II. NOTE CERTAIN SO N STRUCTURE ARCHAIF SO CE ISLAN TYPTH N EI D OF LEWIS-BEEHIVE HOUSES, DUNS, AND STONE CIRCLES. BY W. M. MACKENZIE, M.A., F.S.A.ScoT.

Beehive House*.—Eastwards from the northern shore of Loch Resort on the west coast of Lewis, and between the high lands of Uig and the Fores f Harriso t , run a lons g irregular valley streaked with lochand san watercourses, and occupied towards its eastern extremity by the upper end of the fresh-water Loch Langabhat. On the south side the hills rise in a steep escarpment, while to the north the valley retires in a gentle slope passing intuplandw olo s traversed by shallow glens. It was from this valley that Captain Thomas in 1858 1 made known to the antiquarian world the existence of certain structures of a primitive type, popularly known, from their resemblanc strad ol w e n fori eth mo t hives s "a , Beehiv e Houses, n Gaelii " c styled botliain (sing, bothr o botlian), a word cognate with the English "booth "and " bothie." He foun examples dhi e \vester th n i s n end. Those with whic aboum a hI t to deal lie farther east -where the valley opens out into a scene of pastoral .richness, quiet beautyd an , , over which thdeey esh r boldly move among what were once the habitations of men. clearls i t I largn yi e measur o thit se remotenesdu e s tha mano s t f yo these structures have been preserved t eveYe .n here evidenc- de f o e structio s plentifule nwesteri th n O n . spur f tho se Cuailean Hills, part e uplandoth f s above mentioned e hig th d withi hf an o , o na s mil r o e road to Uig, are the traces of several bothain. Tom-na-Bharabhas, for example, is a knoll deriving its name, according to tradition, from a 1 Proc. Soc. Ant. Scot., vol. iii. p. 127. 174 PROCEEDINGS OF THE SOCIETY, FEBRUARY 8, 1904.

successful pursuit of the Macaulays of Uig by the Morrisons of Barvas, who overtook the plunderers of their cattle resting at this spot, probably in the two bothain whose ruins crown its summit. Of the larger sufficient remains to show clearly the characteristic construction and the circular shape, wit a hdiamete 7 fee f to rbetwee e narronth w entrances which loo kd E.S.E]S an s Tcompanio. It . n measures abou a foot t less. Farther east other ruins occur, and one of these was in summer occupa- tion less than hal a centurf y ago, until supplante a large y b dr modern dwellin constructioe th o gt f whicno oldee hth r buildings paid tribute. A somewhat similar fate has overtaken the group on the eastern extremity of Beiii-a-Chuillein, which were utilised as shielings by the crofters of Beniera up to little more than thirty years ago. Their final ruin was brought about, I am informed, by their being used for harbour- agy poacherb e s aftee grazingth r d beeha s n incorporate e deerth -n i d forest. There are the foundations of several on the hillside above Glen Marstaig, and one, in a rather better state of preservation,' shows traces of adaptation to the more modern type in that its top has been adjusted to permit of rafters and a roof of heather or turf (fig. 1). The weakness of all these structures appears to lie in the upper portion where the curve becomes pronounced toward e apexth s . With, carattentiod ean n e outetoth r coverin f turfgo , which, thoug a hsuperimpose d weightt ye , seems to add to the stability of the building, they may last indefinitely; once tha removeds i t , disintegratio collapsd nan onls ei matteya f timeo r . This is clearly seen in the relative condition of the members of another group across the valley. One is in a nearly perfect state. To this the peculiarity of its site must have largely contributed. It stands in a shallow gully whose sides converge toward e west th whicd s an , h forms part of a shallow depression on the lower slope of Calltraseal Mhor, overlooking Loch a Sguir. Covered with compact turf from base to summit, it looks almost a part of its surroundings, save that its conical regularo shapto d thas i e an ,t herd ther e an destructive th e e work of rabbits has laid bare the grey stones of its framework (tig. 2). A door, low and narrow, but sufficient for the admission of a full-grown

