit n o g a protectese e th y db S sides, and by an area of marshy ground to the W. On the N side, however, which forms the main landward-approach groune th , todas di y firopend man , rising gently toward extensivn sa e tract of machair situated at an elevation of about 50 ft above sea-level. Thus the castle is not advantage- ously situate landware th n do d side. The site has evidently been chosen primarily because of its accessibility to sea-borne traffic protectioe th afforn d ca an beache o t t di n d vessels. Loch Breachacha itself form conveniena s t anchorage, while small craft can be drawn up at high tide in the little sandy bay immediately to the SE of the castle, through which the Breachacha Water flows towards the sea. The pro- montory upon whic castle hth e stands extends southward roca s ksa spine alon westere gth n sidf eo this bay, and at the outer extremity of the promontory there may be seen the remains of a roughly constructed breakwater, which appears formerl havo yt e returne easterln a n i d y directioo t n shelter the bay on its seaward side. Evidence obtained during the present excavations suggests that the level of the ground on the W side of the promontory has risen during comparatively recent times, and that the sea may formerly have come much closer to the castle on this side than t doei s today. 15 | 6PROCEEDING S OF THE SOCIETY, 1969-70 There appears to be no provision for water-supply within the castle, but there was a spring , whilit immediatelf o e otheN e rth springo yt s discharge alon adjacene gth t foreshore.

C. l^O-^O1 primary C. l4-jO-$0'- secondary I later &•5 1 century ! i6&. century M late i6t(i. century 3 early ij&- century M late ijth- century \ late iSth. century 223 later

Tirst-TloorTlan

Ground-TloorTlan (0 2O 30 40 50 M 1O FIG 1 Ground- and first-floor plans of castle TURNE DUNBARD RAN : BREACHACHA CASTLE; EXCAVATION FIELD SAN D SURVEY, 7 1965-15 | 8

SUMMARY OF ARCHITECTURAL DEVELOPMENT architecturae Th l developmen castle th f complexs ei o t somn i d e an ,detail s obscuret bu , maie th n phase constructiof so n followss appeaa e b o .rt oldese Th t surviving portio castle -housee th th f s eni o , whic probabls hwa y erected during the first half of the fifteenth century. This was a rectangular building of four main storeys havin gparapetea d wall-head within which garret-chambera ther s ewa entrance-doorwae Th . y

Garret-TloorTlan Third-Floor Tim

1;' >"'Tg sX'/'/-'/.'/-/''/'/'%t_&&X • , ...I——7- 1 K^~z«^——T-T~v' ———

Second-TloorTlan FI G2 Upper-floor plan castlf so e 158 | PROCEEDINGS OF THE SOCIETY, 1969-70 was wallE situate e t approximatela 1th n di y ground-floor level, while withi thicknese nth f o s the S wall a staircase rose to the full height of the building. Each storey contained a single main apartment, and there was also an associated series of mural chambers and garderobes. After the completion of the lower stages of the tower-house, but probably as part of the same scheme of building operations, the S and E sides of the tower were enclosed by a curtain-wall surmounted by a crenellated parapet. Against the S wall of the courtyard thus formed there was buil single-storeyea t curtain-wal e anglE th S f e o th t da hallcylindricaa ld an , l flanking-tower two storey heightn si curtaie Th entrance-doorwayspierces .o ntw wa y db f whico , firste hth , at the centre of the E wall, was conveniently placed to give access to the tower-house, while the second, situated at the W end of the S curtain, entered behind the W wall of the hall. Subsequently, however cornecourtyare W th S f e ro enclose s sidth ,N fordo s ewa t it entrance-lobby n ma n do . A small lean-to building probably occupied the NW corner of the courtyard abutting the tower- house, while ancillary buildings of turf or dry-stone construction may have stood within an outer enclosure beyon curtain-walle dth . This outer enclosure also containe t leas dbuildina e ton g of stone and lime, namely a kitchen situated immediately to the N of the tower-house, adjacent to the spring mentioned above. This was presumably erected some time during the period between completioe th tower-house th f no courtyard commencemene an th d dan majoa f o t r schemf eo reconstructio sixteente th f no towardd h en century e sth . During the course of these late sixteenth-century alterations the original mural staircase tower-house turnpike-staircasinw th ne a blockes d an ewa , dup e constructe anglE S e th n di between first-floor and parapet-levels. The third floor and upperworks of the tower were re- modelled, while at the same time a new entrance-doorway was formed in the E wall at first-floor level, the original ground-floor opening being either blocked up or restricted in height. Within the courtyard the original parapet and parapet-walk were raised several feet in height and box- were constructe dentrance-doorwayso abovtw e eth e entirTh . e open aref ao the courtyard was filled with sand and boulders, thus forming platforms along the S and E sides of the tower-house between ground- and first-floor levels. Prior to this, differences in ground-level ensuresitd e onth eha d tha ground-flooe th t rtowee entrancelevates walth E f e wa ro l th dn ei several feet abov courtyare levee th eth f o l d outside. These alterations necessitated some re-organisation of the access-arrangements. Hence- forward the doorway in the E curtain constituted the principal entrance, giving access, by means of a somewhat awkwardly contrived open-well staircase, to the E parapet-walk and thence via the E platform to the tower-house. This new staircase appears to have encroached upon the havy ema t bee i cornehall E e d t thiN nth a an ,f s ro perio d that compensatory spac provides ewa d by breaking down the original W wall of the hall and taking in the area of the now superseded entrance-lobb cornecourtyarde W th S f e ro th n yi . The next phase in the development of the castle appears to have been carried out early seventeente inth h century, whe maie nth n work undertake constructioe th s n wa artiller n a f no y defence on the N side of the tower-house to protect the vulnerable landward-approach. This was intended to comprise a low forework, or battery, of multangular plan embracing the kitchen-area, but only the W portion of the battery was completed. To the same period there may also be ascribed a further remodelling of the upperworks of the tower-house in which the stair-head was altere mako dt e roosmala r mfo l turre angleE S te corbelle, th whil t a parapee t eth dou t garret-chambee rooe th f th o f d an r were raise heightn di .

1 The cardinal points of the compass are used for convenience of description; the true orientation is shown on the plans (figs 1 and 2). TURNER AND DUNBAR: BREACHACHA CASTLE; EXCAVATIONS AND FIELD SURVEY, 1965-8 | 159 The last major alteration structure th o st e were probably carrie t somdou e time during secone th dseventeente th hal f o f h century, whe earle nth y supersedes halwa l a.substantiay db l three-storeyed dwelling-house constructed upon the same site. This house incorporated a good deal of early masonry in the lower portions of its walls, and was lit on its S side by windows struck through the parapet of the old curtain-wall. Access to the interior was obtained by means existine ofth g doorway curtain-walle th n si uppee th , r floors being reached fro well-staircase mth e and associated platforms. Abou yeae th t r 175 castle supersede0s th hous w ewa ne ea erectey db d upo adjacenn na t sitet som buildinge bu , t leaseth a f o t s remaine occupation di considerabla r nfo e period there- after, and the roof of the tower-house was still intact in 1843 (NSA, vn (Argyll), 207; cf also pi 16a). One of the buildings thus occupied was the old kitchen on the N side of the tower-house, whico t h alterations were made durin course eighteente gth th f eo nineteentd han h centuries. Certain minor items of repair and restoration were carried out elsewhere in the castle by the proprietor, General Stewart, in about 1930-8 (Simpson 1941, 54).

DESCRIPTION The Tower-house towere Th (figs 1-2 oblons t )i f 8 2 plan measuregd o y b nan W fron o i t s m6 abouE t f 2 3 t transversely over thickneswalln i n i s 6 som t t ground-floosa f e7 r level t risei heigha ; o st f o t about 44 ft. A remarkable feature of the construction is that the walls in general measure about 1 ft more in thickness at third-floor level than at ground level, the upper portion of the inner wall-face showing a marked inward inclination whilst the outer wall-face tends to bulge out (cf figs 3-4). The masonry is of local metamorphic rock bound with shelly lime mortar and showin abundancn ga f pinningseo . largee Manth f yro facing-stones comprise bouldern o t se s edge featura , e found throughou greatee th t castlere parth originae f o tTh . l door windowd -an - openings likewisd an , quoinse eth formee ar , d fro same mth e material walls e bode th th f ys ,o s a but many of the later openings have dressed margins of greenish-yellow sandstone quarried fro Carsaie mth g bed Muln so l (Simpson 1941, 34). These latter openings have chamfered arrises, and the windows show glazing-grooves and bar-sockets. wallse Th , which were formerly harled (p restin e seeie b b 17a) o n t n par n gi ca , t upoe nth underlying rock. Excavation withi towethe n r showed tha surfacrocthe the t highl kis of e y irregular thad an ,t thert leasa deee s ei on tp V-shaped depression running transversels it o yt long axis (fig 6). This declivity was found to be filled with sand covered by loose stones, but it is probable that only the stones were laid by the castle-builders, and that the sand was already in the declivity when building started. Part of the S wall of the tower rests directly on the loose stonefille th . f Towardso angl E S faceworande E e th e sth , th mor f ko e particularlyS e th f o , walls shows trace f repaio s d reconstructionan r , whic havy hma e become necessar conn yi - sequenc insertioe th turnpike-stai e f eth o f no r within this cornebuildinge th f o r . e originaTh l entrance-doorwa walE e l th appears n yi o havt e been lintelled beneatha relieving-arch, the inner portion of the opening having been roofed with a segmental-headed vault. The external facework of the entrance has been restored, except at the foot, where there may be seen some of the blocking inserted when the doorway was closed up or restricted in sixteente th f o d h en century heighe th t ta . Withi doorwaye nth , whic rebateds hi , ther provisios ei n for two draw-bars, the inner one having no counter-socket. The outer face of the E wall is set on the edge of a steeply plunging rock-face (fig 6), and originally the threshold of the doorway lay about 6 ft above the level of the courtyard, access 160 | PROCEEDINGS OF THE SOCIETY, 1969-70

"Developed Section A. A.

Ft. 10 2o jo 40 50 6O

M 10 FIG 3 Section and elevation of castle from E to W

developed Section 'B.'B.

