Previous interpretations of the courtyarded quadrangular structure surviving on the summit at Tarbert Castle, Argyll had suggested this was the earliest upstanding building on the site, and a potent symbol of increased crown authority in the region during the early-13th-century reign of Alexander II. This paper will present a consistent suite of archaeological, architectural and documentary evidence which indicates that this building is a multiphase early-14th-century structure that post-dates the surrounding L-shaped enclosure, as well as documentary evi-dence suggesting that a later 13th- to 14th-century transition from private to direct royal control of the site took place.

The settlement of Tarbert is situated on a low-lying isthmus which links the Kintyre peninsula to Knapdale and the rest of mainland Argyll, separating Loch Fyne and the Clyde estuary from the sound of Jura and the Atlantic Ocean (Illus 1). The Gaelic place-name Tairbeart and its Old Norse equivalent Saltíriseið are toponyms which describe this isthmus and, with sheltered anchorages situated on both sides of the land-bridge, various medieval and later accounts suggest this served as a portage across which boats could be dragged to avoid the long sail round Kintyre (Blaeu 1654: 58; OPS II.I: 32; MacLennan 1925; Dunbar & Duncan 1971: 1; Cheape 1984: 210–11; Kirkeby 1998: 268; McCullough 2000: 20–1).

A succession of Irish annals references to ‘The burning of Tairpert Boiter’ and ‘The burning of Tairpert Boiter by Dúngal’ suggest a power centre of some importance was located here by the early 8th century at least (AU: 712.2, 731.4; ESSH: 213, 227), and within a Scottish Atlantic discourse in which the importance of maritime transport is well established (McDonald 2017), the particular significance of the site is evident in two highly symbolic portage legends which frame a crucial period in Argyll’s later medieval history. The first of these is a Norse saga account describing an agreement struck between the crowns of Scotland and Norway in 1098, wherein the Norwegian:

And the second is found in a Scots poem recounting how Robert I made a similar portage in 1315, and thereby won the allegiance of the West Highlands and Islands chieftains:

Both of these legends are retrospective and of questionable veracity, but meaningful nonetheless. Magnús’s movements during his late-11th- to early-12th-century voyage can be followed through a wide range of broadly consistent documentary sources from England, Ireland, Man and Scotland, and the island/mainland division of territories described in the saga is evident in later administrative boundaries (Duncan & Brown 1957: 193–4; Power 1986). But 12th-century sources describing David I’s share of Kintyre’s revenues and the exclusion of the peninsula’s churches from the Sodor diocese suggest Scandinavian claims to control of Kintyre were apocryphal, or at best very short-lived, and it is probable the saga narrative more closely reflects heightened tensions surrounding control of Scotland’s western seaboard during the early-13th-century period in which it was written (Duncan & Brown 1957: 194; Woolf 2004: 101; Heimskringla I: ix). We might have greater confidence in the Bruce legend, since contemporary charters relating to lands in Argyll and Perthshire appear to confirm that Robert I was indeed at ‘Tarbert next to Loch Fyne’ in 1315 (RRS V: no. 69 reported in POMS, H1/53/77; HP II: 129–34),2 although Barbour’s patriotic late-14th-century narrative describing this visit was written in praise of the ruling Bruce–Stewart dynasty following a long period of political and military turmoil which had included the Wars of Independence (Boardman 1997). While the details of these events should thus be viewed with some caution, the opposing nationalist perspectives presented here are striking and clearly communicate that control of the Tarbert portage was regarded as a necessary requirement for claims to wider political authority across the region in this early part of the 2nd millennium AD (cf Dunbar & Duncan 1971: 2).

It is no surprise, therefore, that the fragmentary upstanding remains of a large medieval masonry castle survive in an elevated position above the south-east shore of East Loch Tarbert, with excellent views out to Loch Fyne and over the anchorage below. A castle in this position belonging to the Earls of Argyll was noted by Timothy Pont (c 1583–1614: 84), with stylised depictions then presented by Joan Blaeu (1654) in 17th-century maps of both Knapdale and Kintyre, before a more detailed plan of the now ruined complex was published by the Ordnance Survey in 1870. Examination of a remarkably detailed set of early-14th-century exchequer accounts had suggested to early scholars that the castle complex was built by Robert I in the decade following his 1315 visit (OPS II.I: 33; RSRS I: lxxi, 52–8), but investigation of the upstanding ruins by David MacGibbon and Thomas Ross (1887: 136–42) resulted in a plan drawing which was consistent with the OS map (Illus 2a & 2b), and a three-phase developmental scheme which suggested the castle had earlier origins.