6 17 PROCEEDING E SOCIETYTH F O S , FEBRUAR , 19048 Y . s kneeshi ma n o n, faces south-easts apparentlwa d an , y shelterey b d a detached wall of its own height. Apart from this the only light to the interior is afforded by an opening in the centre of the roof of abou inche9 t y 6b ,s which migh easile b t y stoppe stona y r mordb eo e probabl a yturf . Opposit e dooth e r three small chamber r ambrieo s s rise in the thickness of the lower wall, one above another, each separated from that below by a single flat stone. This particular arrangement was no tothe y founan rn di example additionao Tw . l ambries ope eithen no r side on the level of the earthen floor, and a third directly opposite to the left of the doorway looking out. Between the last and the doorway a light brow gavh nas e evidenc e cagailtth f er o hearte peat-fire o th f o h . The diameter of the floor is 8 feet 4 inches between opposite ambries, and 7 feet 7 inches from the entrance to the opposite wall. The greatest height to the inside face of the top stones is 8 feet 5f inches. On the more open ground near by are three other examples all more exposed in situation and so in various stages of imperfection. Of one about 80 yards to the south-west quite a quarter has collapsed. The greatest height of wall is on the north-east side, and measures 5 feet 4 inches perpendicularly froe groundmth t presentI . goosa d section, showin charactee gth e buildinth f principae o r Th g (fig. 3) l. stonee sar lonflatd eacd gan an ,h course projects over that below, unti roune th l d converges sufficiently at the top to be closed in the manner indicated above. The weight of the heavier stones alone binds the structure, which in this respect is Cyclopean in character. Gaps and interstices are filled in with smaller rubble. The material is supplied from the schistose gneiss of the district, which in many places weathers, or is easily quarried n suitabli , e presene th slabs n I t. case outeth e r coverinf o g turf has, of course, disappeared, save in patches round the base. It has x ambrie floosi e e floo th Th e higher n r diamete on so . level d up r an , r is 7 feet 6 inches N. and S., and 7 feet 1 inch E. and W. O knolna l abou yard0 5 t s a specimee lasN.W s th i t f o . n also bereft s coveringoit f t otherwisebu , , wit e exceptioth h e finath f l o ncourse , complet e basemenTh e (fig. 4) . t wall, from whose inner edge springs

178 PROCEEDINGS OF THE SOCIETY, FEBRUARY 8, 1904.

e dometh d roofd upoan , n whic t i restsh fee3 n thicknes i s ti , d an s about the same in height. It is in this more massive part of the structure thae ambrieth t r cupboardo s s already mentione e usuallar d y constructed. In this example four are so placed, but three others, 3 feet from the floor, now open to the outside. The floor space is rather larger than is common. It measures 7 feet 9 inches E. and W., e greates fee8 Th inche9 td . an S t sinternad Nan . l heigh f whao t t remains is 6 feet 8 inches, and the height when complete must have been little short of 9 feet. The doorway faces S.E. The structure as t standi s lays t werebari s e typicaa e th , fous r l skeleto e both.th f no One other to the south of these completes the group at Garry-na-Sguir. It is the most ruinous, only about 4 feet 6 inches of wall remaining. Its greatest diamete s aboui r feet8 d besidet an , s ambrie apparentld ha t si y two entrances, facing N.E, and S.W. respectively. Two entrances, indeed, are not uncommon, and they are in all cases opposite to each other. About hal mila f e farther sout e reac hw e ampl hth e margi Loclf no i Chohigerod. Wher e gentlth e y sloping meadow terminates abruptly a rocke foo f th o ty t a e wesridgth to t e stand a compacs t group well suiting in their appearance the comparison to a Kaffir kraal, and strangely impressive in their suggestion of former residence and present abandon- ment. There are seven in all, of which two are in a state of fairly good preservation e entranceTh . s generally look toward e south-eastth s . Their proportions and main features, though varying, are on the whole simila thoso t r e detailed above. Three still wea a partiar l coverinf o g turf ; one is marred only by a hole in the back wall (fig. 5). t wilI observee lb e exampled th tha l al t s described stand well apart irom each other, though associate sam builgroupsthe d in are e on t and , simple plan, varying slightly only in dimensions and in the number of ambries and doorways. The interconnected and more complex groups described by Captain Thomas lie farther west and nearer the seashore.1 It is not unlikely that this fact differentiates the more permanent from e temporarth y habitation originae th f so l builders. 1 Proc. Soc. Ant. Scot., vol. ; iiivol .7 .p. 12 .vii . 153p . . STRUCTURES OF ARCHAIC TYPE IN THE ISLAND OF LEWIS. 179

moro t I pasw e isolatesno d outlying examples t Garry-na-hina e On . e •close to the Uig road, formerly a shieling, was occupied at Captain Thomas's visit (1866) as a residence, but is now abandoned (fig. 6). It presents some special features. There is ,1 chimney, as Captain Thomas: s clearlyi opiniony m t notedi n t i , modera bu ,, n modificatione th n I . main, however a structur s i t i , e exactly resembling those alread- de y scribed, but more carefully preserved. Its longest floor diameter is 7 feet

Fig . Bot5 . t Locha h Choirigerod.

9 inches, and that at right angles 6 feet 4 inches. There are two door- ways facing roughly E. and W. (fig. 7), and three ambries on the level, Jill these being integra le originapartth f o s l structure e doorwayTh . s are 3 feet 3 inches high and 1 foot 9 inches wide. The ambries are .about 3 feet deep and 18 inches wide. Mr Carmichael, in a paper on •the Agrestic Custom e Outeth f so r Hebrides, formin appendin ga e th o xt Eepore Crofteth f o t r Commission, describe e "dothth s cloiche" stone •both beehiver yo , "still, sayse h " , "the shealing Lewie th f so s people." 1 Pro. Soc. Ant. Scot., vol. iii. p. 135.