FI G4 Sectio elevatiod nan castlf no eN froo t mS FIG 6 . Section through tower-hous courtyard ean E d o frot mW A Guano F Sand B Debris, roofirig slate, etc. G Silt C Mortary soil f H Cobble pavement D Occupation deposit J Slab pavement E Loose stones

Key excavateo st d section) 9 sd (figan 7 s FIG? FIG 9 A A Phase I pebble floor B Curb stones B Occupation deposits C C Large stones (Phase II?) D Lin f waleo l face D Phase II cobble E Outer wall E Dwarf wall reducing siz fireplacf eo e F Void from boulder collapse f Perioo t walddou T l F Phase HI cobble (rounded) G Occupation deposit G Phase III cobble (angular) H Clay j H Base of Phase IV timber post 1 15th century mortared flolor (Phase IV strata within the Frenchman's House 2 [Later floors of rough mortared cobbling not shown) 3 Unshaded strata, apart fijom bedrock, consist of sterile brown sandy loan TURNE DUNBARD RAN I BREACHACHA CASTLE; EXCAVATION FIELD SAN D SURVEY, 1 1965-16 8

FIG 5 . Key-pla excavationsf no , 1965-8

presumably having been obtained by means of a timber forestair. The courtyard infilling com- prises clean sand topped with a layer of small boulders 1 ft to 1 ft 6 in deep, which forms a plat- form approximately 2 ft below the level of the curtain wall-walk. Subsequent deposition has covered both this platforwall-walke th d man t proveI . d impracticabl excavato et base th eo et of the sand infilling, but two pilot holes showed that it extended to at least 8 ft below the level of the wall-walk. There are no window-openings in the ground-floor chamber of the tower, which was presumabl ystore-cellara walN e l incorporateTh . aumbrieso tw s , while about midway along the N and S walls, and at a height of about 1 ft above threshold-level, there are opposed socket- holes evidently designed to house a stout timber beam. The purpose of this beam is uncertain, but it may have carried a framework supporting a brazier-hearth on the floor above. Alternatively the sockets may have formed a seating for a mechanism associated with a draw-well now con- cealed by the debris that fills the lowermost part of the chamber. At the E end of the S wall a lintelled doorway, now partly blocked (pi 15a), formerly gave access to the bottom of the mural staircase. This portion of the stair rose in a straight flight, but the upper treads are now blocked, als s i windoe s oth a w that formerl entrance-landinge th t yli . rudimentarA y scarcemen jfte ibas e exist th waldepta th f t eo t a l sa h belo threshole wth d t thibu s , doe in t exten 3 sno t f dvaryin2 alono t whole n i gth g 9 froet f lengtwale m1 th lf ho e anglE th (pS f i e o 14a)wallth N similae n A I .. th f o r d scarcemenen E e founs th wa t a d chambe belo t deptf a 3 t threshole ra wf hth o d ther hara s ei d mortar-fillet furthed an , r tracef so mortar-fillea same twal th S wer t ee a lleve eth f t founotheo a l d rd en points adherinE e th . o gt Resting on the loose stone infill at the bottom of the tower was found an occupation-layer, approximately 9 in deep and consisting of peaty material interleaved with clay. It contained TURNER AND DUNBARI BREACHACHA CASTLE; EXCAVATIONS AND FIELD SURVEY, 1965-8 | 163

FIG. 7 Plan and section of courtyard-hall. (For key see p. 162.)

considerable domestic bone refus wels a esherds a l scarcemene s th etc f I . t supporte dpoorla y made timber floor e occupation-layerth , beneathn i , 6 som t f e2 , could have been formey b d material falling through mortar-fillee th t bu , difficuls i t explaino t t therappareno d n an ,s ei t suppor flooe thoughth e b r y whertfo havo ma t t ei e butted agains wallS e .th t Abov e occupation-layeth e a dee s p wa rdeposi f mortaro t y soil containing many large boulder considerabld an s e lense f browo s n loam. This deposit certainly derived fro late mth e sixteenth-century reconstruction of the tower, and its compacted surface formed a mortar floor to the cellar after this date and sealed the partial blocking of the entrance to the original mural stair. Fragment brokea f so n Spanish costremicaceoud re f o l s ware were recovered from various levels within the mortary layer. Just inside the entrance, but off-centre and not entirely blocking it, was a shallow, roughly conical pit very crudely lined with boulders. Presumably this was normally kept covered with timber, but its purpose is obscure. e originath n I l arrangemen first-flooe th t r apartment r lowetaly o ,b l t rslitli halls - wa , windows in the N and E walls, but the E window was converted into an entrance-doorway during the late sixteenth-century reconstruction of the tower-house. The N window remains in s originait l condition e deeplth , y splayed being ceiled wit ha segmenta l arcd han lintelled toward outes sit r extremity windoE e Th .w appear havo st e been lintelled throughout, originae th t bu l arrangemen obscurew insertee no th s i ty db d doorway (pi 14a)doorwae Th . y itselformes wa f side-posty db s pegge substantiaa o dt l timber lintel, which remain positionn si , while within the doorway there is provision for a draw-bar. lowee flooe th f Th ro r hall rested upon timbers supporte roughln do y formed stone corbels, arrangemenn a t whic repeates uppehe wa th n dri towere levelth f so . Neithe thin o r s floorr no , on any other, was there any chimneyed fireplace in the original arrangement, and heating was probably provide brazierssocket-holey o db tw e .Th s situate wall anothed W e an , dth hign ri p hu same wallth havN y t ee a leveema , th carrien i l d timbers supportin gbrazier-hearta t secondha - floo walS re lintellea l levelth f o .d Toward doorwaen W e sth y gave accesoriginae th o st l mural staircase, which was lit at this level by a slit-window, now much restored. From the stair-landing 164 | PROCEEDINGS OF THE SOCIETY, 1969-70 acces obtaines si anglegarderoba W o S dt e , whicth n ei provides hi d wit hsmala l windod wan vent-shafa t dischargin wall staie W e Th .rbas e th itsel th f eo t ga f formerly spiralled upwardo st second-floo blockew no r . s leveli d up t bu , corneE doorwaA S e rth (pn yi i 14a) give turnpike-staircasse accesfooth e f th o t o st e that was constructed within this angle of the tower in the late sixteenth century, additional space for this purpose having been contrive bridginy db innee gth r re-entranS td maie anglan th nf E e o walls diametea staie s roughls i Th . rha d abouf an ro yt f constructed3 t , treade somth f eo s being of sandston otherd ean locaf so l rubble numbea ; treadf ro s have been restore timbern di . There attempo isn newela t jambentrance-doorwae e a t th Th . f so y have also been restorede th t bu , lintel remains and exhibits a chamfer and dog-tooth moulding (pi 15b). An inserted window in the E wall lights the lowest flight of the stair. The second-floor apartment, or upper hall, is lit by windows in the N and W walls. The segmental-arched window- appeae substantiallb o t r y originale daylightth t bu , - openings have been renewe late th e n sixteentdi h century window-embrasurN e th ; ceiles ei d beneath the arch-head with slab-lintels. A narrow doorway in the E wall admits to a small mural chamber lit by two crosslet-loops. In the original arrangement this chamber was probably more extensive thatodays i t appeari t ni r fo , havo st einsertee beeth y nb dcurtaileS e th n do wala cornee E y lb muraN th whicreturf a e N seao y rf th e lof lhma n nth staircasei n o d an , tower. Alternatively, thi sectio sN chamber e th f slit-windona o y wallb , N t whice li ,s th whn wa i may have forme separatda e apartment reached directly fro uppee mth r hal meany b ldoor a f so - way situated either in the N wall, or in the E side of the adjacent window-embrasure. A doorway in the S wall gives access to the former mural stair via a small vestibule. The loose rubble with which the stairway was filled at the end of the sixteenth century has now 1been removed to show a number of the descending treads. The upwards flight, which appears to have risen spiral-wis thire th do sealee t w floor no d s i stai e , off narrowes Th .ri steeped an r r thae nth anglE S e th inserteand n i ,e likdon e this newelo lattern s slit-windoA .ha , w which formerlt yli the stair between second- and third-floor levels can be seen blocked up in the external wall-face. entrance-doorwae Th y giving accesinsertee th o st d turnpike-stai t thia r s leve rebates i ln a r dfo inward-opening door sandstonmant e th ,bu f yo e rybats have been removed. Between secondd -an third-floor level turnpike-staie sth r incorporate spaciousa s aumbrwallE s it . n yi e third-flooTh r apartmen enteres i t d fro insertee mth d turnpike-stair throug hdoorwaa y similar in type to the one immediately below, and is lit by windows in the E and W walls. Origin- ally, however chambee wallth ,S d enteree s an , th wa rf o d d fro muraen e mth W le staith t a r was lit by windows in the E and N walls. From the original stair-landing access was obtained to a latrine situated towards the S end of the W wall. When the new turnpike-staircase was inserte murae dth l staiadjacend an r t latrine windo werw ne ea w blocked constructean p du n di the W wall. The vent-chute that served this latrine remains visible in the external wall-face. At the same time the window in the N wall was blocked up and a shallow-backed fireplace inserted immediately to the E of the former window-embrasure. This fireplace is roughly formed of loca square-cus l stonha d ean t jamb lintellea d san d windoheadE e .Th w appear havo st e remained undisturbe t thida s period t soma t ebu , later dat outee eth r perhapsilld an , s alsheade oth f o , daylight-openine th g were raise heightn di . originae Th l arrangement t parapet-levea s t eas no determineo yt e ar ll probability al n I . , however, the mural stair in the S wall continued upwards to give access to the parapet-walk,

1 When Boswell and Johnson visited the castle in 1773 this vestibule was being utilised by the Laird of Colprisoa s a l n (Pottle 1936, 266-7). TURNER AND DUNBARI BREACHACHA CASTLE; EXCAVATIONS AND FIELD SURVEY, 1965-8 | 165 perhaps terminating in a cap-house at the SW angle. The tower appears to have been covered by a pitched roof surmounting a low garret-chamber entered from the W parapet-walk. Lighting was presumably obtained by means of dormer-windows or roof-lights. The parapet-walk itself was probably much as it is today, being composed of massive slabs of slate laid alternately in ridge runnelsd san lattee th , r discharging throug parapee hth meany b t f weep-holeso s . Some of the weep-holes have subsequently been replaced by sandstone drain-spouts, and a number of the slate slabs may likewise have been renewed. The parapet itself seems to have had a thick- ness of 2 ft 6 in and a height of at least 3 ft. At each end of the W walk the parapet is seen to be inches offsew offseS fe e a ttth , having perhaps been associated e witon hcap-housea N e th d an , wit hgarderobea whicf o , vent-chutee hth blockedw ,no stils ,i l visible externall alsf y(c o pi 16a.) During the course of the late sixteenth-century reconstruction of the upperworks the original stair was superseded by the new turnpike-stair in the SE angle, which was carried up to terminate incap-housa e givin parapet-walkgS accese th o st uppemeane y b .Th t li r sstai e s flighth rwa f o t of a window in the E wall, of which considerable traces still remain. Durin earle gth y seventeenth-century alteration garret-chambee sth raises rwa d about f 2 t in height, low lateral walls being constructed within the thickness of the parapet-walk, and the gables being correspondingly raisedsame th t e A .tim lintelleea d fireplace, having roll-moulded jambs of sandstone, was formed in the E gable. The outline of the earlier gable-heads is clearly visibl gable-walmasonre W th e gablen eE th i f e yo th l- f (po i d 15c) lesd en san ,N clearl e th t ya wall t thiA . s period, too cap-house ,th turnpike-staie th f eo remodelles wa r thao ds t gavi t e parapet-walkE accese th o st , whenc furtheea r fligh roughlf o t y formed step upwardd sle a o st circular look-out stance corbelle t ove stair-heade dou th r stair-heae Th . d itselvaultes wa fd dan a new window was formed in the S wall to replace the earlier one in the E wall, which was now blocked up. Finally, the parapet was partially rebuilt, being raised to a height of about 5 ft and provided with a series of splayed gun-loops. In general this later parapet-wall is thinner than the original one, while along the E side of the tower it is roughly corbelled out from the main wall- faceoriginae Th . l angl garderobremodelleW s N ewa e th t ea t thi da s period t subsequentlbu , y fell into disuse and was blocked up. At the NE angle of the parapet-walk there is a rough stone seating, perhap sheltesa sentinela r rfo .