Allowing for changes in nomenclature, subsequent interpretations of the surviving ruins have generally followed this three-phase scheme very closely. John Dunbar and Archie Duncan (1971: 7, fig 2) confirmed that each of the site’s ‘three distinct units … appears to be the outcome of a separate period of building activity’ (see Illus 2c) and a detailed description and collection of measured drawings published by the RCAHMS (1971: no. 316, 25, 179–84, fig 175) depicted the walls of the secondary ‘outer bailey’ as distinct from (effectively abutting) the sub-square ‘inner bailey’ (Illus 3). Indeed, focusing on the earliest phase of this scheme, Dunbar & Duncan (1971: 7–8) presented an ambitious typological approach to interpretation of the summit structure; situating this building within a group of early ‘simple rectangular castles of enclosure’ characterised by a ‘substantial curtain-wall of stone, pierced with few openings, enclosing an oblong or square courtyard around which there were ranged lean-to buildings of stone or timber’. This group included primary phase upstanding enclosures at Castle Sween (north Knapdale), Achanduin (Lismore), Duart (Mull), Innis Chonnell (Loch Awe), Roy (Speyside), Hume (Berwickshire), Kincardine (Mearns) and Kinclaven (Tayside) and, although acknowledging that ‘closely datable features are conspicuously absent’ from these buildings, Dunbar & Duncan (1971: 8, 14) suggested ‘they may tentatively be ascribed to the twelfth and early thirteenth centuries’. The relatively large size of ‘Alexander II’s castle’ on the summit of the Tarbert site, moreover, prompted closer comparison with masonry structures surviving at two well-documented eastern Lowland castle sites with strong royal associations – at Kinclaven and Kincardine.

Most recent work at the castle has been sponsored by local charitable trusts (the Tarbert and Skipness Community Trust and the Tarbert Castle Trust) and this has included architectural survey and consolidation of the tower house, laser and geophysical surveys of the wider site, and a community programme of photographic and measured survey directed by archaeologist Roddy Regan (2013). This latter study presented detailed descriptions of surviving upstanding features, and collated historical evidence which informed community excavations to the south-west around the possible location of a medieval royal burgh (Regan 2013; 2018; RCAHMS 1971: no. 334, 191), while preliminary characterisation of the masonry fabric on the main castle site (during a much wider regional survey) highlighted the radiocarbon analysis potential of the exposed mortar materials (Thacker 2016, II: 61). In the interim, however, previous interpretations of the structural development of those upstanding buildings remained unchallenged.

With the portage and its associated legends providing an important geohistorical context, the cultural significance of the surviving fabric at Tarbert Castle has long rested on a coherent narrative of direct control by the Scottish crown, suppression of the unruly West Highland and Hebridean clans, and a relatively early constructional date. Indeed, MacGibbon & Ross (1887: 136, 143) suggested that ‘The object of the castle from its first erection must have been to serve as an entrenched camp or stronghold for a large garrison on the edge of a country which might any day rise up in rebellion, and this explains its plan’, while Dunbar & Duncan’s (1971: 2–3, 14) emphasis on the early-13th-century documentation associated with Kinclaven and Kincardine allowed them to equate the summit structure at Tarbert with King Alexander II’s expedition to Argyll in 1221/2, as reported in the chronicle subsequently attributed to John of Fordun (Skene 1872: 284; Duncan & Brown 1957: 199). Ultimately, as one of ‘a well-defined group of early Scottish stone castles’ (RCAHMS 1975: 26–7) the inner bailey at Tarbert was then itself enlisted as comparative evidence for the interpretation of other masonry buildings across Argyll, and the building became an oft-cited physical expression of the early-13th-century ‘Scottish crown subjugates Argyll’ narrative (Duncan 1989: 528; Barrell 2000: 85; Barrow 2003: 109; Oram 2011; 2012: 79–80, 179, 189–90, 249).

The typological scheme within which Dunbar & Duncan situated the Tarbert inner bailey includes some of the earliest lime-bonded secular buildings surviving in Scotland, but recent archaeological work suggests the wider group was constructed over a longer period than was previously (albeit tentatively) proposed. Indeed, while recent programmes of materials analysis have confirmed that the enclosure walls at Roy and Kincardine were probably raised in the late 12th to early 13th century (Thacker 2020a; 2020b; 2021a), a combination of excavated numismatic evidence, contemporary charters and materials analysis has suggested Achanduin Castle on Lismore was built around a century later – in a very narrow late-13th- to early-14th-century period (Caldwell et al 2015; Caldwell & Stell 2017; Thacker 2020c; 2021a). This is unsurprising, since pre-existing buildings of any age can serve as archetypes for new construction projects, and typological meaning largely inheres in establishing why similar structures may or may not have been built in different periods, but this evidence does suggest that the more chronologically precise interpretations previously ascribed to the Tarbert inner bailey cannot be accepted on the architectural form of the building alone.

From this perspective it should also be recognised from the outset that the early-14th-century exchequer accounts noted by previous investigators are the earliest contemporary sources describing a castle complex at Tarbert and there is no evidence the site was held directly by the Scottish crown before the reign of Robert I. The issuing of charters from Tarbert might imply that some kind of administrative structure was located here in 1315 (see above), but the site does not appear in the catalogue of royal castles handed over to Edward I and The Competitors by their Keepers in 1291 (Stones & Simpson 1978: 100–1; Watson 1991: 28) or in the list of 23 sites passed on to King John the following year (RSRS I: xlix), and the 1292 or later crown source referred to by MacGibbon & Ross (see above) could not be located by this author. Absence from these crown records was regarded by Stewart Cruden (1960: 69) as probable evidence that the castle was either not yet constructed or in pro-English hands, and this lack of late-13th-century documentary evidence for a royal association is implicit in the RCAHMS (1971: 25, 182) suggestion that the inner bailey may have been erected by a dependant of the MacDonalds of Islay before Robert I ‘is known to have taken possession of the castle and to have strengthened it to serve as a seat of royal authority in the area’ (see also Walker 2000: 473). In the absence of direct evidence indicating that the castle pre-existed Robert’s reign, however, it is notable that much of the historical evidence cited by Dunbar & Duncan in support of their early-13th-century interpretation of the inner bailey is also negative: arguing that the castle was not mentioned at all during Norse expeditions in the Clyde estuary in 1230 and 1263 because the portage was so ‘firmly held in Scottish hands’; not recognised in the 1296 correspondence between Alexander of Islay and Edward I regarding Kintyre because the site was situated in neighbouring Knapdale; while the lack of references during the 1296–1314 Wars of Independence was ‘suggestive of neglect or the passing of the castle into private hands’ (Dunbar & Duncan 1971: 7, 13).

