Scott

GLASGOW
ARCHAEOLOGICAL
JOURNAL

VOLUME 10 1983

SUMMARY

The site is a fortified promontory on the W shore of Loch Fyne. Traces of a wooden palisade and huts were found under the rubble-cored stone rampart. The rampart, for which a C-14 date in the 16th century ad was obtained, was fired and partly vitrified, suggesting that it had been strengthened internally with timber. After a period of abandonment the wall was re-fortified, then abandoned again. Later dwellings may have been inhabited as late as the 17th century. A 12th century crucifix is the most important find and will be described in a later article.

INTRODUCTION

The site and its setting
Macewen`s Castle, on the W side of Loch Fyne (NR/91587955), has long been recognised as a defended promontory site, but the exact nature of the buildings within the ramparts was unknown until the excavations of 1968 and 1969.

The promontory is windswept and must always have been so, which is probably why so little soil has accumulated and why the various occupation levels lie so close one on the top of the other. The defended site measures 88 ft on a NE-SW line by 72 ft on a NW-SE line. The area of occupation is level but the ground falls steeply away on three sides. A considerable amount of tumble lies at the foot of the cliff below the rampart, especially at the W side. The grass of the foreshore to the SE is 45 ft below the rampart. The way up to the fort from the burn which runs down to the sea to the SE, and which was possibly the water supply, seems to have been improved by dislodging some of the rock, so making access easier. It seems possible that the rocks to the SW, between the sea and the cliff of the fort, have been broken down, this time to make access more difficult (this was pointed out by R B K Stevenson when he visited the site). A slight bank and shallow ditch outside the line of the rampart can be traced across the neck of the promontory. The solid geology of the site is quartzose-mica-schist. On either side of the headland are shingle bays. About 1/4 mile to the N is the ruin of a Salt House. To the SE the bay is small with a burn running down on the N side. Just above the foreshore the outline of a small building can be seen, 21 ft x 16 ft 10 in with a division through it 11 ft from the N end. The longer sides are parallel to the shore. From its position, so close to the sea, it may possibly have been a Salt House. These Salt Houses were used at a time when salt was an excisable commodity. It was stored under the control of Customs Officials in a lock-fast building, on the shore where the salt would have been accessible for the herring fisherman. At the S end of the little bay a structure is built against the rising ground. It is 23 ft 8 in long and 7 ft wide, open at one end, the other end closed by a wall. The S wall has some large stones in its build (one of them measuring 20 x 23 x 18 in). It now stands 3 ft 10 in high. The N wall is in a more ruinous condition and stands 2 ft 4 in high and 4 ft 8 in broad. Of the various suggestions put forward as to the use of this structure the most probable is that it was a saw pit such as can be seen in a less ruinous state at Portavadie. On the once cultivated land 1/4 mile E of the fort is the ruined settlement of Ardgadden. The area between Kilfinan Bay, the modern road and the burn from Achnaha has several of these deserted settlements which were probably abandoned in the late 18th or early 19th centuries.

The land round the shores of Loch Fyne has always been attractive to settlers. Within walking distance of Macewen Castle are Neolithic burial cairns, Bronze Age cairns, standing stones, duns, vitrified forts, a Celtic chapel, a Medieval motte and a castle.

Other local sites
The Macewen`s Castle ramparts are lightly vitrified. A very much more intensive vitrification can be seen 5 miles S at Caisteal Aoidhe, Ardmarnock (NR/909710). This is a fortified island connected to the mainland by a causeway. The walls were found to contain a wedge-shaped mass of vitrified material 6 ft wide at the top and 4 ft at the base, each side resting against a dry built rubble wall and founded on loose boulders (Honeyman 1890; Christison 1898). This implies that Caisteal Aoidhe had walls very much more firmly built and with more timber in them than those of Macewen`s Castle.

Across Loch Fyne 15 miles N is another vitrified fort on Pennymore Point (NN/047005) where lumps of vitrification can be seen on the E. The finding of iron slag points to a multiperiod occupation (Paton 1928; Campbell and Sandeman 1961). At Caisteal na Sithe (NR/962691) 7 miles S is a D-shaped enclosure. Investigation of the rampart showed that it was faced externally with flat slabs so loosely coursed as to suggest a combination of stone and timber work (Childe 1943, 42). This site is now covered by forestry planting.