The curtain-wall and courtyard-buildings masonre Th originae th f yo l portion curtain-wale th f so generalls li y simila characten i r r tower-housee thao th t f to curtain outee S t part th d bu , f rsan o faceworsE incorporate th f ko e particularly massive boulders curtain-wale Th . noticeabls i harlet s i d wa l dan e thaharline th t g of the W curtain has been keyed to the underlying masonry by means of closely-studded nails. The wall varies in thickness from 3 ft 6 in to 5 ft and in the original arrangement incorporated a parapet-walk situate averagn a t da e heigh f about abovo f t 1 1 te ground-level. This wals kwa drained by a series of weep-holes, above which there rose a parapet some 5 ft in height, pierced with alternate crenelles and open slit-embrasures. Most of the weep-holes still remain, while blocked-up crenelles and embrasures can be seen in the N, W and E sections of the curtain. The parapet-walk seems to have been continuous, except at the SE angle, where it was interrupted roune byth d tower, whic enteres hwa t thida s level only walk E fro e mth . Acces parapete th o st - walk from within the courtyard was presumably obtained by means of steps, but the position of these is uncertain. The parapet-walk of the W curtain was equipped with a latrine placed tower-housangle junctioW e th clos S f th ecurtai o e e o et th th f nd o nean (pi 14b) vent-chute th , e discharging through the wall below. The two entrance-doorways giving access to the courtyard were provided with draw-bars. The doorway in the E curtain, which was virtually rebuilt by 166 | PROCEEDINGS OF THE SOCIETY, 1969-70 General Stewart, appears to have been equipped with a small squint-window, and may at one time have been surmounte commemorativa y db e panel (Sinclair 1899, 368-9). During the late sixteenth-century scheme of reconstruction the parapet was raised in l sideal n s o excep, wherraises t heighS f wa 3 e abouin.y 6 dt betwee eb th y ti d t 1f b t an 4 t t f n2 greatee worw Th facene s k i re pardth f witto h smaller stones tha usee n ar facewor n di k elsewhere in the castle. Most of the original parapet-openings were closed up, and new openings were provided in the form of obliquely aligned loops suitable for firearm defence. At the same time box-machicolations were constructed above the two entance-doorways, the one over the E doorway being provided wit hpeep-hola e overlookin sectio S adjacene e th g th f no t curtain-wall. The N parapet and the N section of the E parapet, which together protected the new first-floor entrance to the tower-house, seem to have been manned from timber walks set about 5 ft above the adjacent courtyard-platform, the platform itself being carried up to a height 2 ft below the original parapet-walk projectinA . g stone shelf, whic havy hma e helpe supporo dt t this timber seee walkb n n courtyard-platforme anglE withica th , N f e o nth , whil esocket-hola e visibln ei adjacene th tower-houste walth f o l havy eema serve dsimilaa r purpose. The S parapet appears to have been manned from a stone walk set about 3 ft above the original one, and drained by a weep-hole and a sandstone drain-spout which still survive, although now blocked up. Any openings that may have existed in either the earlier or the later parapet in this quarter have been obliterated by the inserted windows of the seventeenth-century dwelling- house. The N section of the W parapet was manned from the adjacent courtyard-platform, the latrinNe Wth n ei corne r being raise heighn di provided an t d wit overhanginn ha g vent-chute (pi 14b) to replace the original one, which was now blocked up. This chute collapsed in 1967-8. The arrangements in the S sections of the E and W parapets are not altogether clear, but a blocked-up gun-loo f thipo s seeperioe b nn aboudca t midway curtainalonW e gth . Turning now to the buildings within the S quarter of the courtyard, the original hall appears to have measured internally about 22 ft from E to W by 12 ft 6 in transversely. This building was single-storeyed and its low-pitched roof rose within the surrounding parapet-walk. It was entered frowindowy b m doorwaa t li s n i s wa wall yN d situatee an ,th f o dd towarden E e sth wallsS perhapd d an ,an E s , alsN oe th throug wall W f thes O e . hth e openings doublee th , - splayed window in the S wall and the N doorway still survive, the latter having been somewhat altered and restored. The E window, of which the daylight-opening was renewed with sandstone dressing e sixteentth n i s blockedwindow hN no centurye e th war ,d whil an ,y opening an e s havthay waltema W existe le wereth n di lost whe waldemolishede s nth wa l seconA . d doorway, cornerE situateS e th , gavn di e accesground-flooe th o st r chambe roune th f do r tower. Toward walN centre e ls th recesther th a f s eo ewa s widtsomn i t f whic f he5 o h considerable traces still remain. Thi se appearanc recesth s sha fireplacea f eo t excavatiobu , n showed that occupation-layers, consisting of carbonised material alternating with bands of sandy clay, rest- ing on the lowest of three floors to the hall ran right into it (figs 5 and 7). Thus, the purpose of the recess remains uncertain. Above these occupation layers a very rough floor of crude mortared cobbling had been laid down, while at the same time the recess was restricted in width to approxi- s outeit d r an edg n i e matel 6 delineate t f y3 f curb-stones o flimsa y w db yro e thirTh .d floor, als f poorlo y mortared cobble s probablwa , y laid down whe e latth ne seventeenth-century dwelling-hous s erectedewa t somA . e stage, possibl recenn yi t times e depositth , s withie nth

originae 1Th l parape parapet-wald an tS e th f o k curtain appear to have been between 1 ft and 2 ft lower than remainine thosth f eo g sectione th f o s curtain-wall. TURNER AND DUNBAR: BREACHACHA CASTLE; EXCAVATIONS AND FIELD SURVEY, 1965-8 | 167 restricted recess and behind the curb-stone, were disturbed by the excavation of a small pit (fig 7).i In addition to the hall the courtyard may also have contained a small building constructed against the inner face of the N portion of the W curtain. So much at least is suggested by the existence of a blocked-up window situated low down in this section of the curtain-wall. The SW cornecourtyarde th f ro , however, appear t firss a hav o t t e been unroofed comparativela t a t bu , y earl extendehals e y wal th N dat lwa f e o l eth d westward meeo s t inne e curtaintth W e r facth f -eo wall, thus forming a small entrance-lobby associated with the S doorway to the courtyard. In the W wall of this lobby there is a mural recess, perhaps originally intended as a seat, but which at some later period has been pierced at its base to form a sink-outlet or peep-hole. A doorway lobbe walth N f ye o li ngavth e accescourtyare th o st d beyond blocket s thibu ,swa whep du n the ground-level was raised in the late sixteenth century. It may have been at this period that the demolishedhals e walth lwa W f o l , thus increasin lengtformewidts e e git th th f y hho b r entrance- lobby. Whe e seventeenth-centurnth y dwelling-house e builcamb o tt e abov e earlieth e r hall and entrance lobby, the S curtain-wall and the S section of the W curtain were again raised in height lattee ,th r being chimneyecarriea s a p d u crow-steppe d dan d gable (pi 16a) associatee .Th d parapet-walks, and the box- above the S courtyard-doorway were built up and new windows were formed at first- and second-floor levels. These new openings, some of which have been restored, were constructed with dressed margin f Carsaiso g sandstone (Simpson 1941, windowe th 34)l Al . s appea hav o rt chamfered eha d arrises walS e thosd l th were an , n ei barred. The N wall of the dwelling-house was not founded directly upon the N wall of the hall below, t restebu d partially upo late neth sixteenth-century courtyard-platform, thus allowin buildine gth g an internal width of 13 ft 6 in. The E gable-wall, of which few traces now remain, appears to have been carried across som withit curtain-wallf E e 3 e nth , givin buildine gth internan ga l length of about 28 ft. Access to the ground-floor of the dwelling-house (the area formerly occupied hallobby)e d th an lobtaines y b wa , d both fro doorwacurtain-wale S m th e th n yi frod e an lm th late sixteenth-century stair-lobby associated courtyard-doorway E wit e hth firse Th t. floos wa r reached by means of the staircase itself, which at this period, at any rate, seems to have risen agains innee curtaith E t e r facth giv o nf t eo e acces sparapet-wal e botth o hdoorwat a o t d kyan in the E gable-wall of the dwelling-house. An adjacent doorway in the N wall of the dwelling- same housth t eea leve l apparentl timbea o t d r yle floo r spannin e stair-lobbgth y belowd an , giving e accescourtyard-platformsth o t s e originaTh . l sixteenth-century arrangemen f thio t s staircas altogethet no s ei possibls ri cleart i t e,bu tha gavt ti e direct acces sparapet-wale onlth o yt k of the E curtain. The first floor of the dwelling-house appears to have comprised a single large apartment heate meany db fireplaca f so e situated toward wallN centre e .s th th Thi f eo s fireplac bees eha n very much restored, but appears originally to have been lintelled beneath a relieving-arch. To the W of the fireplace a window, also much restored, looks out over the adjacent courtyard- platform secone Th . d flooprobabls wa r y entere meany d b doorwa a f so associated yan d forestair

1 Dr Simpson (1941, 45) mentions that this recess had been converted into a kind of bin by building a low wall across it, and the wall may be seen in a photograp office th la f hn eo ni Lindsa Partnersd yan , Architects, Edinburgh. This feature had entirely disappeared by the summer of 1965, however, and e evidencth walun-mortares w lo wa l thas ei e th t d d restean de thir th eithe dn o mortarer d cobble- debrin flooo r o rs accumulated abov. eit 168 | PROCEEDINGS OF THE SOCIETY, 1969-70 at the E end of the N wall; there is a fireplace in the W gable-wall, and there may have been anothe missine th n ri gable-wallgE staircas.A corneE S ladder ee o rth apparentln ri upwardd yle s to the parapet-walk of the adjacent round tower. The flanking-to wer This tower (pi 13a) is contemporary with the curtain-wall and its masonry is of similar character, except inasmuch as the plan-form has imposed limitations upon the size of stone selected for the facework. Although generally circular on plan the tower is very irregular, par- ticularly at its upper levels (cf fig 1); the walls show a pronounced batter at base. The tower rises to a height of two storeys and is surmounted by a parapet 5 ft 3 in in height, of which the coping is about 29 ft above ground-level. The flat-topped roof is drained by roughly formed weep-holes simila thoso rt eadjacene seeth n ni t curtain-walls same th t ea leve d an l, ther seriea s e i thref so e canted machicolations. The parapet itself is pierced by splayed embrasures, one of which over- look parapet-wale sth curtainE e th f .ko originae th n I l arrangemen ground-flooe th t r chambe enteres rwa meany d b doorwa a f so y at the SE corner of the hall. This doorway was equipped with a draw-bar and there is an aumbry in the E jamb. The apartment is lit by two slit-windows which enfilade the E and S curtain-walls. At first-floor level the interior of the tower becomes rectangular on plan. This apartment was entered fro parapet-wale mre-entranth e th curtain E n i e d th f tan ,k o angl e betwee curtaie nth n

BREACHACHA 1965-8: Battery Complex

Early Well Covered By Battery Wall

FIG 8 . Pla batterf no kitched yan n TURNER AND DUNBAR: BREACHACHA CASTLE; EXCAVATIONS AND FIELD SURVEY, 1965-8 | 169 and the tower there is a well-preserved garderobe supported on stone corbels. The doorway int towere secureos th draw-barwa a y db apartmene Th . t withi barrel-vaulteds ni flooe th rt bu , wa f timbeo s r supported upon rough stone corbels .dwelling-rooa This wa s therd a m an s ei lintelle wallsS d l thesdal ;an wal windowd fireplaceW E an openingle e th th n n ei si wel e sar l formed with carefully selected slabs and boulders of local rock. The fireplace-chimney rises to parapet-level withi haunce vaulte nth th hatcf hA .o vaule hape N th forme f e tx o allowe th n di d accesparapet-wale th o st meany k b ladder a f so , whil esecona d mean entrf so thio yt s leves wa l provided by means of a narrow doorway on the NW side of the tower, from which a ladder could have been lowered to the parapet-walk of the S curtain. Latterly this doorway communicated wit secone hth dseventeenth-centure flooth f o r y dwelling-house.