This historiographical summary highlights how interpretations of Tarbert Castle have shifted emphasis progressively over the course of the 19th and 20th centuries; from a dependence on early-14th-century documentary sources, to a dependence on the upstanding architectural form of the fragmentary inner bailey summit structure. This process resulted in the surviving castle being ascribed to increasingly early dates and ultimately allowed late-20th-century scholars to situate the building within an early-13th-century historical narrative appropriate to the site’s perceived strategic importance and royal status. This discourse parallels that surrounding Urquhart Castle on Loch Ness: a building suggested to have performed a militaristic peace-keeping role in Alexander II’s reign following the last of the Moray ‘uprisings’ in 1229 (Simpson 1929: 54; 1964), and morphological similarities between the upstanding fabric on both sites are notable.3

Before possible comparators are considered, however, it is important to acknowledge two pieces of documentary evidence which have been largely ignored during previous investigations of the Tarbert site, but which appear to relate directly to administration of the surrounding lordship and the castle’s construction in this crucial 13th- to 14th-century period. The first of these is a c 1250 charter most recently highlighted by David Sellar (1971; 2017), in which a ‘Douenaldus Makgilcriste dominus de Tarbard’ granted ‘free right of cutting, taking, and carrying away all kinds of timber … for the building and maintaining of their monastery and house of Paisley’ (Reg. Pass.: 157; Mitchell 1886: 9–10; appendix 1). And the second is a caveat in the 1326 exchequer roll, which included payments made for various significant structural elements including ‘building the walls of the castle of Tarbart … building the said castle … building a new “pele” at Wester Tarbart … [and] … the making of a moat’, but did not include payments made for ‘the houses within the inner court (clausura) [or] the middle wall enclosing it … for which the constable had not leisure to account’ (OPS II.I: 35 translated from RSRS I: 584). Indeed, while the first of these documents suggests the lordship of Tarbert was in ‘private hands’ by the mid-13th century at least, this latter statement suggests that construction of the castle’s summit structure was a very recent early-14th-century event.

This introduction clearly suggests that the development of the buildings surviving at Tarbert Castle and the site’s role in the later medieval geopolitics of Argyll require some re-evaluation. To this end, the current paper describes work undertaken to establish relative and independent chronologies for the masonry fabric surviving at Tarbert Castle, through buildings and materials analysis. Ultimately, this study would demonstrate that the development of the castle complex has previously been misunderstood, and that physical evidence will be set out below before discussion of its wider implications.

A programme of buildings analysis and materials sampling was undertaken at Tarbert Castle in June and August of 2019 under the aegis of the Scottish Medieval Castles and Chapels C14 Project (SMCCCP). The first of these periods was scheduled to coincide with further community excavations at the south-west end of the site (Thacker 2019a; 2020d; Regan 2019a; 2019b), during which this author sought to characterise the surviving masonry and investigate how the excavated fabric related to the upstanding evidence. This process would inform a sampling strategy: targeting masonry materials considered representative of particular phases of development, to enable lab-based microscopic investigation of material sources and radiocarbon analysis of selected mortar-entrapped relict limekiln fuel (MERLF) fragments (Thacker 2020a). In response to this first round, a second visit was arranged to remove further samples from an area of upstanding masonry considered particularly significant (Thacker & Regan 2019).

Although this investigation was focused on the south-west of the site, it is important to recognise that the wider curtain-walled enclosure is very large, and the ground surface is unusually irregular. Indeed, even the small courtyard on the site summit contains a large irregular rocky outcrop which was never quarried away (MacGibbon & Ross 1887: 137). The three main structures recognised by previous visitors also present striking contrasts in upstanding survival: from the four-storey tower house (which survives to wall-head height); to the inner bailey walls and particularly the cross-wall (which survives to over 2.5m high); to the outer bailey walls which (with certain exceptions) are often reduced to low turf-covered wall-lines only. These contrasts are clearly visible on aerial photographs of the site (see Illus 4) and will be noted at particular locations below. The nomenclature used to identify these structures will follow that adopted by the RCAHMS (1971), feature locations are marked on a modified plan drawing (Illus 5), and a more comprehensive photographic record is presented as electronic supplementary material (ESM).