Barr Iola (NR/938828) 2 1/2 miles NW is a hill fort with a stone-faced wall 10 ft thick linking with masses of outcrop to enclose an oval area 73 ft x 55 ft. Just inside the entrance, built against the wall, is a cell-like structure which might be paralleled by the structure found beside the gateway at Macewen`s Castle (Childe 1931, 419). The site was not excavated. Near it on a ridge is Barr Lagan (NR/942522) a fortified site (DES 1960, 13). There are two small duns within 200 yards of each other (NR/920814 and NR/921816) two miles N (Argyll CC 1915). There is a small dun on the top of an outcrop of rock at Fearnoch farm (NR/930816). The peninsula at Newton (NS/034981) is defended by a 10 ft wide wall enclosing an area of 70 x 60 ft with a further defence across the neck of the isthmus (DES 1965, 9).

Well-defined stone foundations of two hut circles lie between the two small duns and Macewen`s Castle (NR/925804). Some time during the last war a narrow trench was taken through the S one by the late Dr John Rankine. Traces of a hearth and some pottery were found. No report exists.

North of Castle Lauchlan 12 miles N are the remains of the Celtic chapel at Kilbride (NS/007967). The chapel lies so exactly equidistant from the N and S wall of the encircling cashel that the excavator felt that the two were contemporary. Traces of other buildings can be seen both within and beyond the enclosure. Only the chapel was excavated (Knight 1933).

The impressive motte Croc mhic Eoghainne (NR/922 833) 2 1/2 miles N is near the present Ballimore House (Childe 1943). On the top are Campbell burials dating from 1680 to the end of the 19th century; they include Campbell of Otter, Campbell of Ballimore and Burnley Campbell of Ormidale.

The present church at Kilfinan has traces of earlier building, one of the windows dating from the 12th century. There are stones in the graveyard of the 12th to 16th centuries and later. On the Lindsaig burn one can see traces of an early mill. The present Lindsaig mill, dating from the middle of the 18th century, succeeded one built more than 100 years earlier. The one at Fearnoch which has recently been converted to a dwelling house was a farm threshing mill built in the middle of the 19th century. The many foundations reported by Mr A. McIntyre on Lindsaig farm and seen in the surrounding countryside are evidence of continued occupation in the area. All this shows that at no time in its long history was Macewen`s Castle an isolated place without neighbours. The information for this section was largely taken from the Gazetteer of sites in Cowal compiled by the Cowal Archaeological Society.

The Macewens
The headland on which Macewen`s Castle sits rises steeply from the foreshore giving a good natural defensive position. The land on Loch Fyneside which has long been known as Macewen country lies between the Kilfinan burn and Largiemore. Castle Macewen is published on OS maps as Caisteal mhic Eoghainn. The OS Name Book, compiled in 1862-77 gives as authorities the Old Statistical Account and Origines Parochiales which record that it was the stronghold of the Macewens, former lords of the Otter. There is another stronghold associated with the Macewens, a motte-like structure known as Croc mhic Eoghainn near the present Ballimore house.

The Macewens are descended from Aedh Anradham who married an heiress of Cowal and Knapdale in the 11th century. Four generations later Suibhne, in the early 12th century, held Castle Sween. His son Uufgal is known to have given a charter to the Monks of Paisley for two chapels, one in Knapdale and one at Skipness. He also sold the lands of Knapdale and Skipness to the Earl of Monteith. His son John sided with Edward I and Edward II in the Scottish Wars of Independence, perhaps partly because he hoped to recover possession of the lands of Knapdale if the English conquered Scotland. He spent much time in Ireland during these years of war, which probably explains the tradition that the Macewens came from Ireland. It was his grand son Ewen who gave his name to the clan in the 14th century. They settled on Loch Fyneside on land between the territories of the Lamonts and the Maclachlans with which clans they intermarried. About 1450 the Campbells ousted them from this area but, while the Macewens no longer held these lands, they scattered throughout Scotland and prospered. I am grateful to Mr Archibald McIntyre late of Lindsaig, Kilfinan, who has given me this information on the origins of the Macewen for use in this section.

Methods of excavation
The excavation, undertaken by members of the Cowal Archaeological Society and the British Girls' Exploration Society, was run as a training dig under the direction of the author assisted by Miss C Lythe. The workers were divided into three teams, under their own supervisors, on three separate projects, namely the Turf House (Site A), the Stone House (Site B) and the ramparts (Site C). Site D was examined during the second season. The work was done during two weeks in July 1968 and July 1969. Detailed interim reports were written and circulated after each campaign.

Before excavation the rampart round the promontory could be easily traced, but to the landward side it was not so well defined. Grass-covered foundations of two buildings (Sites A and B) were visible within the rampart (Site C). Stone foundations of another building (Site D) lay outwith the rampart to the N.