STRUCTURES OF ARCHAIC TYPE IN THE ISLAND OF LEWIS. 181

Ho refer theio t s r internal arrangement follows a s e walle th th f s: so "n I hut, two, three four o , r feet fro floore mrecesses—Gaelice th ar , , butliailt, Scots, bole-—for the various utensils in use by the people, while in the boso f thmo e thick wall low down nea e dormitoriee grounth th r e ar d s wherein the people sleep. The entrance to these dormitories, slightly raised above the floor, is a small hole, barely capable of admitting a person to creep through." I have found nothing in those examined to

Fig . Interio8 . Botf o r t Garry-na-hineha , showing ambrie r bolesso bedd an , .

answelattee th o rt rdescriptio n : indeed e Garry-na-hinth n i , e example strae heatheth d wan r bed .a lin f bankestonesy o eb p u d, lies against e walth l opposit e fireplaceth e (fig. 8) . The slabs of the present building are unusually long. About 4 feet up six complete the circle, and one of these is 4 feet 3 inches in length. course Th e above these projects considerabl stone fee2 on inches 8 e tyi : s widee roofTh ,. contrar customo yt threy s closeb i , p e to dston e onth e slabs; the longest, in the centre, have a span • of 2 feet 8 inches and a widt f 1ho 1 inches. 182 PROCEEDINGS OF THE SOCIETY, FEBRUARY 8, 1904.

e bothThith s i sdescribe a mor n i de summary manne y Captaib r n Thoma s Botha s , Cnoc Dubh, Ceann Thnlabhig, d judgean 1 d from its- peculiar features and from personal information to be of comparatively recent construction representatioe Th . chimneye th f no , howevere th n i , section figured by Captain Thomas is rather misleading. He shows a carefully constructed vent opening on the roof of the building, whereas it appeared to me a roughly formed aperture in the solid base completed

Fig . Bot9 . Eilean ho n Fear Chroithir.

in an unskilful, patched fashion from the outside (indicated by a slight e mattee datth Th f o ef o . e rsid 7) projection on figsi e , e 6 ; th n o n building I shall refer to presently. The foundations of a companion structure are clearly traceable a few feet farther east. My last example occupies a lonely position on the east side of a small island outsid Little eth e Berner e moutth t f Loca o ah h Eoag t I wa. s complete, minu e turth s f covering yearw s fe sinc ha sa , et agobu , collapsed (fig. 9). It has a floor diameter of 6 feet, converging to 4 feet 1 Proc. Soc, Ant. Scot., TO], vii. p. 161. STRUCTURE ABCHAIF O S ISLANE C TYPTH LEWISF N DO I E3 18 . e standinth f o gp portionto e inche8 th t , a swhic abous hi fee4 t t high. entrance Th e e wholefaceth n s o e bothgooa , east , Th is d . deal smaller than is ordinary. The island is locally known as Eilean Fear Ghroifhir, " the island of the men of Croir "—a farm at the northern extremity of the Great Bernera formerly occupie croftersy db . They came s herewa I , informed, to collect shell-fish, and no doubt also to utilise the patch of pasture. When we seek local information as to the construction of these bothain we meet with nothing but professions of absolute ignorance. " Wher- ever placed," writes Captain Thomas nativee th , l "al s agree e on tha o n t know builo wh st them thad an ,t they e werfather t madth r eno y b esno grandfather f persono s living.w no s " This general statement, however, he qualifie reportiny b s assertion ga builnpersos a "y o wa thab t e nwh ton is still alive" (1858), and, later, gives as his conclusion "from various circumstances" that the. both at Cnoc Dubh "was made about ninety years back" (1867).1 Struck with thes uniquo etw e exceptions I ,mad e careful inquir n informana f o y n ever i ty excellentl wa y y qualifieo t d speak on the subject, with the result that he refuses to admit the exceptions reiterated an , e entirsth e local ignoranc e origith f f theseo no e buildings. " My grandfather," he writes, " whom I well remember, lived heramond ' beehives ean e gth year5 fathey m 10 ' s: r live years7 d8 d an , I am 50 years—perhaps the longest pedigree that can be found in Lewis or Harris." The whole district which I traversed in his company is thoroughly familiar to him, and he himself lived in his youth in the "beehives" of Glen Marstaig. "I have to say," he continues, "that the ' beehives' were as ancient and mysterious to my grandfather as they were to my father and me. There is not the slightest recollection of eve e traditioo builth n wh t f themo n y moran , e tha f thoso no wh e Druidicae Dund th ol builr so e th t l e particulaStones.th n O " r point t issuemphatis i a e eh declarinn ci g tha cannoe h t t accep e statementh t t recorded by Captain Thomas—which, I may note, has passed into general

1 Built by a man Smith from Callernisli, says a local informant, practically cor- roborating Captain Thomas's statement. 4 18 PROCEEDING F THO S E SOCIETY, FEBRUAR , 19048 Y .