Between the south angle of the inner bailey and the tower house, the south-east curtain wall at Tarbert Castle is approximately 2m wide and appears to be continuous along the south-east sides of both enclosures (Illus 5 & 6; ESM Illus S1–S6). Upstanding survival of the external wall face is limited, although a relatively complete area bounding the inner bailey (measuring 20m long × 1m above current ground levels) presents a single vertical face without visible architectural details. These degraded remains suggest the wall face was informally constructed, presenting a poorly sorted and poorly bonded mixture of irregular rubble blocks with faces up to 700 × 250mm. Higher ground levels within the inner bailey have placed even greater limits on visibility, but a 5m-long stretch of the internal wall face was exposed by excavation in the south corner of the outer bailey, revealing a 200mm-deep scarcement intake which divides the vertical internal wall face (here surviving up to 1.6m high) from a roughly battered foundation footing (of approximately 450mm high). These wall faces have also been constructed from a mixture of undressed schistose rubble blocks (including slabs with faces up to 800 × 200mm) laid in informal level courses. The wall retains extensive lime mortar evidence, although dissolution has reduced this to such a friable condition that it was challenging to identify samples coherent enough to be safely removed, and internal (possibly partial) mortar coating evidence was fragmentary. This mortar appeared compositionally consistent throughout these exposed areas of the wall, however, and could be confidently characterised in situ as a wood-fired limestone-lime, tempered with a reasonably fine foreshore aggregate (Thacker 2020d). The internal wall face is separated from the adjacent cross-wall by a straight joint (labelled Straight Joint 1 in Illus 5), which is clearly visible in all upstanding and excavated contexts and indicates the cross-wall is secondary to (abutting) the internal face of the south-west curtain wall at this location.

Stretching from a rounded west angle to a right-angled corner with the outer bailey south-west wall, the fragmentary remains of the inner bailey north-west curtain wall is also approximately 2m wide (Illus 5 & 7; ESM Illus S7–S11). This feature is generally visible as low turf-covered footings only. The loss of the external wall face has exposed some limited volumes of lime-bonded core rubble, however, and a 1.2m+ stretch of internal wall face composed of well-bonded schistose slabs survives at its north-east end behind Straight Joint 2 (Illus 5). Although the two are not visibly continuous, this wall face fragment aligns with the much more fragmentary internal face of the inner bailey wall above. As at Straight Joint 1, the relationship between this internal wall face and the adjacent cross-wall was visible in the upstanding masonry, and the superficial vegetation covering the fabric behind Straight Joint 2 was removed to allow further investigation (Thacker & Regan 2019). This internal wall face is associated with limited volumes of a very friable lime mortar surviving to approximately 200mm deep, and this could be confidently characterised in situ as a wood-fired limestone-lime tempered with a poorly sorted fine mixture of rounded foreshore gravels (Thacker 2020d). A series of undressed rough rubble quoins form the fragmentary remains of a much-robbed right-angled corner by which this internal wall face was bonded to the adjacent outer bailey south-west curtain wall, and this angle is abutted by the adjacent cross-wall at Straight Joint 2 (see below).

The outer bailey south-west curtain wall rises from the remains of the west angle-tower to the right-angled corner with the inner bailey north-west curtain wall described above (Illus 5; ESM Illus S12–S15). The upstanding remains of this wall are generally limited to low turf-covered footings and core rubble and the corner with the inner bailey north-west curtain wall displays missing quoins (Illus 7). Excavation across this wall, between Straight Joint 2 and a protruding mid-tower, exposed the fragmentary remains of a substantial (3m-wide) gate, which displayed red-coloured dressed sandstone external reveals (square to the external wall face) with chamfered external arrises, a probable portcullis slot, and doorframe rebates (Regan 2019a; Illus 8; ESM Illus S12 & S13). Internally, these reveals are rubble-built and splay to 3.5m wide, constructed from large roughly squared blocks and narrower slabs laid to formal courses, and extending past the line of the internal face of the curtain wall on the south-east side. A moderate concentration of red sandstone spalls was also visible in the core rubble close to the north-west reveal, where extensive volumes of constructional mortar in full face–core–face cross-sections has also survived. This material appeared compositionally consistent and can be characterised in situ as a wood-fired limestone-lime mortar tempered with a poorly sorted/bimodal foreshore aggregate grading to 15mm (Thacker 2020d).

The external south-west face of the west tower was revealed by excavation for a width of 2.1m from a rubble-built south angle. Surviving to 1.9m high and constructed to a batter of approximately 1:10, this wall face has been tightly built in well-bonded rough courses of narrow schistose slabs grading up to 1.15 × 0.21m (ESM Illus S14 & S15). No protruding foundation plinth is apparent at the base of this wall, although the adjacent external wall face does display a low battered plinth. The internal wall faces of this structure have been neatly constructed using smaller mica-schist slabs, with the remains of a doorway in the north-east wall framed by dressed red sandstone blocks with chamfered external arrises. The constructional mortars surviving in this building present a range of dissolute textures, but appear compositionally consistent throughout, and can be confidently characterised in situ as a wood-fired limestone-lime tempered with foreshore aggregates.