The rampart showed marked outer and inner ridges with a slight hollow between. Trenches CI and CII were cut through the rampart where this feature was most distinct. The area CIII was cleared to define the entrance, the probable guard room and a circular structure of a later date lying across the entrance. Where occupation material ran up to and under the rampart to the S a cut, AVII, was made to follow this level. Several other cuts were made to confirm the line of, and to examine the nature of, the rampart at other points.
 

The walls of Site A, found to be of turf, were cleared showing two opposing entrances and a bench of stones at one side. All but a small part of the floor associated with the turf walls was cleared. Under this level a complex of Phs associated with occupation soil was found. The Stone House (Site B) was cleared, the walls defined and the floor uncovered. Under it, as in the Turf house, Phs of an earlier period were found. At Site D a stone house with large irregular boulders in its walls and one bowed end was cleared. Planning, 1/2 in to 1 ft, was carried out at all stages of the excavation.

When the new OS 1.50,000 metric map of the area was printed the name Caisteal mhic Eoghainn was removed and replaced by Dun.

THE EXCAVATIONS

Palisade and hut circles
A level of dark soil with flecks of charcoal and chips of bone was found to lie under the floors and walls of the Turf house (Site A) and the Stone House (Site B). It continued, with less black in the soil and fewer chips of bone, under the rampart, where a post-hole was found. This post-hole (no. 2) and post-hole 1 in area CII, also under the rampart, indicate that the earliest defence was a palisade. No. 1 was cut 3 in. into clayey, heavy soil with packing stones round it. It was 10 in. in diameter. No. 2, cut 4 in. into the natural rock, was 9 in. in diameter with packing stones round it.
 

These post-holes found under the floor levels of the Turf house and of the Stone house and outside the wall of Site D belong to the first occupation of Macewen`s Castle. Two distinct types of holes could be distinguished. Nos. 3, 4, 5, 6, 7 and 8 in Site A, and nos. 10, 11, 12 and 13 in Site B and no. 14 outside Site D, were small and slightly built. No. 14 lies beyond the line of the rampart as seen today so the palisaded rampart may have extended further N than the stone-built one. All of those in Site A were cut into the natural clayey soil or into rock. Two of those in Site B were also cut into soft rock. It would seem therefore that this group belongs to the time of the palisade.

The post-holes of the other type were much larger and more substantially built; some of these were cut into the subsoil. One stone used in no. 16 was 22 in long. Nos. 17 and 19 had stones 22 in and 24 in long set into their sides. No. 9 in Site B was so much better made than the others in that area that it probably belonged to the period of the large holes. It was firmly constructed of narrow stones 12 in long, the tops of which were under the level of the foundations of the wall of the Stone house. It is difficult to use the smaller and earlier post-holes in Site A to reconstruct a hut or huts. Others of the series are bound to have been lost during the later occupations. The patch of occupation soil found outside the E side of the Turf house and going under it shows that the hut or huts probably extended in this direction. In Site B post-hole 13, cut 7 in into the rock, could have been the central post of a hut about 20 ft in diameter the walls of which included nos. 10, 11 and 12. Pieces of red burnt clay, possibly daub, were found at this level. The large post-holes 18, 17, 19, 16 and 20, could possibly lie on the line of the wall of a hut about 30 ft in diameter. No. 3 - 18 in in diameter, cut 3 in into the rock and with no remaining packing stones - seems too slight to have held a central post as its position suggests. Dark occupation soil was associated with the postholes but, except in one or two very small areas, it was impossible to distinguish two separate levels. The area of occupation of the period of the huts must have covered much of the flat ground on the promontory. The patch of occupation soil found under the cobbles of area CIII and the position of post-hole 14 stress this point.

The ramparts
The rampart is clearly defined round the edge of the promontory except for a stretch beside the Stone House (Site B) but is less clearly marked on the landward side. At no place does the wall rise more than one course above the present level of the turf. The rampart has marked outer and inner ridges most of the way round, and the depression between these ridges varies in breadth from 4 ft to 6 ft.

That the first defence was a palisaded wall was suggested by post-holes 1 and 2 found under the foundations of the rampart. The holes were shallow and not firmly packed so that the posts must have needed support, perhaps from an earth bank. The dark soil found at the base of the rampart may represent the remains of this. The posts were not so close together as to make a solid wooden wall. However, as the holes were found in the only two trenches cleared to bed rock, and since these trenches were only 3 ft and 5 ft wide, it is probable that the posts were not set far apart.