Kitchen and battery Immediately to the N of the tower-house there is a polygonal embracing a much altered, rectangular, single-storeyed building which incorporates a substantial fireplace. This outwork (pi 16b) can be identified as an artillery-battery, and it has been established that it post-dates the earliest phase of the rectangular building (figs 1 and 8). In view of the large terminal fireplace (pi 15d) and the proximity of a water-supply the rectangular building probably originally functioned as a kitchen. This interpretation raises certain difficulties, however, since there is no convenient access either to the tower-house or to the courtyard-buildings. Possibly the situation of the kitchen was determined partly by the position water-supple oth f face partld th yt an y tha y b groune remainine th tth n do gcastle sideth f eso was subject to flooding. The kitchen, together with other ancillary buildings now vanished or burie sandy db , probably stood withi outen na r enclosur eartf eo timbed han r knowsucs i s ha n to have existed at the end of the sixteenth century (infra, p 173). Excavation withi aroun d hamperes nan wa kitchee water-table ) dth 9 th d y db n an (fig8 , e5 s but the relative chronology of the structure was established. Kitchen-building, Phase I. The masonry of the kitchen is generally similar in character to that of the tower-house and curtain-wall, the facework incorporating many large slabs set on edg roughld ean y brough courseo t generoue th pinnings f n y o i sb e 3 t sus buildinf e 4 Th .1 s gwa wide walapproximatels internallE e wa l th thickd n i t widfireplace yan f 6 t 6 y Th .f e b s y2 ewa 2 ft 9 in deep, the N wall of the kitchen being thickened to accommodate it. The length of the kitchet beeno ns determinenha d excep inferencey tb walE alignes e i l Th . dcorne E exactl e th rn yo tower-house owalth fW le align th bosa d f rocn so o esan k which projects belo wale wf th o l the tower. Adjacent to this boss of rock an occupation-layer was exposed beneath later cobbling. Hence it may be inferred that the kitchen extended as far as the tower and abutted upon it, giving a length of 27 ft 9 in on the W side and 25 ft 9 in on the E. There are no traces of the kitchen to be seen on the N wall of the tower-house, however, and it is probable that the tower was re-harled after the demolition or collapse of the kitchen (infra, p 170). The hearth was living rock, and in front of it an occupation-layer covered the sloping surface of the rock. At the other end of the kitchen the occupation-layer rested on a pebble floor laid over waterlogged silt or sand. The water-level here prevented satisfactory excavation. fireplace th f corneo e buildine eE th th e f ro th go T rested partly ove springra . Boulder beed sha n clae packeth y f alongsido p d to int e sprine o th efoundatioth a s ga dry-ston a walle d th r an , nfo e wall built aroun e sprindth foro gt mpolygonaa l 'well-chamber' thisf o smala S , e th l reco T . - tangular well (pi 17c), also spring-fed, and approached by small rough steps, was built against kitchene walth E f originae e o l th Th . l doorwa kitchee th o yt n presumabl closy . yla eby Kitchen-building, Phase . BeforII battere eth constructes ywa kitchee dth eithed nha r been 170 ) PROCEEDINGS OF THE SOCIETY, 1969-70 slighted or had been allowed to become derelict. That there was some considerable interval between this dereliction of the kitchen and the construction of the battery is indicated by the fact that the harling of the tower is sealed by the battery (pi 17a). This harling, although only partially preserved today, except where protected by the battery, appears to have been applied afte collapse kitchenth runreasonable t th no f s eo i t .I supposo et e tha towee tth r would have been re-harled after its reconstruction and that the damage to the kitchen took place before this. frese Th h naturharlinge th f eo , where preserve batterye th y db , suggests tha intervae th t l before the construction of the battery is unlikely to have been of more than a few decades. The second phase of construction of the kitchen-building is contemporary with the erection kitchee batterye walplintth o w reduceW th s ff lo o le nwa ha Th .whic o dt h forme base dth e of the rear wall of the battery. The N end of the wall was reduced to approximately the same level and a gable for a lean-to roof placed upon it. A low wall, with two doorways with a splayed window between s builwa ,t acros e kitchenth s , thus formin a gsingle-roome t connecteddhu , presumably, with the functioning of the battery. Excavation in front of the fireplace showed tha t thia t s time flooa ther s largf o re wa e cobbles (pi 17b), which seale Phase dth eI occupatio n layer. To the E of the fireplace both this floor and the earlier occupation-layer could not be identified because of disturbance over the soft clay adjacent to the spring. The outer face of the N wall had fallen away except at the NE corner, and this wall was refaced in similar style to the rear retaining-wall of the battery. This masonry is crude, irregular badld an y pointed re-facine fireplace-walTh e . th o gt offses i l t from what remain originae th f so l wall-face. At some stage the polygonal well-chamber became silted up. The silting was partially removed and an elliptical dry-stone walled bay constructed within it. The elliptical bay partially collapsed outwards into the soft silt of the polygonal bay before it also became silted up. The silting of the polygonal baycontaineda sherd of sixteenth-seventeenth-century Frenchpolychrome ware. The elliptical bay may belong to Phase II of the kitchen-building: alternatively it could be contemporary wit constructioe hth late th e f seventeenth-centurno y dwelling-hous castlee th n ei - courtyard. Kitchen-building, Phase HI. The hut was divided into two unequal chambers by a stone partition and the splayed window was blocked up. A neat pavement of carefully selected round cobbles (pi 17b) was provided for the larger chamber and this pavement extended outside the apronn space a th rese s f a eo Th t.t betweehu hute towee battere ngiveth s th th , d wa r nyan pavemena angulaf o t r cobbles which extended int smallee oth r chamber. This smaller chamber was probabl yprivya hearth-aree Th . fireplace th f ao e same wasth t ea , time, reducee th y db constructio dwaro tw f f n o t eacwalls a e h sideon , . Presumabl t whehu ne state th y th f ethis o swa it was visited by Dr Johnson and James Boswell in 1773 (Pottle 1936, 267-8). Johnson 'gave some charity t famila largee livehu o o th t ' e n ydwh ri Th roo . in m 4 whict f 7 hy b measure n i 6 t f d9 was then known as Teigh Franchich, i.e. the Frenchman's House, but the family, apparently, were Scots. The neat cobbled floor was set in loose mortary sand resting on the Phase II pavement. There seems little pavemenprovisioe reasow th ne r a nfo f no t unles risina s g water-leved ha l mad t necessaryei . rectangulae Th r well (pi 17cbecomd )ha e choke stepde th wit f s o hwhic p rubbleto he Th . entered crosses ha coverewa a t dy i d b d drain whic frod cornee hcobblle e m th th f ro e aproo nt what appeared to be a sump a few feet SW of the rectangular well. This feature could not be investigated fully becaus evidence b y f lacfurthea eo f f timema ke o o t rbu , ris water-leven ei l in wet seasons. This could have been occasioned by the run-off from the spring becoming choked. TURNER AND DUNBAR: BREACHACHA CASTLE; EXCAVATIONS AND FIELD SURVEY, 1965-8 | 171 At the presumed SW corner of the old kitchen-building, the angular cobbling overlay the Phase I occupation-layer, but was not preserved right into the corner of the kitchen. Kitchen-building, Phase . AfteIV r soiaccumulated ha l d ove cobble th r e pavemene th f o t living-room of the Frenchman's House, this room was utilised as a byre and three stalls made with loose brick floors. The battery The outer wall of the battery (pi 13b) is neatly made, using the same technique as the curtain tower-housee th d an t displayinbu , g noticeably less variatio stone-sizen i fiele facee th d Th o . t s are well mortared and join at oblique angles except for the abrupt right-angled return to the tower-house. Excavation (figs 5, 8 and 9) showed that the rear of the battery-wall, which varies in thick- ness clay-bouns wa , d (pi 17a) t appear.I s tha thice th t k clay-bonded walbuils e wa lth t firsd an t

FIG. 9 Section x-x through kitchen from S to N (For key see p 162.)

mortared facing and parapet applied to it. The upper levels of the body of the battery were of earth, capped with a layer of boulders. The boulder-layer was irregular and did not provide anything resembling a pavement. The layer was interrupted by a shallow (partly traceable summey dr vegetatioe a f 1968th s o r a f o )d whicnen e marhth t wideneka d toward towee sth o rt a broad depression (fig 8). A gunner standing in the ditch would have only his head and shoulders above the parapet of the battery. There is no evidence that the parapet was ever higher than it todays i , namel yparapewidte e abouth th ; f ho ft t2 seem. in varo 3 st t f y 4 fro o t m n i abou6 t f t2 battery-wale th f o d Then le E appear 'fado st e away'. Excavation here showed tha suba t - stantia e battery-wall parth f o t l lies buried e clay-bondeTh . d inner wall endss neatlywa t bu , apparently never here given a mortar-face. This clay-bonded wall is linked to the re-facing of the fireplace, the linking being in dry-stone masonry. The contrast betwee re-facine n th fireplacneatle e th th o d gyt ean mortare d facine th f go battery suggests that the battery-wall was originally intended to continue behind the fireplace, and tha space th t ere-facine betweehavth o t d s ean gbee wa t n i r somn filleFo e. dreasoin e nth work was curtailed and the parapet of the battery returned to exclude the unfinished end. Excavation showed thaclay-bondee th t d battery-wall partly covere dcirculaa r settinf go boulders place trenca n di h runnin trencdepte e th th f restrictiogo ha ho t ( N-Se Du . trencn no h imposee b o deptt insuranc r d dfo hha e reasonspressure th d watef )an e o r this feature coult dno 172 | PROCEEDINGS OF THE SOCIETY, 1969-70 completele b y excavated bouldee s th t wa wall a bu ,r d f settino an , p to g e appeareth e b o dt probably another water-hole or spring. The outflow from this spring would have passed to the W of the castle. The re-entrant space, above the blocked well, between the fireplace and the end of the battery-wall beed ha ,n deliberatel bottoe th yt mfilleA . therdin e were layer f loa so sand man d and above this a sandy deposit containing coprolites. This was covered by a substantial layer of boulders. Blown sand had accumulated above the boulders, and in this there was a layer of collapse fro chimnee mth numbewels a y a s a l turf-linesf o r . Rabbit-burrowin complicated gha d the stratification throughout. It is probable that the battery was intended to embrace a vaulted chamber based on the formee th kitchef e wallo r th W d beekitchenf d e so en nth ha n d N reducean . n BotE a e o dt hth approximately level plinth and could have acted as the haunches of a vault. The very rough kitchene th f o , W whic e natur wale th th ho lt f retaineo battery e earte th s th f ho , provides support- ing evidenc r thi efo unlikels i thesit i r ysfo that this rough wallin intendes gvisiblewa e b o dt . Such an arrangement as that here suggested is closely paralleled at Brodick Castle {infra p 177). The eastern apron Excavation outside the E entrance to the courtyard showed that there was here a well- made pavement of large cobbles crossed by an open conduit (pi 17d). This pavement overlay earlien a r slab-pavement which also forme conduite base th th df e o . Sealed betweeo tw e nth pavements were found two small sherds of late seventeenth-century Westerwald stoneware. The upper pavement can be placed as being contemporary with the dwelling-house within the court- yard and its associated vestibule and stairwell. It extended, although not at full width, as far S as the flanking-tower, where it was directly beneath the corbelled latrine. To the SE of the pave- ment, in line with the open conduit, a waterlogged area was partially excavated and produced certaia n amoun woof o tfragment d dan f shoeo s s {infra 180p p , 185). An attempt to examine the stratigraphy below the lower pavement failed because of the water-level. There was no cobble-pavement outside the door in the S curtain corresponding to the later . OutsidE e th on eo et thi s entranc rougea h pavemen largf o t e boulders sloped down froe mth threshol approximatelr dfo . ft 0 y1 Exterior westthe to The land immediately to the W of the castle rises slowly from a low 'cliff' in the blown sand at the back of the beach. This ground is marshy during wet seasons, but behind it firm machair rises more steepl bence castlw th whicn o ne yht o e e standshth . smalA l trenc immediatelt hcu latrine-chute th f o tower-house basye W th th f eo t e a e showed tha land-surface th t e herresule risend th e f ha depositios o ta , f blowno n sand, approximately centurieslaso e th tw t n i t f . 2 Below this, waterlogged deposits were cappe considerably db e quantitie heatherf so . t seemI s probable tha medievan i t whicla timese e h inles n th cam o a thert f s o tp eu e wa castle th e wall thin so s side least ,a higt ta h spring-tides. This would accord wit late hth e sixteenth- century description quoted below (p 173).