and unqualified acceptance;: adding that all it may mean is that the havmay nema helped "to renovat brind an eancienn ga t 'beehivea o t ' modern summer shieling, as many now living here have seen done. Captain Thomas's story must have originated from such tampering." The Glen Marstaig ' beehives' with which he was personally familiar were, he informs me, " reduced to summer shielings in 1832, and remained in .use for that purpose until 1872, when the people were deprive grazine th f do orden gi r tha groune th t d migh convertee b t d into a deer-forest." From a recent Bluebook -we find further that the Scaliscro deer-forest, to which, in part, this addition was then made, was formed in 1850 t ,i having been previously, unde e Seafortth r h ownershipa , sheep-farm groupe Th . s above Loch-an-Sgui Locd an r h Choirigeroe dar included in the Morsgail deer-forest, constituted out of a sheep-farm also abou same th t e t thestimeBu e. bothain, informany m t tell, " wersme e never occupied within the memory of living man." recorde Th f earlso y traveller e Highlandth n si s affor scarcels u d y yan light on our subject. Wherever we meet with any description of a typical Highland dwelling, invariable w ya for havn mi t ei stil l familiar to us (fig. 10). The modern shieling or airidli, when it is not a mere turf hovel simpls i , y suc hdwellina miniaturn gi e (fig . heae 11)f W o r. e openin th e shielin t th a f gn o th geme season earl Junn i y e settint ou g with wood, heather ropes, etc., for the repair of the temporary dwellings. This is a contingency that could scarcely arise in the use of stone-roofed e crofterbothain.th r ,Fo indeed mose th e roo s tth ,i fvaluabl e parf o t his house. When he is shifted from one site to another he carries his roof with him, 2a natura l econom countra n yi y where growing timbe- un s i r known witnese On . s from Harris stateCroftee th o dt r Commission that when the people of his district in his youth went to the shielings, the huts hille th sn wero thed ey ha " like ordinary houses buil stonf turf.o td ean " 3

A reference to beehive structures is almost certainly intended in the Cf1 . Prehistoric Scotland, Munro, p. 338. 2 Evidence Crofter Commission (1884), vol .. 903iipp . , 905, 930, 943. :i Ibid.,-p. 853.

186 PROCEEDINGS OF THE SOCIETY, FEBRUARY 8, 1904.

words of Martin (c. 1695) speaking of Skye, where none, however, is said now to exist. After describing some "earth-houses," he writes : " There are several little stone houses built above ground capable only of one person, and round in form. . . . They are called Tey-nin-druinich, i.e., Druids' House." A similar attribution, significant, like the Greek " Cyclopean," only of helpless ignorance, occurs in Pennant, who, in his description of the Broch at Glenbeg, Inverness-shire, writes : " Almost contiguous to this entrance, or portico, was a small circle formed of rude stones, which was called the foundation of the Druids' houses. ... I was told there were many e kinotherth f do s scattered ovee valley.th re Th " " grotesque groupe" of shielings which he describes in Jura were con- structed of "branches of trees covered with sods."1 Pennant indeed gives a description with illustrative drawings of what he calls a Pictish building in Caithness, as supplied to him by Mr Pope of Reaye appearancTh . s a tha"i ef beehive,o t e dimensions—th " a wal 4 fee1 l t thick enclosin n are a gf '22o a feetd measurinan , e th o gt sprinarce feet—wit2 th h 1 f go e grounth h d plan, showing eight lodging rooms of an oval form in the heart of the wall, suggest rather a Broch. I nee t enteno d r intdiscussioy oan typee f theso th f neo bothain,s a that has been thoroughly dealt with elsewhere.2 This primitive ap- proximatio e "archeth o nt d roof" extends from Greec o Greenlandt e : abundant examples are found in Ireland, where its pre-Christian character bees ha n fairly established. O'Flaherty, writin f thosgo Connaughn i e t n 1684i , speak f theso mtermn i s identical with those derived from other sources : " nobod thef made.s o ymy wa know an "lono w gag sho We may note also the examples from Holyhead Island in Wales,3 and that at Chysauster, Gulval, Cornwall,4 which have been placed in same th e category l thes Al e circula.ar e dry-stond an r e built t havbu , e

1 Creel Houses Boswell'e Se , s Hebrides,e Tourth o t 30th August. Proc.e Se Soc. Ant., vols. iii. . Mitchell', A vii. r Si ; s Past then i Present; Munro's 2 Prehistoric, Scotland. 3 Archceological Journal, vol. xxiv. 4 Ibicj,, vol. xviii. STRUCTURES OF ARCHAIC TYPE IN THE ISLAND OF LEWIS. 187