Where excavated the rampart was found to be set on earth and stones above the rock. It had an outer face fairly regularly built with carefully selected but undressed stones. The inner face was not so well built. The space between these faces, which were on average 5 ft apart, was filled with earth and stones, with a suggestion of a spine of stones. Among the stones were patches of dark soil, possibly the remains of packing with turf or scrub wood. On the outer face at CII, where the wall was well built 4 ft down the slope of the cliff, a horizontal slot was found. It was made by a long stone resting on smaller ones. The soil inside was very dark brown and there were small lumps of vitrified material. In the core of the wall and 2 ft below the turf line a transverse stone set on smaller stones covered another such timber slot; it too was filled with very dark soil. At the same level in CI dark earth with charcoal was found but there were no slots found in that cut. The impression the excavators had was that the walls had been timber-strengthened rather than timber-laced.

Trench CIV was cut where the rampart had been narrowed and was lower. It seemed that there might have been a sea-gate at this point. Excavation showed that this was not the case but that the same method of construction had been used as at CI and CII. On the landward side outside the rampart a low bank and ditch ran across the neck of the promontory. A trench 40 ft long and 2 1/2 ft wide was dug across this (off the plan). There was only a 9 in difference in level between the crest of the bank and the foot of the ditch. Bedrock was 1 ft below the turf at the lowest part of the ditch and 2 ft at the crest of the bank. There were stones in the soil of the bank which can never have been a very effective defence and was possibly made by shovelling up the soil from the ditch. It has probably levelled off considerably in the centuries since its construction.

The entrance to the defended area was shown before excavation by a closing of the outer and inner faces on each side of a lowering of the rampart. Excavation of Site CIIIa showed massive stones lying across the width of the rampart, 10 ft wide at this point. These stones defined the W side of the entrance. The E side had been largely destroyed, its line being just traceable. The space between was filled with a jumble of stones and earth, probably from the collapsed side walls and from later deliberate fill. At a depth of 2 ft 8 in below the turf line the stones were smaller and harder packed, and some stones were laid flat with clay among them. This may have been the floor of the entrance. Under this rough paving, dark earth overlay the bed rock.

In this trench there was a considerable amount of vitrified material, not fused blocks of stones such as are found in the cores of regular timberlaced walls which have been fired, but rather small loose pieces of stone with bubbles and droplets. Fragments lay under and among the loose filling, above the firmly packed floor, among the massive stones of the cross wall and they were more concentrated against this massive wall than elsewhere in the trench. This concentration implies an abundance of wood at the gateway and suggests a massive wooden structure. Some of the stones were covered with a green glaze so like that of mediaeval pottery that a small thin flake of stone so covered could very easily be mistaken for a sherd of a green glazed pot. Small lumps of vitrified material were found in all parts of the ramparts examined. What was completely absent was the hard core of vitrified material seen in regular timber-laced ramparts as at Dunagoil and Finavon.

Finds: Part of a pitchstone armlet (0.5 in wide and with an inside diameter of 2.3 in) lying in tumbled stone on the inner edge of the entrance 5 in below turf. A fragment of lignite, possibly part of thick disc, was 5 in below the turf. A quartz point was in the fill of the entrance. A chipped hone was found among the rampart stones at the edge of the entrance.

Rampart at CIIIb
To the E of the gateway the rampart swells to a thickness of 10 ft, contracting again to 5 ft further N to give the impression of a chamber within the wall. When cleared this chamber was well defined (12 ft x 10 ft) and edged with large stones. There was some obvious tumble above a packing of stones (8-10 in in size). This packing of stones was removed from the W half to a depth of 2 ft 6 in when a level of blackened earth and small stones was found suggesting a possible floor. This may have been the floor of a guard room beside the gateway though no entrance to it was found. Later it must have been deliberately filled and packed with stones. J G Scott suggested it might have been the base for some sort of a tower, perhaps wooden, built in mediaeval times.

Finds: A fragment of copper plate with rivets 0.5 in long came from tumble in the guard room. A group of 7 chuckie stones was found together.

Reconstruction of the rampart
There may have been a considerable interval between the firing of the rampart which caused the vitrification and the building of the Turf House. In trenches CI and CII it was found that an outer wall of large stones had been built against the rampart. The steep slope below this was packed with stones. Considering the steepness of the slope it is surprising that no buttress or lower spine of stones had been built to prevent the wall from slipping down the hill. Instead wooden posts were used to stabilise the outer rampart. This is shown by post-hole 23 built just outside the retaining wall 1 ft 3 in below its base and by two stake-holes filled with carbonised wood 3 ft below no. 23. This post-hole 7 in in diameter, 4 in deep and packed with small stones, contained carbonised material which gave a C-14 dating of ad 1530 ± 70 (GaK - 2046). Further reconstruction of the wall can be seen at the SE and SW corners which have been squared. J G Scott pointed this out as a typically mediaeval feature.