HISTORY During the thirteenth century Coll formed part of the MacDougall lordship of Lorn, but followin defeae gforfeiturth d an t earle Robery th b f Alexande yo en n i so I t s hi f Lor o rd nan TURNE DUNBARD RAN : BREACHACHA CASTLE; EXCAVATION FIELD DSAN SURVEY, 3 1965-17 | 8 fourteenth century the island, together with other extensive territories, was granted to Angus Og of Islay (Gregory 1881 ; Dunca24 , Browd nan n 1957, 205-6) terme indenturn a th f y so B . e made between John of Islay and John of Lorn in 1354 Coll reverted to the MacDougalls, but the MacDonalds evidently continued to exercise certain rights of superiority (MacDonald 1950, 160; MacPhail 1914,75-8). Some timedurin secone gth d quartero fifteente fth h century Alexander, Lord of the Isles, granted the island to John Garbh, son of Lachlan Lubanach of Duart, and founder of the family of MacLean of Coll, (Gregory 1881, 71; RMS, iii (1513-46), no. 712). Durin secone gth dsixteente halth f fo h centur yfeua d developed betwee MacLeane nth f so Coll and the MacLeans of Duart and in 1578 Lachlan MacLean of Duart is known to have captured and garrisoned the castle (RFC, iii (1578-85), 132-3).1 A similar incident occurred following the death of Hector MacLean, 5th of Coll, in 1593 (Sinclair 1899, 370-1), when it was reported thae MacLeanth t f Duaro s d seize ha te 'castel dth f Bretacho l ; quhilk house thay intend to dimoleis and cast down'. The castle was not recovered until about the end of the year 1596 (RFC, v (1592-9), 354; Gregory 1881, 269-70). An anonymous account of the Western Isles written between 1577 and 1595 contains an interesting description of Coll and its castle 'quhilk is ane great strenth be reason of the situation thairof veri seae neith ,o t rquhil k defendi hale sth f thres thairofhe ed wallan , s abou rese th tt of the castell, and (?ane) thairof biggit with lyme and stane, with sundrie gude devises for defend- ing of the tower. Ane uther wall about that, within the quhilk schippis and boittis are drawin and uttermose th thire d th salvit dan d t An wal. f tymbeo l earthd an r , withi quhile nth haile kth l cuntrigudee th f keipie so e ar tymn ti troublef eo weirisr so ' (Skene 1876-80, iii, 436-7). Durin thire gth d quarter seventeente th f o h centur MacLeane yth s wer disputn ei e with Campbelle th 167n i d 9 an sDonal d MacLea f Colno l garrisone castle dth e against Archibald, 9th Earl of Argyll, for a short time, afterwards surrendering upon terms (Hist. MSS. Comm., Reporh 6t t Appendix, 622, 632-3).2 Donald's son, Hector, 13t f Collh o housw , ne buil ee th t of Breachacha, and it was there that Boswell and Johnson were entertained prior to their inspec- castle tioth f Octoben eni o r 1773 MacLeane Th . s retained possessio f Colno l until 1856, when it was sold to John Stewart of Glenbuckie and Lorn (Simpson 1941, 28).

DISCUSSION

Chronology Although the castle has generally been ascribed to the fourteenth century (MacGibbon Rosd an s 1887-92, iii, 119; Beveridge 1903 ; Simpso55 , n 1941 suggestew , no 52) s i t ,i d thaa t secone datth n ei d quarte fifteente th f ro h centur equalls yi y acceptabl architecturan eo l grounds preferree b o t historicas n i do d an l grounds. The architectural evidence is admittedly sparse, for the intractable nature of the building materials employed consequene th d an , t absenc f originaeo l dressing mouldingsd san , makt ei difficult to base any conclusions upon stylistic analysis, while the general form of the structure parallelee b n ca d elsewher Scotlann ei d throughou Middle th t e Ages. Nevertheless towere th , - hous t Breachachea uncharacteristis ai unvaultethas n ci i t ti lackd dan s original mural fireplaces, features that might be considered primitive and thus indicative of early date. Unvaulted tower- houses seem to be relatively more common in the Western Isles than in most other parts of

e detaileth f 1C d accoun f thio t s episode given i n Simpson 1941. 29 , detailee th f C 2 d account give Simpson i n 1941. 31 , 17 | 4 PROCEEDINGS OF THE SOCIETY, 1969-70 Scotland, however, possibls i t i d ean that this particular chronologicafeaturo n s eha l significance but results from a difficult1 y in obtaining local supplies of rubble-slabs suitable for barrel-vaulting. Certainly the building-stone available on Coll does not lend itself to this type of construction, while timber floor-joists could have been imported fro mainlane mth d without undue difficulty or expense. The absence of original mural fireplaces is more remarkable, but may likewise have been due to the peculiarity of local conditions, and in particular to the difficulty of obtaining adequate supplies of firewood on an almost treeless island. The main fuel on Coll was no doubt peathav y thid sean ma , been burn braziea n i t r standin stona n go e heartmiddle e th th n hf i e o floor. The of Kiessimul and Sinclair on the adjacent island of Barra are similarly unpro- face videdth t d thaan , t numerous Irish tower-houses firss a , regulato builtn d rha , fireplacer so chimneys (Leask 1951, 93), may perhaps be explained in the same way. It has already been pointed out that the tower-house at Kiessiemul shares two of the main pecularities observable at Breachacha, namely the absence of vaults and fireplaces; and a detailed comparison of the two buildings reveals other similarities (fig 10). Both are constructed without the use of freestone dressings and incorporate narrow mural stairs and a number of mural chambers. Moreover towero tw e s th ,hav e almost identical overall dimensions (Breachacha 32 ft 6 in by 28 ft; Kiessiemul, 31 ft 9 in by 29 ft), and each was conceived as the nucleus of a small courtyard castle incorporating a compact group of domestic buildings enclosed by a strong curtain-wall havin associaten ga d angle-tower placed diagonally opposit towere th o et - house itself. Lastly, it may be noted that in each case the curtain-wall appears to have been constructe same pars th d a f e o tschem buildinf eo g operation tower-housee th s sa t withoubu , t any attempt having been made at structural integration by the use of bond-stones. These similarities strongly suggest that Breachach Kiessiemud aan contemporarye ar l d an , tha singla t e designe responsibls rwa initiae th r le fo lay-out f botso h castles. Unfortunatele yth absenc f readileo y dateable stylistic feature markes a s i s t Kiessiemuda t Breachachaa s i t i s a l , castle th bees d eha nan variously ascribe twelfthe th o dt , thirteent fifteentd han h centuries, while opposing views have been taken of the chronological relationship between the tower-house and the courtyard. (MacGibbon and Ross 1887-92, iii, 51-6; RCAHM (Hebrides) 1928, no. 439; Mackenzie 1927, 163; Cruden 1963 ; Simpso,45 n 1967, 162-3) place th attemp.o t e t Thi no s si t a fresh analysis of the castle, but it should be pointed out that survey operations carried out by e Royath l Commissio Ancienn no t Monument 1967-n si 8 have shown thaput-loe th t g holes for timber parapet-walk hourdsd san , whose existenc bees eha n take conclusivs na e evidencf eo e earlth y medieva e castll datth f eeo (Cruden 1963 ; Simpso45 , n 1967, 162-3)facn i f o te ar , secondary constructio tha d originae nan th t l parapet-walk thif s o wer s t opinio e typeeno th n .I n joint-authore oth f thif o s s repor texistin e nonth f eo g building t Kiessiemusa l nee ascribee db d to a date earlier than the fifteenth century on architectural grounds. Turning now to a consideration of the historical evidence it may be observed that Brea- chacha Castle does not come on record until 1542, when it is referred to as the 'castle of Coll' chartea n i r grante Jamey dJohb o t snV MacLean Colf o d l (RMS,3r , in (1513-46) . 2787)no , . paucit e vien th I f w o earl f yo y record material relatin Hebridese th go t , however, little importance attachee b io st thio dt s date omissioe referenc.Th y an castle f nth o o et (and likewis Kiessieo et - mul) in Fordun's list of island strongholds (Fordun, Book 2, chap. 10), compiled sometime during the second half of the fourteenth century, (fig 11) is more significant, however, for this list seem othen i s r respect reliablee b o t s , providin gfairla y complete recor f castledo s then existing in the Western Isles. It is also noteworthy that in the indenture of 1354 between John

1 E.g. Castle Mao Caistead an l l Uisdein, Skyed an , 1 Kiessimul and Sinclair, Barra. FI 0 GComparativ1 e sketch-plans (partially reconstructed Breachachf )o a Castl Kiessimud ean l Castle 176 | PROCEEDINGS OF THE SOCIETY, 1969-70