about five time e amounth s f flooo t r spac e Hebridesf thoseo th n ei , with the walls, to which they are all practically reduced, three to four times a& roosurviveds o n thick ha fs A e questio. th , n whethe casy t wasan ei n i r- formed in the familiar " beehive " fashion of enaorbellation is admittedly doubtful; while the extent of floor space to be spanned, varying from 4 feet3 o , t increase 5 1 improbabilitye th s , especially whe e notnw e the- polygonal character of the stones used. No doubt, however, attache e methoth o t sf constructio do f similano r dwellings in the islands of the Mediterranean basin, where, says Mr Arthur Evans, there exists "a continuous zone of the Cyclopean tradition of domed chambers." From the mainland, too, we have precise testi- mony. ProspeM . r Castanier s Hi*toirehi n i , de Provence dans UAtitiquitt',1 writes :— " On trouve, encore, en maintes regions du Bas-Languedoc et de Provence, des cabanes rustiques, placees generalement aux angle s champsde s t que , i ont gard a forml e e antiqu e: elles s'elevent sur un plan circulaire jusqu'a nne hauteur de deux metres ; a ce point nait la calotte de four qui les reeoudre et que forment des moellons bnits, plats, poses en encorbellement et par assises- horizontales. Le plus souvent une large pierre plate serf de clef a la voute. Ces constructions n/ont ni cement i poutren , , ni tuile? i chevronn , s; elle s com- posent souvent une piece unique et possedent une seule ouverture, etroite et basse. Elles servant aujourd'hui d'abri x cultivateurau s r lepa s mauvais temp s; parfoi s ceux-c cacheny s' i t pour attendr gibier.e el " He states, further, that the same methods of construction are to be found in the fortified positions at La Malle, La Tourre, L'Andido and other places, and concludes that all these go back to one time, and are the work of one people. They are alternatively called " Celtic" or " Ligurian." And, just as in the Hebrides, we find the traditional work- manship persisting among the Italian and French shepherds of the Alpes- Maritimes, who still build circular stone huts after the beehive fashion, about 8 feet high and as many wide, contracting towards the roof, which is also of stone. Through the kindness of the Kev. J. E. Somerville, Mentone whoo indebtet ,m a mI r mos fo df wha o t I thav e sai thin di s . 1178 Volpp , , 179i. . . 188 PBOCEEDINGS OF THE SOCIETY, FEBEUAEY 8, 1904.

•connection, I am able to give an illustration of such a hut, one of •several said to be ancient, which has been repaired and roofed by the shepherds, who further provided it with the side openings from which they might shoot foxes who came to prey on the lambs (fig. 1 IA). Returning now to the bothain of the Lews, we may conclude that in all the cases which I have described we are dealing with the early shieling a peopl f o s e semi-pastora n theii l r mod f lifeo e . l Theal e yar

. Shepherd'FigHA . sMaritime Huth n i t e Alps.

on garrys, or grassy knolls, which in their freedom from the damp moss afforded a resting-place for man and beast. The neighbouring lochs teem with fish. Probably they were occupied for a longer time than e whercasw th e no ecustos i th e f pasturagmo e survives people Th .e depended mainly on their cattle for subsistence, and their winter •store of butter and cheese was made in the shielings. It is not necessar supposo yt e that every example, wherever found a shieling s i , . e elaboratTh interconnected ean d structure f identicao s l type,1 described 1 Probably "communal dwellings.• " STRUCTURE F ARCHAIO S E ISLANC TH TYP F LEWISN O DI E . 18&

by Captain Thomas and others, probably indicate, as I have already sug- gested more th , e permanent habitation neighbourhooe th n si tillee th f ddo groun d neae fruitfuan dth r l sea. Whil e latteth e r migh coursn i t e of time be abandoned, the former, as still satisfactorily serving a need, would continue, as they have done, in use to the present day.

STONE CIRCLES. Iii the matter of the stone circles of the Lews, that at Callernisb, which is rather a circle of small diameter with a central cairn and con-

Fig. 12. Concentric Circles near Calleruisli.

verging lines and avenues of standing stones, has, by its magnitude and imposing appearance, cast all others into unmerited shade. Yet even significance th d picturesquenesan e f thio s s majestic monumene b y ma t enhance consideratioy db face th t f slightethano o themselven i tw t t rbu s strikin interestind gan g groups stand w aboulo a e east a miltth n o ,o et ridge striking southwards fro e highroamth shore th f uppeo o edt r Loch Eoa g; whil e abou coupla t f mileeo s farther east, flankin Uie gth g road, are one shapely circle and some minor fragments, still erect, of another. Westwards e Grea th clif w f o lo tf , Bernerae againth n o , , stand three 190 PROCEEDINGS OF THE SOCIETY, FEBRUARY 8, 190.4.

conspicuous stones overlookin e Narrowth g f Earsliadiro s e merTh e. enumeratio f thesno e facts seemsuggestivee m o st . Moreover, these minor circles are not all of the same type. That nearer .the Callernis hconcentrio tw roa f s madi o d p cu e rings (fig. e 12)Th . outer consists of eight stones, and there are four lying. The greatest breadttriangulae on f o h r ston fee3 inches0 s 1 ei t . This outer rins gha a rather scattered appearance e inneTh r. ring s markei (fig ) y 13 . b d four stones f whico , he eas threth t n sido e e stand close togethee th ; r