Finds: 3 sherds of mediaeval green glaze ware were 4-8 in under the turf.

Post-rampart structure (CIIIc)
A circular structure 20 ft in diameter was built across the inner side of the gateway and against the rampart. This structure had a well-built wall 2 ft 6 in thick enclosing a paved area, and it made use of the rampart as part of its own wall. The encircling wall survives nowhere more than two courses high. It had a well-defined entrance with long stones laid crosswise emphasising it. A hole for a door post was found at the W end. Just to the S side of the entrance the wall widened to 4 ft over a distance of 4 ft 6 in. This wider section of wall did not taper into the circular wall but joined it at right angles. Some four layers of cobbled paving were removed and each one had slightly smaller stones than the one below. The bottom layer was of flatter, largish stones. At the rampart side these lay on bedrock which sloped inwards. Earth seems to have been used to level up the site towards the bed rock. A depression just inside the doorway had much charcoal in it, and the cobbling at the S end was less firm and had only one layer in places. Under the bottom layer was black occupation soil. This may have been connected with the period of the hut circles.

From its relation to the rampart this structure must come late in the sequence of buildings on the site. There were no finds to date it or to give a clue to its function. That it was built with some definite purpose in view is certain. The cobbled paving would have ensured dryness; perhaps it was built to store grain.
 

The Turf House
The Turf House (Site A) measured 34 ft x 26 ft and lay on an axis NNE by N-SSW by S, the measurements being taken from the crests of the walls. It was oval rather than rectangular in plan. The remaining walls, 2 ft 6 in - 3 ft thick, were built of turf. The separate turf layers showed clearly, especially in the section at the SW corner where seven could be counted. There were no post-holes associated with the Turf House. The doorway on the E was 6 ft wide and had fine pebble cobbling at the entrance extending 2 ft inside. The cobbling was firmly and carefully laid and paving stones lay at the threshold. Opposite this doorway, in the W wall, was a narrow opening 1 ft 6 in wide which had two layers of turf across it. The turves in the wall had been laid sloping down to the sides of the opening thus showing that it was an original feature. No holes for door-posts were found at either doorway.

Two layers of turf were uncovered set against the inside of the wall at the N end. As these were set on the floor level and were overlaid by slip from the wall it seems as if there had been a narrow turf bench in that part of the house.

Some re-construction of the wall had taken place. In the trench cut on the S side a rebuild could be seen in the form of fresh turves set on top of collapsed ones. On the W wall to the S of the doorway stones had been set among the turves on the top of the wall. This was the only part where stones were found to be used in the wall. A platform of stones placed on a base of turves against the W wall may have served as a sleeping or working bench. A scatter of stones on the other side of the house may have served a similar purpose.

There was a hearth area between the bench and W wall; it had grey ashy material with red burnt clay and some charcoal, was edged by small stones to the N, but was substantially built. The floor which lay 9 in - 1 ft 2 in below the present ground level was unexpectedly nebulous. There was some cobbling, some tramped earth, some flecking with charcoal. The cobbling did not extend S of the bench. It was on this level that nails, a fragment of copper, stone discs, perforated stones and chips of bone were found. A plotting of the finds shows that there must have been a difference in the use of the two ends of the house. There were more nails and fragments of metal to the N and more fragments of bone and charcoal to the S.

On the floor level to the N and E of the main doorway a patch of loose pebbles, similar in size and shape to those used in the doorway, was noted. There is no real explanation for this feature, but perhaps they had been collected to extend or repair the entrance. A James I Fleur de lis groat found while clearing down to occupation level is dated to the first half of the 15th century. Unstratified finds found during the initial clearing of the turf walls included fragments of glass bottles and a 12th century crucifix 4.5 in long.

Finds: 12th century crucifix (to be discussed in a later article): unstratified. Fleur de lis groat of James I and II: unstratified. 5 pieces of green and amber glass bottles. Iron knife blade fragment 8 in long, broken. 12 fragments iron nails and nail heads and 1 large nail, all very much corroded. 2 perforated stones, perhaps spindle whorls, from the floor level. 3 chuckie stones found together on floor level. 3 small shale discs or counters, from 1.2 to 0.9 in in diameter, from the floor level. 2 schist discs, 3 in and 4 in diameter, from the floor level. 1 fragment of lignite, possibly part of a disc. 2 flint scrapers with secondary working. 1 flint blade fragment with secondary working. 3 flint flakes and chips. 1 quartz flake, possibly used as scraper.