Stretching between the north-west and south-east walls of the inner bailey, a cross-wall approximately 42m long divides the inner and outer baileys, with a gate approximately mid-way along its length (Illus 5; ESM Illus S16–S21). Viewed from the outer bailey, the upstanding masonry of this wall is dominated by considerable volumes of exposed lime-bonded core rubble, standing over 2.5m high in the north-west with some small but significant areas of wall facing (Illus 9). At its north-west end, the outer face of this feature displays an upstanding area of wall face measuring 2.7 × 1.0m, comprised of a bimodal mix of angular blocks with large planar faces up to 500 × 250mm, laid in a relatively formal masonry style with some evidence for 300mm-high courses. This masonry is also associated with continuous and coherent mortar evidence up to 600mm from the wall face, and some fragmentary mortar coating evidence, which can be confidently characterised in situ as a wood-fired limestone-lime, tempered with a poorly sorted aggregate mixture grading to 6mm (Thacker 2020d). Presenting different course heights but no quoins, this cross-wall masonry (face blocks, core rubble, and constructional mortar) clearly abuts the internal face of the adjacent inner bailey north-west curtain wall at Straight Joint 2 (Illus 5 & 7), and the contrast in wall face survival and mortar coherence (see above) is notable.

Examination of the upstanding and excavated masonry reveals that this cross-wall also abuts the internal face of the south-east curtain wall at Straight Joint 1 (Illus 5), forming an obtuse angle of approximately 100° at the south corner of the outer bailey. Excavation had exposed a 5 × 2.3m area of this cross-wall, revealing an offset scarcement (here 250mm deep) separating the main vertical wall face from footings 250mm high, which appear to have been bedded on organic soil. In a striking contrast with the adjacent south-east curtain wall, a large area of mortar coating survives on this excavated cross-wall face (obscuring its masonry style), and observations made during sampling suggest this coating is continuous with mortar visible in masonry beds to at least 50mm deep. With high concentrations of included limekiln relicts, this material can be confidently characterised in situ as a wood-fired limestone-lime, tempered with a bimodal mixture of foreshore aggregates grading to 16mm (Thacker 2020d). Direct comparison indicates this material is of similar superficial character to the mortar associated with the adjacent south-east curtain wall, although once again the cross-wall mortar is much more coherent and here appears to be included with higher concentration of marine shell inclusions. With no evidence for quoining, both faces of this cross-wall and its associated mortar coating clearly abut the adjacent south-east curtain wall at Straight Joint 1 and overbuild its scarcement intake.

Excavation revealed internal wall faces on three sides of the inner bailey north-east range to the south-east of the cross-wall gate (Illus 5; ESM Illus S22 & S23; Regan 2019a). The walls forming the south-west and north-west sides of this cell are of similar character and are bonded to one another with alternate long stone slabs – clearly indicating contemporaneity. The relationship between these features and the adjacent cross-wall at the north angle was obscured by the remains of an oven-like masonry feature, and examination of the turf-covered external faces did not clarify this latter relationship any further, although the chamfered red sandstone block defining the gate to this summit structure was noted.

Where examined most comprehensively (in the south-west wall) the mortar of this north-east range is visible in continuous bed-core contexts, while excavation also revealed a very small area of mortar floor. Despite some superficial variation in colour and texture attributable to taphonomy/dissolution, this material appears compositionally consistent and single phase. In situ characterisation determined that this was a wood-fired limestone-lime, tempered with a bimodal to poorly sorted mix of foreshore aggregates, dominated by submillimetric materials but including larger gravels, stones and marine shells to 30mm+ (Thacker 2020d).

Informed by this analysis of the surviving masonry, an assemblage of material samples was removed from upstanding and excavated contexts to enable lab-based investigation. Hand-measured in three planes relative to adjacent wall faces and datum string lines, sample locations were recorded photographically and plotted onto a plan drawing of the site (Illus 10; ESM Illus S24–S35). A label was also pinned to the masonry to allow sample locations to be identified on the excavation drawings.

Analysis of the wider sample assemblage remains at a preliminary stage, but microscopic examination of the MERLF assemblage indicated this was composed of four fragments of Quercus sp (40%), four fragments of Betula sp (40%) and two fragments of Corylus sp (20%) charcoal (Table 1). Unusually for an SMCCCP study, an initial round of radiocarbon analysis was hampered by lack of carbon in selected samples, but fragments removed from the cross-wall and outer bailey south-west gate have returned determinations with δ13C values reasonably close to the expected −25‰, and these both calibrate to similar 13th-century date ranges (Illus 11 & 12; Table 2).

Analysis of the upstanding and buried masonry at the south-west end of Tarbert Castle for the present project has revealed that the cross-wall dividing the inner and outer baileys is an inserted feature, secondary to curtain walls of similar character on both sides. A fragmentary right-angled return was recognised in one of these earlier walls and, although close to a gateway which is more complex than currently understood, this suggests the L-shaped configuration of the castle enclosure was conceived of from its inception (Illus 5–7; ESM Illus S1 & S2, S7–S9). This corner feature and the south angle of the west tower have been constructed from undressed schistose rubble blocks similar to those used in the general wall faces, and evidence for dressed freestone in these earlier phases is currently limited to the chamfered red sandstone blocks framing the cross-wall, outer bailey and west tower gates and doorway. Confirmation of the provenance of these materials must await further analysis, but similarities with the metamorphic outcrops visible around the site suggests the schistose rubble blocks were quarried locally, and a source from nearby Arran is suspected for all sandstone materials. In situ examination also suggests that all masonry features were bonded with lime mortars manufactured from a wood-fired meta-limestone and tempered with foreshore sand/shell/gravel aggregates. These source materials are also consistent with the coastal character of the site and the Dalradian metamorphic geology of the wider locality, which includes a Loch Tay limestone previously exploited for lime manufacture (Roberts 1977; Stephenson & Gould 1995; Thacker 2020e). This wood-fired interpretation has been confirmed through microscopic sample analysis, and the Quercus/Betula/Corylus composition of the examined assemblage is remarkably consistent with the predominantly oak/birch/hazel character of local semi-natural woodland populations (Ratcliffe 1977; Birks 1993; Tipping 1994).