Fig. 13. Inner Eing of Concentric Circles near Callernish.

fourth is opposite, rather to the 'north. Two of the former group are fee6 d t an respectivel 5 y above ground e solitarth ; y stonfeet8 s i e. 'There is no central mound or cairn or other enclosed feature. About a quarter of a mile due west, where the ridge, after sinking to e moorth , rises slightly e companioagainth s i , n circle singla , e rinf go jnuch taller, stones (fig. 14). Five still stand grounde th ; n threo .e ear e latter-Oth f , what look e mosth s t massive measure fee6 s inche2 t s fee 8 uncovereds a y inchet7 b s i sd thican , t whaka t would have been, when erect, the top. From this to a square stone opposite the diameter rjneasure e largesfeet0 6 Th s . t erect stone f triangulao , r shape0 1 s i , STRUCTUKES OF ARCHAIC TYPE IN THE ISLAND OF LEWIS. 191

fee inche2 t s above ground e mosTh .t interesting feature, however, is a large stone-built cairn occupying about half the internal space (fig. 15) d havin,an gsmala s centrit l n i chambee t shapepi r o rd likea large round-bodied bottle wit a hshor t neck n thii , s respect resembling the pit in the Callernish cairn, which is, however, rectangular. The highest surface of the cairn is about flush with the surrounding peat, whic s beehha n cleared away from both circle a dept o t s h varying from

Fig . Circl14 . e near Callernish, showin gfive fouth e f stoneso r . centrae inches8 th 1 foo1 n o I t tl . spac f thio e s circle roun caire dth n were found, when it was laid bare, four square holes containing charcoal.2 The circle beyond Garry-na-hine overlooking the Dig road is also conspicuously placed. s formei t I y fiv1b de stone l abousal ) (figa t16 . size ,southere thath t ta n extremity regularit,e wherth n i ep ytherga a s ei of their positions, bein gfee8 inche9 t s 3 feehig inche2 y t b h s broad, and varyin thicknesn i g s fro m6 inche t bottoa s o aboumt t 8J inches e neatopdiametee Th th e closelr. th f o yr circular enclosed spacs i e 1 Tursachwn Ceann Thulabhig. 2 Proc. Soc. Ant., vol. iii.

STRUCTURES OF ARCHAIC TYPE IN THE ISLAND OF LEWIS. 193

about 30 feet 6 inches. In the centre is a rather dilapidated cairn abou fee6 t t across ,2 fee wit d t a hstonen hig n ho e (fig. 17). Quite two feet of turf have been removed from inside the ring of stones. Less tha nquartea mila f o r e e othe intmooth e oth n e r o rth sid f o e road are three small stones standing about 3 feet above the moss, the remains of a circle of which all other portions have disappearedl (%. 18). •

Fig. Cair17 . centrn ni f Circleo e beyond Garry-na-hinc.

All the stones making up these groups are formed of the schistose t appeano gneise localityo havo th t d r f t seo bu bee, n quarriee th n di immediate neighbourhood. DUNS. The word 'Dun,' as cognate with O.E. 'tun,' modern 'town,' is applicable to any form of fortified position, the original ' town' also havinsimilaa d gha r significance exhaustivn A . l suceal lishf o tsite s 1 Tursachan Airidh nam Eidearan. VOL. XXXVIII. 13 194 PROCEEDINGS OF THE SOCIETY, FEBRUARY 8, 1904.

in Lewis has been compiled,1 and to this I purpose making but one addition, with elucidatory note certain so n others. As in the case of the circles, one example dominates the whole—the well-known Dun at Carloway, which is of the type of structure known as a broch. One side still rises to near the original height, and well shows the peculiar contour of the external wall (fig. 19). Its inner face display characteristie th s c opening e inter-murath o st l galleries -(fig. .20). e dooTh r sid , howeveris e , reducee lintee leveth th f lo o lt (figd . 21).

Fig. 18. Portions of a Circle near Carry-na-liine.

The solid wall here is 10 feet thick. Inside the passage rises to the central court, though the slope is no doubt largely due to accumulated debris. The diameter across the court is about 24 feet. The section expose righe th n td o contain lowese staire th ruiTh e f .sth tno gallery, with those above, extend l rounal s standine dth g portionwalle th sd an , mose ar t carefully constructe f easildo y handled stones pretty muca f ho size. It is 5 feet 6 inches in height. Captain Thomas in his elaborate description 2 speak sea f beehivo f t so ebasemente cellth n si , lower still. The section on. the left shows the twin walls (fig. 23). The outer is 1 Arcticeologia Scotica, vol. v. p. 365. 2 Ibid.

STRUCTURE ARCHAIF O S ISLANE C TYPTH F LEWIS S O DI E7 19 .