FI castle e GWestere th Th 1 1f so n Isle recordes sa Forduny db Brodick1 . Lochranza,2 Rothesay3 , Dunyvaig,4 , 5 Chonnuill, 6 Duart, 7 Arcs, 8 Cairn na Burgh More, 9 Tiree, 10 Borve, 11 Tioram of Islay and John of Lorn by which Coll was restored to the MacDougalls (MacPhail 1914, 75-8), great importance was attached to the possession of the castle of Cairnburghmore (men- tione Fordun)y db nearbe th n i ,y Treshnish Islands referenco n t bu castl, y mads ei an n eo o e t Coll itself. MacDougall hostages retainee werb o et Johy db Islaf no y unti recoveree h l d Cairn- burghmor handsn eJohd ow int s an ,n ohi undertoo kname neveth f egrano o rt y castle an tth o et TURNE DUNBARD RAN : BREACHACHA CASTLE; EXCAVATION FIELD DSAN SURVEY, 7 1965-17 | 8 of the clan Tynwyne', who were enemies of the MacDougalls. The inference is that in 1354 Cairnburghmor thio t Westerse y parregardeth s ke f e wa o t th n s Islesd a tha d an ,t thers ewa castlo n Colln eo . The local political situation changed markedly, however, when the MacLeans established themselve Coln so l sometime durin secone gth d quarte fifteente th f ro h century possessioe th d ,an n of a suitable castleon the island is likelyto have been regarded as a prime necessity by the incoming family. It is reasonable to suggest, therefore, that Breachacha was erected at this time. It so happens tha establishmene th t MacNeile th islane f o tth f Barrn do so generalls ai y regardes da deriving fro mchartea r grante Alexandery db ,Isles e Lorth Gilleonao f t ,do n MacNei 142n i l 7 (RMS, n (1424-1513), no. 2287), so that the erection of Kiessiemul Castle may likewise be ascribed to the same period with some plausibility, a conclusion which fully accords with the architectural evidence already considered indicatin castleo g tw e contemporary thae ar s th t . Moreover, the MacLeans and the MacNeils, although sometimes at enmity with each other, were closely connected by marriage. Political, as well as geographical, considerations, therefore, favou probabilite th r y that both families engage same dth e designe their rfo r respective castles. majoe Th r reconstructio tower-house th f nassociateo e th d ean d raisin curtain-wale th f go l sixteente th f dateo e b d h aboun o dt en century importeca e e th th t y b d Spanish vessel (p 184) and by the dogtooth moulding on the lintel of the inserted stair. This moulding may be compared with similar ornamen t Dundarava t e (1596), Haggs (1585) Maclellan'd an , s Hous Tombd ean , Kirkcudbright (1582 and 1597). Moreover, this reconstruction was evidently carried out before the castle was visited by the writer of the 1577-95 account (supra p 173), for the phrase 'sundrie gude devises' is hardly one that could have been applied before the construction of the machicola- tions and the raising of the parapet. Sinc Duare eth t MacLeans occupied Breachacha fro deate mth Hectof ho f Colro 159n i l 3 until c 1596, it can be implied that the reconstruction was virtually complete by 1593. If the Spanish costrel arrive Coldispersaresule n da i th s Spanise la f to th f lo h Armad founderind aan g of mans shipit f 1588n ysi o , the datine reconstructioe nth th f go tower-housee th f no t leasta , , furthee b n rca confine perioe th o t d 1588-93 active Th . e feuding between ColDuard an l t a t this time would make major reconstruction difficult, but Breachacha was occupied and garrisoned by Lachlan MacLean of Duart in 1578 and it is not improbable that damage caused during this unsympathetic occupation necessitated some reconstruction. r SimpsoD n (1941 equate) 49 , battere dth y wit secone hth d wall mentione 1577-9e th n di 5 account, i.e. the refuge for ships, but this is unconvincing because he did not, at the time of his paper, realis trus eit e character facn sixteenth-centure I . th t y account implies that onl innee yth r curtain was of stone and lime. It seems more likely that the battery was not erected until after accoune th writtens wa t . This fitweln i s l wit hgeneraa l patter f developmenno whicn i t e hth castl s strengtheneewa d e reconstructiofirstlth y yb e tower-housth f no inned ean r curtain n(i about 1588-93) secondld additioan ,e th artillery-defencn y a y b f no vulnerable th n eo e landward side at about the beginning of the seventeenth century. Differences in the masonry of the recon- structed batter e othere tower-housth th f n o hande ,yo makd on curtaian d e , t ei th ean n no unlikely that both operations were carried out at the same time. The battery may be compared with the artillery defences at Brodick Castle, Arran, and Whittingehame Castle, East Lothian. finae Th l reconstructio upperworke th f n o towee wely th f ma lrso have been carriet dou soon afte MacLeane th r f Colo s l recovere castle dth 159n ei orden 6i mako t r e good damage MacLeane donth y eb knowe Duartf ar so o havo nt ,wh e threatene demoliso dt place hth e during their occupation of it in 1593-6 (supra p 173). Indeed, it is possible that the 1588-93 reconstruction had not quite been completed, and that the final work on the tower-head represents a continuation | PROCEEDING 8 17 S OF THE SOCIETY, 1969-70 earliee oth f r building-operations. This wor probabls kwa y carrie t aboua t same dou th t e time constructioe asth batterye th f no . The late dwelling-house in the courtyard cannot be closely dated. Its erection detracted from the defensive character of the castle, however, which probably implies that it took place afte dispute th r e betwee MacLeane nth Eare th f Argyl f o Colld o s an l l which ende 1679n di . dwelling-house Th contemporars ei y wit finae hth l modification e opeth no t sstair-wel e th n i l courtyard and is probably contemporary with the final phase of the pavement outside the E entrance. This pavement sealed two minute fragment Westerwalf so d stonewar latf eo e seventeenth- century date (p 185).

APPENDIX I

Ecological material

A. ANIMAL BONE. (By R Harcourt, BVMS, MRCVS) Introduction In view of the relatively short time-span of the material from this site, the apparent similarity of size and type throughout for each species, and the small size of the total collection, the material from the pre-1600 deposits has all been amalgamated into one group. The extremities of bones were measured across articular surfaces only. No attempt has been made to express any ages in years because it is the writer's belief that, in the absence of reliable evidence of how fast earlier animals matured, any such attempt gives a spurious impression of precision. There is no doubt however that certain teeth in a particular species erupt later than others and that certain long bone epiphyses fuse early, somintermediatn a t ea e stag otherd ean s late lifen ri . This latter sequencs ei very similal specieal r thusd fo rsan , with these criteria possibls i t i , arrivo eage-grout n a t ea p system which divide givesa n population into juvenile, young adul fulld an ty mature r olddo . Description Materialof All the usual domestic animals are present: cattle, sheep, goat, pig, horse, dog and domestic fowl. Remains of fox, whale, a small dolphin, fish and molluscs were also identified. The total number of identifiable minimuspecimene th d abous an m0 s wa number 30 t maie th nf s o specie s represented were: sheep, 20; cattle, 10; and pig, 3. The weight-ratio of cattle to sheep in any stock population, excluding extremes, is, in the writer's experience, about 10:1, both for advanced selected stock and for primitive unselected thus i t sI . evident tha tmea e mosth f t o teate thin no s beefs sitewa . Sheep dimensioe measurable Th th l al f no e bone showe sar see e Tabln b i nn thaeca 1t tI . these animals were smal slended lan r with bones closely resembling Soaye thosth f e.o This typ sheepf eo , referrino gt body-dimensions not wool, occurs with little alteration throughout the archaeological record both in Scotland (Ryder 1968, 127Englandd an ) . Cattle The bovid remains were much more fragmentary than those of sheep and only one bone was measurable. This was a metacarpal which had the dimensions: 181 tl, 47pw., 26 msd., 51 dw., msd% (see table 1) 14.4% tl. The estimated height is 108 cm, or 43 in (Fock 1966), a very small animal. Other domestic species Bones of horses, pig, dog and domestic fowl were identified, but were too few to merit special comment onle yTh . specime goaf no t identified with certaint distae th s l metacarpalhalya wa f fo . TURNER AND DUNBARI BREACHACHA CASTLE; EXCAVATIONS AND FIELD SURVEY, 1965-8 | 179

TABLE 1 MEASUREMENT SHEEF SO P LONG BONES tl. pw. msd. dw. msd.% Height t shouldersa . tl n i m c (dimension ) mm n si Humerus 114 - 13 25 11-4 - - 24-26(7- - ) - Radius 130 25 14 20 10-8 143 27 16 24 16-8 28(1) - - - Femur 156 - 13 32 8-35 - 11-8 1 2 - 161 1 2 2 2 2 5 11- 2 21 8 1 Metacarpa 7 10 l 3 2 1 2 8 5 8-0 01 8 1 Metatarsa 5 12 l 3 2 9 1 8 5 9-2 61 7 1 5 12 tl. = total length; pw. = proximal width; dw. = distal width; msd=midshaft diameter; msd%=m.s.d. as percentage of total length. Figures in brackets tl. indicate the number of specimens measured. To estimate the height, the tl. of metacarpal is multiplied by 4-86, and that of metatarsal by 4-68. The result is only approximate (Tsalkin, V 1 (1961. Cited in Ryder 1969, 49).

Wild mammals representes wa x radiua Fo y ulnad b san , both complete thin presenco e svien th g I . f wdo o f eo site and the apparent absence of fox from the fauna of many of the Scottish Islands, this identification was checked with particular care becaus lon e confusee eth b g n bone ca mala dx f so ewitfo h thosf eo some terrier-type dogs. The marine-mammal remains consisted of three fused cervical vertebrae of a small delphinid and a portion of bone, probably from the jaw of a whale, but this last identification is not absolutely certain. Age at death of cattle and sheep The specimens from fully mature animals outnumbered the combined total of those from young adult juvenilesd san . This reflect pattersa n which appears consistentl nearln yi periodl yal suggestd san s thaslaughteree th t d animals were killed only when somr fo , e reaso r anotherno , they wer longeo en r more valuable alive than dead possiblA . e explanatio bonee th younge e f somr so th n fo f eo r animals si that they were from cases of natural death. There is no reason to doubt that the meat in such cases was eaten more often than not. This possibility is too often forgotten. tima t eA when wool-productio maie th ns purposnwa f keepineo g shee whed pan n cattle, both oxe cowsd nan , wer motive eth e powe r ploughinfo r g (Trow-Smith 1957) self-evidens i t i , t tha kilo t t l young animals would be lunacy. It is probable in all periods, and known for the medieval (Trow-Smith 1957) that farm-stock matured slowly and that losses from disease were, from time to time, high. This would inevitably mean that stockmen woul t oftedno n have dare tak o drist e killinf eth ko g beasts other than those, already referred to, that had reached the end of their usefulness. To do so would have jeop- ardise vitae dth l breeding-stoc reducer ko numbee dth steerf ro s availabl essentiar efo l ploughinf o d gan ewes and wethers for wool production. In addition to the fact that there is no literary evidence for the practice (Trow-Smith 1957), these considerations show that the so-called 'autumn killing', unfortunately firmly entrenched in popular mythology, cannot ever have occurred. The slaughter of a few older cattle sheed an p carefully fattened durin lase gth t summe autumd ran theif no r usefusame th t el thinlifno s ei g at all.

Acknowledgments indeptem a I Miso dt s Judit KinhE confirminr gfo d an identitr e fox-bonee gth he th o f t y o d san FraserC F r , D both formerl Britise th f yo h Museum (Natural History) r examininfo , dolphine gth - and whale-specimens. 180 | PROCEEDINGS OF THE SOCIETY, 1969-70 B. FISH A number of fish-bones recovered from the pre-1600 deposits were examined by Dr K E Banister of the British Museum (Natural History). The only positively identifiable bones were cod but there appeare some b o det non-cod bones fro earle mth y deposi courtyare th n i t d house whiche coulb t dno certainly identified.

C. UNWORKED WOOD Fragment f unworkeso d wood from waterlogged deposits have been identifiee stafe th f th o f y db Royal Botanic Gardens, Kew. From deposits c 1700 (peaty deposit outside the cobble-apron E of the curtain, and the garderobe-deposit W of the tower-house) the following species were identified: Quercus (oak), sp. of robur type; Pinus (Scots pine), of sylvestris type; Alnus glutinosa (alder); Larix (larch sp. or Picea (spruce) sp.; Betula (birch) sp.; Prunus of avium (bird cherry) type; Fraxinus excelsior (ash); and nut shells of Carylus avellana (hazel). From the only deposit firmly datable to before the late 16th century which produced samples onle ,th y species identified were: Quercus of robur type (th esample)e bulth f ko , Betula sp.; Pinus sylvestrisf o type Picear Larixd o . . an ; sp sp Mr C E Hubbard, of the Royal Botanical Gardens, writes: 'The Trees represented in the samples as unworked wood are species which grow readily in Scotland. They could have grown on Coll before extensive clearance was carried out.' Most of the specimens were from small branches of a size unlikely to have been imported into Coll for building-timber. It is known from Boswell's account that there were virtually no trees on the island by 1773, but this discovery of unworked small branches in deposits that could not be much, if any, earlier than 1700 suggests that clearance could not have taken place much before that date.