This fuel assemblage is too small to allow statistical significance (and independence between closely located samples cannot always be demonstrated) but it is notable that those samples removed from the cross-wall and inner bailey north-east range wall were dominated by birch (four birch, two oak), while those from the curtain walls (two oak, one hazel) and south-west gate (one hazel) included none. Other more general contrasts in material character were noted in situ, including increased coherence and survival of cross-wall mortar and masonry (compared with the adjacent curtain wall fabric) and varying marine shell temper concentrations. This comparative evidence is striking and appears consistent with the stratigraphic relationships on display – perhaps suggesting that a significant time period separated these constructional events. But this interval cannot be easily quantified and identification of a thick mortar coating on the external face of the outer bailey north-east curtain wall during previous mortar survey (for Thacker 2016) highlights that such contrasts are often exaggerated by taphonomical/environmental pressures and/or contextually specific constructional practices. Indeed, close examination failed to clearly identify surviving evidence for an early internal mortar coating fossilised within either of the highlighted straight joints.

MERLF fragments from the cross-wall and outer bailey south-west gate have returned radiocarbon determinations which calibrate to very similar 13th-century date ranges, at both 68% and 95% confidence. With only single determinations available it is not yet possible to situate these data within a comparative model which would allow more precise chronological constructional estimates, and the potential for residuality in the samples precludes using the cross-wall determination as a terminus ante quem (TAQ) for the adjacent pre-existing curtain walls. Each of these calibrated dates should be regarded as a terminus post quem (TPQ), which usefully indicates that both features were constructed in the later medieval period although, since these determinations were returned by relatively short lifespan tree taxa, a 13th-century or early-14th-century date for both seems most likely.

The quadrangular courtyarded structure which dominates the late-19th-century plan of Tarbert Castle published by the Ordnance Survey (1870) reflects the increased height of the turf-covered masonry fabric on the site summit (Illus 2a & 4). With similar images presented in later developmental schemes, this standalone interpretation appears to have hardened up as the interpretive balance shifted away from the documentary evidence, and earlier constructional estimates were imposed on this inner bailey building (above; see Illus 2). These interpretations were effectively led by an architectural-historical typology and, although the straight joints at each end of the cross-wall are evident on the RCAHMS plan drawing (1971: fig 175; Illus 3), their conjectural interpretations of the relationships between adjacent features are not consistent with the evidence now visible.

The radiocarbon and architectural evidence associated with the outer bailey south-west gate and north-east drum-towers presents a consistent late-13th- to early-14th-century picture, and a similar red sandstone framed gateway in the north-east curtain wall has now also been revealed by excavation (Regan 2020; see Illus 5). Both gates present issues of interpretation regarding portcullis winding mechanisms which suggest further complexity pertains and the continuation of the south-east reveal of the south-west gate past the internal curtain wall face is salient. But the 3m width of this latter opening, which faces the burgh and isthmus, is similar to the arch-headed portcullissed opening surviving in the gatehouse at Urquhart (Simpson 1964) and this south-west gate probably represents the main castle entrance. Indeed, the north-east drum-towers are small salient circular features separated by over 10m of curtain wall and, with access from the foreshore requiring negotiation of a steep incline, it is probable that the sandstone framed opening situated between these towers represents a postern gate. This tower/gate configuration has similarities with the postern opening at Coull Castle (Aberdeenshire); an irregular curtain-walled enclosure with round angle-towers and a drum-towered gateway previously ascribed to the early 13th century (Simpson 1924). But elsewhere in Scotland these features are more generally ascribed to a later period, and in the wider Firth of Clyde region this includes the probable late-13th-century phases at Rothesay (Bute), Dundonald (Ayrshire) and Brodick (Arran) (Cruden 1960: 72–80; Dunbar 1966: 31; Dunbar 1981: 47; Ewart et al 2004; Anderson & Dixon 2011; Dixon et al 2015; Thacker 2019b; Illus 1).