3 feet thick and the inner 2 feet 4 inches, while the interspace measures 2 feet—givin totaga l thicknes fee7 f inches4 tso . Dun Carloway e callee belongclasb th f whay o sdo t ma stshor e duns. A more striking example in site occupies the top of a small •headlan de northerath t ne Greapointh f to t Bernera, lookinn o t ou g •East Loch Roag and visible from Dun Carloway. It is known as Dun Stuidh, and has not so far, to my knowledge, been recorded. The site is

. FigPortio23 . buildinf no Stuidhn Du t ga , Great Bernera.

0 fee9 . t abov s approachei a level se d e an , d fro e lanmth d side ovea r narrow saddle-shaped isthmus, traversed by a deep chasm. The position thus resembles that of Dun Arnisfcean on the western shore of the main island, below Dell, which is even more difficult of access. •A portio e inne th f waln o n ro l fac f whao e s practicalli t n isolatea y d •stack is all that remains of the original defences of the latter. Of Dun Stuidh, however, the ruins are considerable though chaotic, and e outsidoth n e grass-grown. Concentric wall e clearlar s y traceable, and to the right of the entrance which opens on the east side, looking 198 PKOCEEDINGS OF THE SOCIETY, FEBRUARY 8, 1904.

seawards a smal s i , l chambered portio 2 feen 6 inchet s wide which has escaped being stopped up by the tumbled masonry. Opposite the land approac e oute hpatca th f ro hbuildin g shows clear (fig.e 23)Th . distance fro e extremmth e e extremouteth o t r e inne cone rth fac -f o e centric walls varieapparene th feet0 d s1 an fro,o t mgreates9 t diameter of the structure, including the walls, measures 39 feet, giving for the inner court a diameter of about 20 feet. The ground plan is roughly circular.

• SHpTTWa: «*v«i'. "ifllr ' "»

. FigPortio24 . buildinf no t Duga n Nicisabhat, Kirkibost.

Reduced to little more than green ramparts one within the other, but showin place masonre on eth t ga y rising fro e rockmth y site (fig. 24)s ,i Dun jSTicisabhat in Kirkibost, on the southern shore of the same island of Bernera. Captain Thomas gives an alternative name of Tidaborough. s stoneIt s have been utilisee constructioth n i d f adjoinino n g houses. t riseI s fro edge a steem th f eo p escarpment abou a quartet mila f o er froe e shorenort e th th mentranc Th hn / o sides s extremi It e . e diameter is between' 50 and 60 feet over an irregular grass-grown enclosure, while the inner court measures about 30 feet -inside the wall, which has an apparent thickness, at the base, to which it is reduced, of STRUCTURE F ARCHAIO S E CISLAN TYPTH F LEWISN DO I E 9 19 .

nearly six feet. In its present condition, however, a superficial view of materiae th l affords little tha definites i t . My next example helongs to the class of lake duns such as Dun Barvas, Dun Shadir, and many others in the Lews. It stands on—covers, indeed—a little islan n Locdo h Bharahhae th e souther th f o t a d t nen Great Bernera—abou 0 fee8 t t froe shoremth , accordinor , Captaio gt n Thomas, fathom0 2 1 s froe grasth m s i approache line t I . d ovea r

Fig. Du25 . n Bharabhat, Great Bernera.

causeway, now appearing above the water as starran or stepping-stones (fig "e . Traditionth 25) n I .Macaulaye th f so s " 2 Captain Thomas notes that he has never seen such stepping-stones " placed in a straight line, but alway a curve, n i s " whic considere hh s "an obvious advantagn i e defence.e presenth n I t" case causewath e s certainli y y curved, more probably following a slight ridge in the lake bottom ; but the width and solidit e approacth f o y h throug e comparativelth h y shallow wateo s r near shor enegativ o appeart e notiom y o ean st defensivf no e stratagem. 1 Archceologia Scot., vol. v. p. 391. - Pros. Soc. Ant. Scot., vol. xiv. p. 363. 200 PROCEEDINGS OF THE SOCIETY,'FEBRUARY 8, 1904.

On reachin islane gth e find w d , restine nativth n eo g rocwitd kan h but a narrow margin between it and the water, a thickly lichenecl wall risin a height mosa o t g0 feef 1 o t8 inchet t extendind an s g for abouroundy wa quartea t e . tumbleth w Thereaftef lo o r a ds i wal t i r l of irregular shape t followi s a , e contou th islandse th f ro ^ enclosin gmasa s of loose stones overgrown with honeysuckle and brier. The ground plan is thus difficult to get clear; but such as it is, it may be found in the pape f whico r hI hav e spoken above. diametee e areTh th 1 af withio r n

Bharabhatn Du Fig. 26 . , interior. the extreme walls varies from 25 to 36 feet. The central stones would seem als indicato ot structurea f someo e sort, .whose natur purposed ean , however, cannot be discerned. Between it and the line of wall to the south is a crescent-shaped space, measuring at the most about 18 feet. e erecTh t e nortportioth n hf o wal o n side s i l , facin e steppinggth - stones d shows'twan , o openings e loweth , r hidde y theb n ' vegetation. These have an average width of about 2 feet 4 inches, and together give an open space 6 feet high-(fig. 26). From 'this westwards the wall'is solid, but to the east runs a .gallery flagged on the 'top—which forms 1 Arckceologia 'Scot., . 391volp Plat. d v ., an e 48.' STRUCTURES OF ARCHAIC TYPE IN THE ISLAND OF LEWIS. 201

uppee th r surfacwall^ie th f o eusuae nth l broch fashion chokes i t I d. with fallen material (fig. 27)t woulbu , d seem s Captaia , n Thomas affirms1 e ruith n w •—aninsa rathee dh r better condition—to have been occupied