The Stone House
Examination of the ground to the S of the Stone House (Site B) showed that the rampart had gone out of use before the house was built. In front of the E wall of the house the remains of the rampart had been flattened with a rough paving laid over it to provide a level space between the wall and the edge of the cliff. Bedrock just below the present ground level must have made it a difficult site to build on.

The Stone House lay on an axis NNW SSE (for the sake of brevity the end walls are referred to here as N and S and the long walls as E and W). It measured 24 ft x 12 ft within the walls which were on average 2 ft wide; they now stand 2-4 courses high. Some of the bedrock appears to have been flattened to accommodate the corners of the house. Tumble filled the interior of the building and spread outside the walls, though outside the N wall there were fewer stones. Specially large stones had been chosen to make the outside corners. No dressed stones were used in the build. At the N end the inside corners were rounded giving a slightly bowed appearance to that end. The bottom course of the wall appears to have been faced with clay, inside and out. The walls were set on a bottoming of small stones which served to level up the irregularities where the wall was built on bedrock. A doorway 2 ft wide was towards the S end of the E wall. The entrance had been laid with firm paving over which was a pebble cobbling partially tramped in among the paving. Over this was a later, looser paving. The doorway had been blocked so firmly that it was difficult at first to locate it. No other entrance was found. Just to the N outside the doorway was a platform of small stones, probably a butt stance.

A rough paving lay all round the outside of the walls, close to the one on the S, less well laid at the N. A half quern was used in this paving on the E side, 4 ft N of the door. This quern is 2 ft in diameter and has moulding round the central hole. On the other side is a recessed square.

Bedrock was very close to the surface everywhere inside the house. The charcoal-flecked soil of the early occupation associated with the postholes and huts had levelled this up somewhat. There were two concentrations of black soil of this period, one in a hollow of bedrock which ran under the wall at the NW corner. The floor associated with the Stone House, made of hard packed pebbles and extending over the whole floor, was less firm towards the S end. Two levels of cobbling could be distinguished in places, with more charcoal-flecking among the lower cobbles. Rough benches were built against both ends of the house, the one to the S being better built with flat stones set above smaller ones which levelled up the bedrock. A half quern was incorporated in the one at the S end. In the centre of the house a complete quern stone had been laid among the cobbling. This quern, 3 ft in diameter and larger than the usual hand mill, is probably from a click mill (a small mill set across a stream using a horizontal wheel). It had been used as the base of a hearth and was made of extremely friable stone which disintegrated after being uncovered. Quern stones are a feature of this house, three other fragments being found there.

An area of firm clean clay lay against the E wall obscuring the doorway and spreading fan-wise into the house. At the centre of the patch against the wall it was 9 in thick, it ran 7 ft along the wall and reached 5 ft into the house. Clay was also found against the W wall opposite the door.

Finds: One small sherd with black shiny glaze, Staffordshire type ware, probably mid 18th century. Schist disc 4.5 in in diameter. 5 fragments of iron nail heads, much corroded, found under the tumble of stones inside the NE corner. 1 fragment of a pitchstone blade.

Site D
The position of this building seems to indicate that it was constructed when the ramparts were not needed for defence. Shelter from the wind may have been sought as it lies in the lee of the main site. The house lay on an axis WNW-ESE (for brevity referred to as NW and SE in this report). It measured 20 ft x 12 ft inside. The walls were set on a foundation of small stones. Large, irregularly-shaped boulders were used in the walls along with other more conventionally-shaped stones. Remains of turf found among the tumble showed that stone combined with turf was used in the upper courses. Turf would certainly have been needed to fit into the irregularities of the large stones used. These large stones were more prominent on the NW and NE sides; on the SW side the smaller stones, in places two or three deep, were not built in regular courses. The house had a straight NW wall and a slightly bowed SE end. An entrance at the SE end had been blocked with small stones. Another doorway, one half of which remained, was set in the NE wall. There was a hole for a door post just inside the wall. Bedrock was nowhere far from the surface in the interior. There was a possible bench at the NW end but this part of the house was not cleared to bedrock. Otherwise the whole area inside the house had a rough cobbled floor in which two periods could be distinguished. Firm cobbling from the door in the NE wall led to an area of hard-packed black earth with small stones. Since this hearth area, 4 in deep, overlay cobbles it belonged to the second period of occupation. Outside the NE wall and E of the door another area of burning was uncovered. This patch of red burnt earth with some charcoal ran under the wall so may have belonged to an occupation before the house was built. Outside the NE wall, just E of this burning, the soil was very peaty; this might be the remains of a peat stack.