The portcullis evidence at Rothesay follows a similar pattern of interpretation, with ascriptions varying between pre-1220 and more probable post-1230 dates (see Cruden 1960: 29–30; Tabraham 2002: 16), although all agree that (as at Lochindorb) this feature is associated with a secondary phase of construction and the primary entrance was draw-barred only. Indeed, this developmental scheme is also likely to have prevailed at the nearby castle of Skipness, where an early domestic block and chapel (see below) were absorbed within an impressively formal sub-rectangular curtain-walled enclosure, displaying quadrangular angle-towers and a portcullissed gatehouse framed with red sandstone blocks (Graham & Collingwood 1923; Cruden 1960: 55; RCAHMS 1971: no. 124). The RCAHMS (1971: 178) erred towards a slightly later turn-of-the-century date for this programme of remodelling than was previously proposed (Graham & Collingwood 1923), and once again suggested that an association with the MacDonalds of Islay and Kintyre might pertain, but Richard Oram’s (2017: 256–7) interpretation that ‘the late-13th-century drum-towered gatehouse at Brodick and high-quality mason-work at Skipness underscore the parallel entrenchment of regional domination by the Stewarts and their cadets’ is a more convincing attribution. Stewart patronage of the Tarbert drum-towers and gates cannot be ruled out, therefore, although the surviving fabric does not readily conform to the architectural formality presented by these more upstanding buildings. Indeed, the lack of spreading bases prompts comparison with the round-towered enclosure at Caisteal Dubh, Pitlochry (Perthshire); a building which has been compared to the late-13th-century castle at Inverlochy but is also associated with Campbell of Lochawe’s accession to the Atholl earldom in c 1326 (MacGibbon & Ross 1889: 109–10; Cruden 1960: 61–2).

In the current absence of evidence indicating these gate and tower structures were inserted features (cf RCAHMS 1971; cf Dunbar & Duncan 1971), construction of the large L-shaped enclosure at Tarbert can also be cautiously ascribed to this same late-13th-century or later period. Displaying a secondary relationship with this enclosure which now clearly separates it from the other buildings included in the Dunbar & Duncan (1971) scheme, and drawing further attention to the lack of projecting angle buttresses or towers5 currently associated with this structure, this model would suggest that the inner bailey masonry was also raised in the late 13th century or later. Indeed, an early-14th-century interpretation for the inner bailey building ranges and cross-wall appears consistent with the radiocarbon evidence discussed above and with the character of the occupational evidence revealed from within the north-east range by excavation; and these secondary structures very probably represent ‘the houses within the inner court (clausura) [and] the middle wall enclosing it’ highlighted in our introduction which constable de Lany’s report suggests were constructed in 1325/6 (OPS II.I: 35; RSRS I: 58; Regan 2019a; see also above).

The MacGillechrist grant to the monastery at Paisley, also highlighted in our introduction, appears to represent the only surviving contemporary 13th-century reference to political administration of the Tarbert lordship. Although undated (Reg. Pass.: 157; Appendix 1), this account has been ascribed to c 1250 by both Dugald Mitchell (1886: 9) and David Sellar (1971: 29; 2017: 38), and recognition that the first witness is one ‘domino Hugone de Parcliner perpetuo vicario de Kylmacolme’ (Appendix 1) allows us to associate this document with the Loch Fyne ‘Tarbard’ with greater confidence.6 The inclusion of the northern part of this Kintyre parish in the ‘lordship of Knapdale’ was recognised by Cosmo Innes (OPS II.I: 37 n12), who contrasted this administrative unit with the ‘land of Knappedoll’ north of the Tarbert isthmus granted to the MacDonalds of Islay in 1335 (Rot. Scot. I: 463), while Dunbar & Duncan (1971) supported their contention that the more southerly lordship boundary was in place by 1261 with reference to a description included in MacFarlane’s 17th-century Geographical Collections (OPS II.I: 32). This latter source claimed no earlier date for this arrangement than the MacDonalds’ later medieval forfeiture, but it is reasonable to suggest that this contrast in boundaries reflects a process of secular negotiation which post-dated the fossilisation of these parish boundaries – an event which itself probably took place during a 12th- to early-13th-century period when control of the wider region was dominated by Somerled, his powerful son Ranald (‘rex insularum, dominus de Ergile et Kyntyre’; Sellar 2000: 195) and subsequent MacSorley descendants. Indeed, the location of Saddell Abbey suggests that Ranald was based somewhere in Kintyre (Woolf 2004: 105) and, although his main caput site is yet to be identified, Tarbert must be considered amongst possible locations.