APPENDII XI Finds A. BUILDING-STONE Sandstone Sandston widels ewa margine y th use r df windowfo so s durin late gth e sixteenth-century recon- structio laterd nan . BS1. Worn fragmen rybaf o t t with simple roll-moulding (not illustrated).G SubmitteF r M o dt Dimes, BSc, of the Institute of Geological Sciences, who writes: "The specimen does not match examples of Carsaig stone, which was formerly quarried at Carsaig, South-West Mull, in the Institute's collection. Nevertheless, I feel that it may be assigned to Car- saig stone, for the stone is very variable and it would be difficult to suggest any other provenance fo. Accordinit r traditioo gt sandstone-quarried ol e nth s nea Nun'e th r s pass, wes f Carsaigo t , supplied stone use ecclesiastican di l buildings stone f Scalpo Th . s ei a Beds, Middle Lias, Jurassic in age.' Roofing slate Fragments of roofing-slates were numerous in the superficial deposits. They contained the matrices of iron pyrites crystals similar to those found in slate from Easdale and Luing. The shape was usually sub-rectangular, the upper end being pointed. Many of the slates retain the corroded ends of their iron fixing-nails longese Th . t example found 1s llonn i wa i 4 wid y n £gi b approximatel y eb y ^§thickn i e .Th only slates which compared with Boswell's description (Simpson 1941, 42n) were the large slate slabs used on the wall-walk at the top of the tower-house. An example of the slate was examined by Mr Dimes, who writes: "The obvious localitie slate Ballachulise th ear r sfo Easdaled han e . th Thes n placeo o etw e sar same slat blace eth kbeld carbonaceouan t s slate with pyrite f Ballachulisso h extended through to the Oban district. The slate is very fine-grained, blue-black or black, and the beds vary slightly TURNER AND DUNBAR: BREACHACHA CASTLE; EXCAVATION AND FIELD SURVEY 1965-8 | 181 in colour and grain. The chief differences are noticed in variations in the quantity of pyrites which occurs most cubesoftein o i t states Ti s n ?na i t I . d that roofing slates fro islande mth s of Easdale have been worked for 300 years. I do not think it is possible absolutely to differentiate between Ballachulish and Easdale Slate, but, from close comparison with specimens in the Institute's col- lection thinI , Breachache kth a specime comparabls ni e with Ballachulish slate.'

B. STONE OBJECTS (fig 12) Workedflint A number of fragments of flint were found, some of which had certainly been deliberately struck. These were found almost exclusively in the occupation layers sealed at the base of the tower-house. S.I Thumb scraper. From reconstructio tower-housene bas e layeth th f eo t ra . S.2-4 Three piece f workeso d flint each wit hheavila y worn edge. Possibly crude strike-a-lights. From occupation layer at the bottom of the tower-house.

r-—-v

^^-^s^-J

FIG 12 1965-8 excavations: small finds

Worked Slate 5.5 Small fragment of slate with incised 'doodle'. From occupation layer at the bottom of the tower-house. 5.6 Incised disc, approx. 1 diameter^n i . Wedged tower-hous e int walth E of S o l levee th f lo t e a reconstructioe th n layer. 5.7 Spindle-whorl. From earth make-u batterye th f po .

. METAC L Iron tools, etc. Fe.l Partangea f o t d tool, possibl ychisea gouger o l . From below Phas I cobblineI g inside eth Frenchman's House (not illustrated). Fe.2 Stub end of a door-hinge support. From topsoil outside entrance, but possibly originally from the entrance (not illustrated). 182 | PROCEEDINGS OF THE SOCIETY, 1969-70 Copper Alloy objects (fig 12) Cu.l Fragments of two decorated bronze book-clasps. Thought to date from c 1500 by J P Harthan, Keeper of the Library, Victoria and Albert Museum. From occupation-layer at bottom of tower-house, closely associated with pottery I.I below, and the needle, Cu.2. Cu.2 Bronze needle. From occupation-layer at bottom of tower-house; closely associated with pottery, I.I below, and the book-clasps, Cu.l above. (Not illustrated). Cu.3 Part of a bell-shaped bronze object. Mr S Maxwell of the National Museum of Antiquities of Scotland writes: "The bronze fragment is very difficult to diagnose. The unbroken edge is so regular that I find it difficult to believe that it has been joined to anything else, i.e. that it could have been a bell, or the base of a candlestick, as the shape suggests. A cup doesn't seem possible, for the same reason, unles base s th muc s ei h wider tha necke nth wit s a , inkwell.n ha ' From inside the eastern chamber of the Frenchman's House, stratified with fragments of leather similar thoso t e describe below, presumably deposite earle th ydn i eighteent h century.

D. NUMISMATICA . N.I Coin submitted to Dr J P C Kent, FSA, of the British Museum, who writes: "The coin is a "double tournois" of Henry IV (1590-1610) or Louis XIII (1610-1643) of France. The diameter of 21 mm tends to favour Henry IV but this size still occasionally persists under Louis XIII.' From superficial deposits insid Frenchman'e eth s House quoto t , r KenteD As . , 'suc objecn ha t could onl residuae yb lats a l180s ea 0 eve remotesn i t Argyll' coi e probabln ,th n ca consideree yb havins da g entered the deposit during the collapse of part of the fireplace. It may be taken, therefore, as some slight confirmation of the seventeenth-century date argued for the construction of the battery and the re- modellin Kitchen-Buildinge th f go interestins i t I . noto gt e thaonle th ty coia n s sit e founth ewa n do founs thad wa Frenc dt an i t associatee hon d wit buildine hth g knowlate th e n eighteentni h centurs ya the Frenchman's House.

. POTTERE Y (fig 13) Wheel-made pottery was not found on the site in any context that could be place earlier than the end of the sixteenth century. Prior to that date the only pottery used seems to have been coarse, hand- made vessel same th f es o traditio Hebrideas na n sixteentCraggae th f o d nh wareen centure e th .Th w ysa importation of wheel-made wares, but Craggans continued in use and were found on the site in all deposits down to the nineteenth century. Group I: From occupation-layer at the base of the tower-house sealed during the late sixteenth- century reconstruction 1.1 Small bag-shaped, hand-made vessel. Thin but coarse, black-brown ware with some attempted external burnishing. Decorate nece randomlth y b kn do y disposed circles, apparently producey db stabbing with the end of a straw or truncated quill. Found in close association with the bronze clasps, Cu.l, above. A vessel decorated by the same technique has been reported from Tiree (Mann 1908, 327-8). Thialss oswa bag-shaped, wit uprighn ha t neck decoratioe Th .thref n i o s p e nto neawa n i t e rowon s- the rim, the others on the shoulder of the pot. There were no datable associations and it was described 'decoratea s a d Craggan' tere mTh . Cragga normalls ni y applie primitive th o dt e hand-made vessels made in the Hebrides down to the nineteenth century. A collection of Craggans from Bousd, Isle of Coll, has been illustrated (Mann 1908, 327-8) and ascribed to the nineteenth century. These vessels were flat-bottomed and undecorated. 1.2 Sherds of tiny, bag-shaped Craggan. Coarse brown ware with grey core. Some crushed stone temper. Decorate necn do k with small horse-shoe shaped indentations, approximatel rowsyn i , apparently produced by stabbing with the end of a straw or truncated quill. 1.3 Neck of Craggan. Crude, brown-black ware with coarse micaceous sand temper. TURNE DUNBARD RAN ; BREACHACHA CASTLE, EXCAVATION FIELD DSAN SURVEY 1965- | 818 3 1.4 Rim of wide-mouthed Craggan. Crude brown ware with a little coarse micaceous sand temper. Carbon deposi exteriorn o t . 1.5 Rim-sher smalf do l wide-mouthed Craggan. Thin, coarse grey-brown ware with some micaceous sand temper. Decorate conicao tw y db l pits placed near rim, apparently made after firing. smalo 1.6Tw l dishe hand-madef so ward re , e wit tempero hn .

1.5

1.3

1.4

m.2

/ I m.i 1.3 \ EZ.2 H.3 E.4 2.1 J

SI .3

^^fei_^^SI4 m.2 FIG 13 1965-8 excavations: pottery 18 | 4PROCEEDING S OF THE SOCIETY, 1969-70 Group FromII: mortary reconstruction layer footat of tower-house - late sixteenth century II. 1 Costrel of Merida ware. Sherds distributed vertically throughout layer in the SE corner of the tower-house. Hurst G FSA, Not J y MA :e,b Complete pear-shaped costrel with small vertical neck and thickened rim, rounded outside. Two handles with triangular section coming straigh frop bodu e t themd th yan n turning down sharplo yt joi shouldere nth . Flat slightly splayed base. Har fabrid dre c with many mica inclusions. This red micaceous fabric is typical of a group of wares made in the Extramadura region of south- west Spai near o Meridao n rt i kile tert knowe n.Th th mno site t e Meridnsbu ar a war proposes ei r dfo this group. This fabrilona s g cha lif e through from Roman time sherde mand th san f yo s still have Roman characteristics suc criss-cross ha s burnishing. Thi confuses sha issue dth Spain ei n where there are few stratified examples and this ware is still labelled as Roman in most Spanish museums.1 exportes wa t I Americao dt , firswhers wa t t recordeei d from Jamestow contextn ni firse th t f halso f of the seventeenth century though at that time it was thought to come from Mexico (Cotter 1958). Durin lasyeare n gth te t s Merida war bees eha n foun sixteen do Countriesw n siteLo e Britain si th .d nan Most of these finds are single examples or sherds but there was a large group of many hundred sherds from Plymouth in contexts dating between 1550 and 1650 (Oswald 1969, 122-4). These include large globular costrels, bowls, jugs or bottles. Thirteen of the British sites are in southern England or South Wales. There is only a single example on the east coast at Durham, while on the west coast there is only singla e fin t Chesteda r between Breachach findo Soutn tw si d ahan Wale t Kidwellsa Bentond yan . Beside Breachache sth a example ther foue ear r other datable costrel thif so srecognisedr typfa o es . earliese Th froe tar m Guy's Hospital, London Hythed ,an , Kent contextn ,i late th e f fifteentso earlr ho y sixteenth century. The next example was found at Tintern Abbey, Monmouthshire, but the form is more ovoid tha Breachache nth a example. belonThi y dissolutioe th sma o gt n period middl e sixie nth th - f eo teenth century fourte Th . h exampl Parson E foun s Refectore J ewa th y db n si t Durhaya contexa mn i t seventeente th f o d en datablh e centuryth o e t same th . f eThio basi s si c type clearls agaii t nbu y more developed wit bode becomw hth yno e very globular handlee shouldee th ,e th th o t sd raisen an ro p du rim sharply moulded and undercut. closese Th t parallel Breachache th o t s a costre groua e f threar lpo e from Middelburg, Nether- lands,2 unfortunately not from a dated context. The profile of these is almost identical only being a little large sizen ri . It will be seen therefore that this form has a long life at least from 1500 to 1700. The Breachacha exampl Tinterclosee s ei th o rt n exampl weln e place e thaca lb Durha e d nth d an betweee mon n themn i date wit ovoie hth d shape beginnin givo t pear-shapeep a expanto o g t e th t da d form which later became globular secone datth A . n e i sixteent de halth f fo h century suggestes a , archaeologicae th y db l evidence, therefors i e quit orden ei cannot rbu confirmede tb hopes i t I . d that more dated example thif so s typy ema com ligho et t whic amplify hma y this suggested developmen r refuto t . Thereit undatee ear d examples from Londo Guildhale th n i Ashmoleale Museuth n i d man n museu mt Oxforda . 11.2 Sherd of shoulder and body of wheel-made vessel. Hard grey-brown ware with even olive- brown glaze externally. 11.3 Semi-circular sher f largedo , wheel-made vessel. Hard grey-brown ware with even olive- brown glaze. Possibly deliberately chipped int odisa c shap lated ean r broken. (Not illustrated). 11.4 Sherds of undecorated Craggan ware similar to those found in the occupation-layer sealed below. (Not illustrated).