Finds: 5 iron nail heads, very much corroded, from tumble at the NW end. Flint scraper from inside the NW wall. 2 stone discs 2 in and 1.3 in in diameter, from above cobbling.

DISCUSSION

The site of Macewen`s Castle is an attractive one with good land behind it and bays on either side, and the promontory itself has wide views up and down Loch Fyne. That there have been people living on it for a very long time is shown by the flint artefacts found scattered all over the site. There were three scrapers, one of them definitely Neolithic in character, a small broken blade, Mesolithic in character, as well as four flakes of flint. Flint is not native to the area. There were also three flakes of quartz which were possibly used, two as scrapers and one as a point. These were similar to the quartz artefacts found at Auchentagan, Glendaruel (Marshall 1977).

The range in time of the other finds, from the 12th century crucifix to the Victorian glass bottle fragments, underlines this continued frequenting of the site. Nothing dateable was found in an archaeologically sealed deposit.

The palisade and huts
The earliest evidence of occupation came from the group of small post-holes set into natural soil or bedrock and which were linked by a charcoal-flecked level to the palisade post-hole under the rampart which ran round the promontory. Although only two of these palisade post-holes were found, from the position in which they were found - in the only two cuts in the rampart which were taken down to bedrock - it can be assumed that the posts were not set very far apart; neither did they make a solid timber wall. As the postholes were small and shallow, some support would have been needed. Earth underlying the stones of the rampart suggests that this may have been provided by an earth bank.

The use of wood to form a palisaded enclosure is a basic idea and has been so used from Neolithic times right up to the present day. Palisades surrounded homesteads and settlements in the first millenium BC (Ritchie 1970). Palisades on defensive sites which were succeeded by stone ramparts have been found at Balloch, Kintyre (DES 1975, 10) and at Craigmarloch Wood, Renfrewshire (DES 1965, 34).

The stone rampart
The palisade was followed by a timber-laced wall set within the earlier defence. At Macewen`s Castle the wall was built along the line of the palisade.

This palisaded defence may not have lasted for very long. The settlers in the huts with the small post-holes may have built the huts with the large post-holes or they may have been succeeded by another group of settlers. As the levels of occupation associated with the two types of post-holes were so close, one on the top of the other, it is unlikely that there was any great lapse of time between these phases. It may be assumed that the stone rampart was built on the line of the palisade by the people who lived in the huts with the large post-holes. This later rampart was found to have had timber and probably turf within the stonework. The short and irregularly-placed timber slots indicate timber-strengthening rather than timber-lacing. Excavation showed that there had been an entrance to the NNE 10 ft wide and 10 ft long with a solidly built guard house beside it. The vitrified stone found all round the rampart shows that the site must have been fired. Nowhere was there a core of vitrified material. The concentration of vitrified material at the gateway suggests a wooden gate.

The firing of the rampart which produced the vitrification probably brought this phase of occupation to an end. Then there was seemingly a period of abandonment until the Turf House (Site A) was built. With walls 2 ft to 3 ft thick, seven turves high, probably a thatched roof, a working bench, a hearth and a lightly cobbled floor it would have been a comfortable enough place to live. The doorway with its distinctive pebble cobbling had no door posts so may have been closed by a heavy skin screen hanging from the lintel.

After the period of abandonment the rampart was repaired and altered. The NE and NW corners were squared and massive stones were set against the face of the old rampart at the W. This outer wall was strengthened by posts. A C-14 date in the 16th century ad was obtained for charcoal from one of these post-holes. The tumble in the guard room was consolidated with other stones making a firm base. A tower, possibly wooden, may have stood on this as was suggested by Alcock (1969) when discussing the widening and flattening of ring banks beside entrances to defended areas. One can assume that the rebuild was done by the men who built the Turf house. The lack of stratified artefacts which are dateable makes it difficult to date the Turf house and the reconstruction of the rampart. The Fleur-de-lis groat is of the 15th century. Two of the sherds found under the turf of the rampart at trench CII have been dated to the 15th or 16th centuries while the third sherd from the same place is possibly of 17th century date. The documentary evidence for the occupation of the Kilfinan area by the Macewens (Moncrieff 1967) states that they came from Ireland, where they had spent the years of Bruce's war, in the early 14th century. About 1440 the Campbells ousted them from the lands of Kilfinan and Otter. Perhaps after the Macewens had given up the land the Campbells built the Turf House and altered the rampart to establish their dominance.