Fig. 27. Dun Bharabhat, interior of gallery.

e solia stair Th y e dotheb .th wal n r o l side would see o havmt e flanke e entrancth d e (fig 8 fee . s ti 28) thict I . k abou fee6 t t froe mth base Dun I .n Bharabha e appeae dealinw b t o t r g wit brocha h which 1 ArchcKologia Scotiea, vol. v. p. 392. 202 PROCEEDING E SOCIETYTH F O S , FEBRUAR , 19048 Y . either possessed certain peculiar s featuresubjecte s ownwa it r f o o ,s d to some later modifications of the original structure. deao t I havlw witduno eno htw s which t plainlcomno eo d yunde r

. FigDu28 . n Bharabhat, southern extremit wallf yo . the category of brochs. On the main island, off the eastern corner of Bernera n BaraghloumDu s i , t I occupie. e cresa smal th sf o t l bluf5 2 f feet above sea level, with half a mile of narrow sea loch on its west side and a short bight on that opposite. A ditch or dry moat apparently STRUCTURES OF ARCHAIC TYPE IN THE ISLAND OF LEWIS. 203

filled up with debris cuts acrosn eithe o e level d th e sr an th , sid e ar e foundations of a curvilinear building of stone connected across the hollow by a line of stone causeway. About a third of the length a transvers y b f of o et f wallcu eac s i h, dividin t i gint a olarge d an r a smaller compartment e lattey takth , f o havma whict o er e ew h been a sleeping chamber. Enclosing the whole is a broken line of circumvallation. The structures are parallel to each other and point roughly east and e westoute Th s rathei r. r pear-shaped whild an s ,wester it e n division

Fig. Roug29 . h sketch-pla llaran Du f .no (Scale inch.n a fee4 o 2 ,t t )

measures 18 feet by 10 feet 5 .inches, the other has mean measurements e inne feet0 Th 2 fee 0 ro.y 1 fbuildinb t fee0 2 t e easter s i longth n gi n compartment and 7 feet 2 inches in the other, while it has an average breadt f abou o he inne feet2 Th 1 t r.e roughl th face f o s y built walls are mostly bare, but the outer are grass-grown. The thickness of the walls, reduced to their present condition, does not exceed a few feet. narroe Ath t w bridg ditce fee3 1 eth ts hi wide . A similar arrangemen f buildingo t s withi a ngrass y e ramparb o t s i t found on the summit of the headland known as Dun Mara (fig. 29) on the west coast of the island below Cross, Barvas, and a few miles from the 204 PROCEEDINGS OF THE SOCIETY, FEBRUARY 8, 1904..

circumvallatioe ButtTh . n her clearls ei y define averagn a t da e distance of about 6 feet from the remains of the stone buildings. To the right of entrance th workse whico th t o p et u ,h ther a gentl s ei e rise e tracear , s osemicirculaa f r hornwor interioe th f earthko f O r .structures , botf ho whic dividee har d transversel precisela n yi y similar manne o thost r f o e e leffirse th tth e compartmen to it son case e 2e 4on ,th feer d fo tan t e otheth feer 8 r fo e longeto th n r axis, easwestd an t , wit averagn ha e width of about 12 feet; while its companion is 19 feet by 10, wite smalleth h r sectio n additioi n n rather ill-define d brokean d n i n outline. thesn I e twin structures surrounde n earthea y db n rampar seee w tm to have to do with a type of fortification hitherto undescribed so far as e Lewth concerneds si . Captain Thomas reports1 d thaol te "onth f o e people" informed him that Dun Mara had been " as round as a bottle," and that eighty years before that time it was "quite entire," in which informatio e Captaith n n comment s " certainlwa e h s y mistaken.o S " e stonth fas ea r portion concernede ar s I hav, e observed their likn i e groune th d plan, though generally there associate n groupi d s connected by low passages, in the island of Berisay, which Neil Macleod fortified held threr an dfo e years (1610-1613) agains e Mackenzieth t f Kintailso , t lasa whtd obtaineoha d possessio islande knows i th t f I no . n froe mth indictment at his trial that ISTeil ran " divers, forays" on the main island, and these constructions in the neighbourhood of fertile districts may.have.served as bases for this purpose.2 In 'any case, it is not likely, that they-ca datee nb y timdan e posterio e lasth t o great r t internecine struggle which convulsed the Lews. 1 Arclueolorjia Scotica, vol. v. p. 372. 2 There is no water supply in either case.