Group III: From silt ofapsidal spring-bay 111.1 Sherd of polychrome ware from Saintogne. White ware with pattern in mauve and green under lead glaze. Sixteenth-seventeenth century. widf o 111.em mouthe2Ri d grey-brown Craggan. 111.3 Sherd of pinky-brown ware with coarse sand tempering. Decorated with horizontal parallel rows of stabbed cylindrical pits.

11 am greatly indepted to Mr Juan Zozaya, of the Provincial Museum in Soria, for his help in identify- ing this fabric. Zeeuwe th n 2I s Museum. TURNER AND DUNBAR: BREACHACHA CASTLE; EXCAVATIONS AND FIELD SURVEY, 5 1965-18 | 8 Group FromIV: below cobblethe outside east entrance IV .Bas1 e angle sher f seventeenth-centuro d y Frechen Bellarmine with mottled 'tiger' glaze. (Not illustrated). IV.2 Sher f Westerwaldo d stoneware. Deep blue decorated with embossed roundel greyn si . Late seventeenth century. IV.3 Sherd of Westerwald stoneware. Grey with vertical line of raised 'flowers'. Mauve bands of panel. Late seventeenth century. IV.4 Rim sherd of small Craggan. Group FromV: garderobe deposits west tower-houseof browf o m V.nRi Isurface d Craggan ware. Decorate nece diagonath y kb t d a l impressions. Pro- bably early eighteenth century. Group Post-1750VI: Craggans VI. 1 Rim and neck of Craggan. Coarse brown ware with coarse micaceous sand temper. Carbon deposi exteriorn o t . From nineteenth-century roof collapse over mortar floo tower-housen i r . Associated with nineteenth-century stoneware. VI.2 Cragga f blacno k ware with some micaceous sand temper. From sand dune behin firee dth - Frenchman'e placth f eo s House burielase th t largy db e collaps masonryf eo . Associated with tinglazed earthenware (delft brown-glazed an ) d earthenware. Probably eighteenth century. VI .Uprigh3 t nec f beakeko f grey-browo r n Craggan ware with some coarse sand temper. From upper fill of the spring-bay. VI.4 HalCraggaa f o f n from guano abov blockine eth earle th yf g o mura l stai t second-flooa r r level. Nineteenth century.

. GLASF S plaif o nG.m glasIRi s bowl, approx diamn i .4 , ^ thickn i . From below Phas cobblI e II th n ei Frenchman's House. (Not illustrated). Bottl2 G. e stamp: 'Coll 1816' inside circle. From beneath tumble just above cobble outside French- man's House. (Not illustrated).

G. LEATHER Two groups of leather fragments were found in waterlogged deposits. The first was in the garde- robe-deposits W of the tower-house, the other in wet sand immediately above the peaty deposit SE of the cobble apron outside the E entrance. Both groups were treated by the laboratory of the London Museum and were examined by Miss J M Swan, of Northampton Museum, who wrote: "The two groups from different deposits both contain what would be typical "sweepings up" from a shoemaker's or repairer's workshop (they would be one and the same person in the eighteenth century). The leather is cattle-hide, imperfectly vegetable-tanned - the tanning has not reached middle th e satisfactoril splittinleathee w th no d s i rtwoyn an gi t .unusual Thino s si . [The group from outside the E entrance] includes part of a heel, probably worn on a man's right foot (though the shoe would have been straight, i.e. made for either foot) as Mr J H Thornton of the Northampton College of Technology thinks that it seems to be worn on this side. It is the usual stacked leather type, fastened together with wooden pegs Thorntor M . n think mors i t si r les eo s complet cond ean - sequently, as it is quite low, would most likely come from a shoe and not a boot, dating post 1720. There are two fragments which belong with it and have become separated, the topmost of which is the seat, and the stitch holes are visible round the outside, which attached the heel to the shoe. There are only three other pieces in the [group] which may be from shoes: one piece which may be the toe end of a vamp, though it shows no signs of having been stitched; another which may tonguee th f o ,p dat to t fro be cu m th als o1720c othee th ; lacrpara e whicb f eo ty anklehma - shoe which could not have come from the same shoe as the tongue. [The group from the top of the garderobe deposits W of the tower-house contains] one piece which is certainly from a shoe and could be from the same shoe as the fragment of tongue. This is part of the strap of a buckle brokes usuae ha sho th d n lei an plac e wher prong e buckle eth th f so e weake leathere nth leathe.A r stra thif po s widt moss hi t likel havo yt e come fro mman'a s shoe datin betweeo gt n 167d 0an 186 | PROCEEDINGS OF THE SOCIETY, 1969-70 1735, though the recuo yd groupe r th later rese f leather-workerss i ][o tTh . ' debri could san d very well have been fro mshoemaker'sa . Both [groups] contain shoemaker's debri therd san s ei nothing to suggest that they are not from the same shoemaker's workshop,' similarite Th y betwee leathee nth r fragment groupo s tw fro e smth fro m oppositee sideth f o s castle suggests that both deposits were fro firse m th eighteente t th hal f o f h century, possibly froe mth third decade. (Not illustrated).

. WORKEH D BONE ) (fig12 Bn.l Possible tooth fro mbona e comb, longr-Tn Oi square?n i . From occupation-laye base th et ra tower-housee ofth . (Not illustrated). Bn.2 Needle, 2f in long, with an approximately square eye. A primitive type that can be closely parallele exampley db s from prehistoric contexts. From occupation-laye t basra tower-housef eo . Bn.3 Possible tally. Sliver of worked bone of rectangular section 1^ in by $ in by ^ in. From occupation-laye t botto ra tower-house e th mf o . (Not illustrated). Bn.4 Antler pin from garderobe deposit W of tower-house. Early eighteenth century. Bn.5 Bone handle wit hgroovea d pattern from superficial deposit Frenchman'n si s House. Nine- teenth century.

. WORKEJ D WOOD (fi) g14 W.I Mitred end of batten. From below the cobble outside E curtain, seventeenth century or earlier. (Not illustrated).

W.6

FIG 14 1965-8 excavations: pulley-block TURNER AND DUNBAR: BREACHACHA CASTLE; EXCAVATIONS AND FIELD SURVEY, 7 1965-18 | 8 W.2 Bottom of the leg of a chair with mortices for cross rails. From garderobe deposit W of tower-house. Probably deposited early eighteenth century. (Not illustrated). W.3 Fragments of pegged mortice and tenon joints. From garderobe deposit W of tower-house. (Not illustrated). Notche4 W. d stic tallyr ko . From garderobe deposi tower-housef o W t . (Not illustrated). W.5 Timber, approximately 4 in by 3^ in by 2 ft 1 in long, with complicated chamfered tenon at one end and a large mortice with two peg-holes at the other. Possibly from the roof of the tower-house. From top of waterlogged deposits W of the tower-house. (Not illustrated). Pulley-bloc6 W. k from rigging .rectangula e Froth filf e mo lth Frenchman're welth f o l s House. Brow NationaA (Drawe r th f M n o y nb l Museum.)

REFERENCES

Beveridge 1903 Beveridge, Erskine, Coll and Tiree, London, 1903. Cotter 1958 Cotter , ArchaeologicalL ,J Excavations t Jamestown,a Washington, 1958. Craden 1963 Cruden, Stewart, The Scottish Castle, Edinburgh, 1963. Duncan and Brown 1957 Duncan, A A M, and Brown, A L, 'Argyll and the Isles in the Earlier Middle Ages', PSAS, xc (1956-7), 192-220. Fock 1966 Fock , 'MetrischJ , e Untersuchunge , Metapodienan n Einiger Europaischer Rinderrassen'. Diss, Munich, 1966. Fordun Fordun, John, Chronica Gentis Scotorum, various editions. Gregory 1881 Gregory, D, The History of the Western Highlands and Isles of Scotland, London, 1881. Hist. MSS. Comm. Reports of the Royal Commission on Historical Manuscripts, London, 1870-. Leask 1951 Leask, H G, Irish Castles and Castellated Houses, Dundalk, 1951. MacDonald 1950 MacDonald, C M, The History of Argyll up to the beginning of the sixteenth century, Glasgow, 1950. MacGibbo Rosd nan s 1887-92 MacGibbon Castellatede Rossd Th an , T Domestic,d D , an Archi- tecture of Scotland, Edinburgh, 1887-92. Mackenzie 1927 Mackenzie MackayW , Medievale Th , Castle Scotland,n i London, 1927. MacPhail 1914 MacPhail , HighlandN R J , Papers, , Scottisi h History Society, Edinburgh, 1914. Mann 1908 Mann ', AM potter ,L y churn fro islane mth Colf do l with remark Hebridean so n pottery', PSAS, XLII (1907-8), 326-9. Statisticalw Ne e Th A AccountNS of Scotland, Edinburgh, 1845. Oswald 1969 Oswald , 'MarkeA , d Clay Pipes from Plymouth, Devon', Post-Medieval Archaeology, 3 (1969), 122-42. Pottle 193 6Bennettd Pottle an , BosweWsH A C F ,, Journal Hebrides e oToura th f o t with Samuel Johnson, LL.D., London, 1936. RCAHM (Hebrides) 1928 Royal Commissio Ancienn no Historicad an t l Monument f Scotlando s : Inventory Ancientthe Historicalof and Monuments Constructionsand Outerthe in Hebrides, Skye anSmalle dth Isles, Edinburgh, 1928. RMS Registrum Magni Sigilli Regum Scotorum, Edinburgh, 1882-1914. RFC Registere Privye Th th f o Council of Scotland, Edinburgh, 1887-. Ryder 1968 Ryder , 'ThL eM , Evolutio f Scottisno h Breed f Sheep'so , Scottish Studies, (1968)2 1 , 127-67. Ryder 1969 Ryder, M L, Animal Bones in Archaeology, Oxford, 1969. Simpson 1941 Simpson, W D, 'Breachacha Castle in the Isle of Coll', Glasgow Archaeol Trans, x (1941), 27-54. Simpson 1967 Simpson, W D, Portrait of Skye and the Outer Hebrides, London, 1967. Sinclair 1899 Sinclair MacleanA , Clane Th , Gillean, Charlottetown, 1899. Skene 1876-80 Skene, W F, Celtic Scotland, Edinburgh, 1876-80. Trow-Smith 1957 Trow-Smith, R, A History of British Livestock Husbandry to 1700, London, 1957.

The Society indebtedis Civilthe to Service Department granta for towards the cost of this paper.