Post-rampart features
Once more there was a period of abandonment. The now ruinous rampart was not needed as a defence by the builders of the Stone House (Site B) who flattened what was left of it to make a platform outside their door. The house was well built despite the difficulty of the uneven ground and the bedrock which was so near the surface. Two periods of use could be distinguished. In the earlier floor flecks of charcoal and fragments of bone were found which were absent from the upper cobbling, so it may have been used first as a dwelling and later as a work place. One feature of the house was the number of querns found, the click mill stone on the floor, two half querns and two other fragments found inside and the half quern set in the paving outside the house. While the development of Iron Age querns of Scotland has been worked out (MacKie 1971) and it is known that bun-shaped querns are early it is not yet possible to date mediaeval querns typologically. In the 17th century when water mills were built landlords charged for the grinding of the meal and naturally discouraged the crofters from using their hand-querns. In some cases they went further and actually broke the quern stones so that the laird's mill would have to be used. Something of this kind may have happened round Kilfinan and the broken querns were collected to use in and about the house. The little platform of stones outside the door may have been for a water butt or for a barrel to collect urine which was used as a mordant for dye.

It is difficult to understand why the doorway was blocked and what was used as the entrance after that. It is also difficult to explain the mound of clean clay set against the blocked door. It must have been collected with some purpose in view, perhaps to use as clay luting on the walls as is seen on the walls of the houses in Clachan, Albean forest near Tummel. Then something happened to prevent this being carried out, perhaps an epidemic, a series of bad harvests, or a change of purpose.

The circular structure (CIIIc) may belong to the same period as the Stone House. It was built after the rampart had passed out of use. Apart from that fact there is nothing in its build or in objects found to date it. There was no tumble round the well built encircling wall to suggest that it ever stood higher. As there was a clearly defined doorway with a door post it must have been an enclosed area perhaps with a brushwood fence on top of the wall. The layers of paving would have provided a dry base for storing purposes. The cobbling which lay to the E of the structure may have been part of a yard. The full extent of the cobbling was not established.

It seems that the rampart was not in use defensively when the stone house Site D was built outside the rampart and in the lee of the fort. It could have been built either before the turf house people reconstructed the rampart or after it had gone out of use. There is nothing to date the house either typologically or from finds. The walls were partly built with large irregularly shaped stones as was found in the long-narrow-house at Glenvoidean, Bute (Marshall 1969). There was nothing to date this house either. At Glen Lednoch, Perthshire there are shielings of three periods with the same irregularly-shaped large stones in the build of the middle period houses. The hearth indicated that it was used as a dwelling-house. Bedrock must have made an uneven floor even if levelled up with earth and stones. Perhaps we make too much of the roughness of these floors. The furniture, stools and benches may have had three legs, a design that adjusts to an uneven floor (Kenyon 1957). There is a 1660 record that Duncan Campbell of Ardgaden, having no peat on his hill, was allowed to cut peat at Strone in the Barony of Otter. The possible remains of peat stack found outside the N wall might be connected with this. If so, Site D would come late in the sequence of buildings at Macewen`s Castle.
A catalogue of the finds is in microfiche.

ACKNOWLEDGMENTS

We are grateful to the late Dr John Rankine of the Otter Estate and to his successor as owner of Fearnoch Farm, Mr Martin Bergius, for so willingly giving us permission to excavate the site and to camp on the hill above at Ardgadden.

The drawing office of the Royal Commission for Ancient Monuments prepared the plan of the site for publication. Mr Sam Rennie, head of the training of surveyors at George Wimpey and Co., building contractors, Bishopbriggs, planned the area round the site as an exercise in practical surveying. Mrs Susan Gell of Cambridge drew the finds for publication. The photographic Department of the Glasgow Art Galleries and Museums photographed the crucifix. All this help towards the illustrating of this paper we acknowledge with thanks.

We also appreciate the help and advice we have had during the excavation from visiting archaeologists, especially Mr J G Scott, and Dr Horace Hairfurst. Dr Anne Robertson reported on the Fleur de lis coin. Mr John Hume examined the nails and Mr Eric Talbot assessed the pottery.

We were fortunate in having a grant from the Rollo Lamont Fund through the Clan Lamont Society which paid for the C14 reading.

As director I give special thanks and praise to those who did the actual digging, members of the Cowal Archaeological Society and the girls of the British Girls' Exploration Society, an enterprising group organised by Miss Rowe of Skipton. The McIntyre family of Lindsaig gave invaluable help with the scything of the site before work began.

BIBLIOGRAPHY

This paper is published with the aid of a grant from Mr R W B Morris.