David Sellar’s (1971; 2017) suggestion that the MacGillechrist lord who held Tarbert in the mid-13th century was likely to be a member of the Donnsléibhe kindred suggests that further interpretation of the region’s geopolitical development is possible. Preferring the 17th-century genealogy of the O’Clery Book of Genealogies (OCG),7 Sellar (1971: 27) recognised that three of Donnsléibhe’s sons – Suibhne, Ferchur and Gilcrist – were ancestors of significant landowning clans in the Loch Fyne region and highlighted that the first of these can be identified with the ‘Swineruo’ described in a 17th-century narrative account of McIver-Campbell history as ‘Thane of Knapdaill and Glasserie’ (HP II: 82). Contemporary documents describing the territories administered by subsequent generations of the MacSweens and MacGillechrists are consistent with this title and, although these clans also held estates in neighbouring Cowal during the period, these Knapdale and Glassary lordships are also strongly associated with a network of 13th-century castles (Illus 13). This includes Castle Sween itself, a building whose generally accepted c 1200 constructional date is consistent with a traditional 16th-century account (‘Leabhar Chlainne Suibhne’) reporting that Suibhne was the patron (Sellar 1971: 27), with various Romanesque architectural details associated with the primary enclosure (RCAHMS 1992: no. 119, 245–59), with excavated numismatic evidence from a secondary phase of occupation (Ewart & Triscott 1996: 521, 526; Holmes 1996), and with recently published radiocarbon data associated with MERLF fragments removed from the upstanding walls (Thacker 2020e). To the north-east, a very precise mid-13th-century constructional estimate has been generated for Castle Fincharn, again from MERLF radiocarbon data, suggesting that this masonry block can be associated with a 1240 crown charter effectively granting (or more probably confirming) the lordship of Glassary to ‘Gillascop mac Gilcrist’ (HP II: 121–3; RCAHMS 1992: no. 130, 283–6; Thacker 2017; 2020a; 2021a). On the southern boundary we can cite the well-known c 1261 charter granted to the Paisley monastery by ‘Dufgallum filium Syfyn’, which describes his ‘castrum … de Schyphinche’ (Reg. Pass.: 120–1; White 1873: 182–3) and suggests that the masonry block and chapel absorbed into the later enclosure at Skipness (see above) had already been constructed by this date (Graham & Collingwood 1923; RCAHMS 1971: no. 314, 165–78), while a 1247 grant to the parish church at Kilcolmanell suggests that the lordship had been held by this same Dubhgall MacSween for some time (Dunbar & Duncan 1971: 6). Other seigneurial sites associated with the kindred include Castle Lachlan in neighbouring Cowal, which was the probable site of a castle building included in a 1314 documentary source, associated with a grandson of Gilcrist known as Lachlann Mòr (RCAHMS 1992: no. 118, 237–45; Sellar 2017: 35). And while the 13th-century ascription previously associated with the small upstanding curtain-walled enclosure at ‘Ardechaisteol’ Duntrune (MacGibbon & Ross 1889) has been challenged by the RCAHMS (1992: no. 128, 276–82), the position of the ‘terra Johanni McGilcrist’ in the list of lordships included in Baliol’s 1293 Sheriffdom of Lorn suggests that the surrounding lordship of Ardscotnish (Kilmartin) was held by Gilleasbuig MacGilchrist’s brother Ewen (Eugenius) earlier in the 13th century (NAS Roll of Parliament PS1/2, 1293/2/17 in RPS; HP II: 121 n2). Sellar’s identification of ‘Douenaldus Makgilcriste dominus de Tarbard’ (Reg. Pass.: 157) as a contemporary of these MacSween and MacGillechrist magnates, therefore, further consolidates this remarkably consistent documentary and architectural record, and completes the kindred’s stranglehold on secular administration in Knapdale and Glassary down to the mid-13th century at least (Illus 13).

This confluence of architectural and historical evidence represents a process of geopolitical consolidation and growth in which the kin-network’s core Knapdale and Glassary territorial block was subdivided into smaller secular administrative units which are generally co-extensive with parish boundaries (Thacker 2016, I: 146–8). The early date associated with the primary enclosure at Castle Sween suggests that these important developments were already underway by c 1200, but the Sween/Gilchrist clans also clearly retained extensive holdings following Alexander II’s 1221/2 reorganisation of the region’s administration (Duncan & Brown 1957: 199–200; Duncan 1989: 528) and their holdings may have extended into northern Kintyre in this same period at the MacSorleys’ expense (cf Cruden 1960: 55). Whether a member of this kin-network also administered royal power more directly, in line with the later-13th-century bailie of Loch Awe and Ardscotnish recognised by Steve Boardman (2006: 20), is unknown.8 But the charter indicating that Gilleasbuig MacGillechrist held his Glassary settlements direct from the crown in 1240 clearly indicates that he had accepted the Scottish king’s authority by this period (Barrow 2003: 139; Oram 2012: 179) and a similar arrangement might well have pertained at Tarbert. Ultimately, these sources suggest that the Donnsléibhe kin-network was more powerful during Alexander II’s reign than is often acknowledged, and Tarbert was held by a more devolved administration during this early-13th-century period than has been previously recognised.

Against this evidence suggesting the Donnsléibhe kindred reached their floruit in this early-13th-century period, the rapidity with which the MacSween chieftains lost control of their lands once Alexander III reached maturity in c 1260/1 is striking and highlights an aggressive shift in later 13th-century crown policy towards the region (Duncan & Brown 1957: 212–13; Duncan 1989: 576; Barrow 2003: 141). As is well known, the main beneficiaries of this process were the Stewarts, who were granted extensive lands to the south of the river Clyde in the 12th century, acquired Bute by 1200, and made extensive territorial gains across Cowal and Arran then Knapdale in the later 13th century (Barrow 1980: 68; Boardman 2007). In 1262 Dubhgall MacSween granted Skipness, Killislate (south Knapdale) and other lands to Walter Stewart Earl of Menteith, who then confirmed the 1261 MacSween grant to Paisley (Reg. Pass.: 121–2; Sellar 2017: 36), and a grant of Glassary lands to the same monastery was witnessed by Walter Stewart and Malcolm son of the Earl of Lennox in this same year (Reg. Pass.: 132; Cowan 1990: 122). That the MacSweens were already making grants to Paisley suggests they had already been drawn into the Stewart orbit before the mid-13th-century, since this Cluniac house had been founded by this incoming family as early as 1163 (Reg. Pass: iii), and on the face of it these charters present an orderly succession – with the 1261 MacSween grant witnessed by Walter Stewart, and the 1262 Stewart confirmation in turn witnessed by Dubhgall and Murchadh ‘filio Malcmur’ MacSween (Reg. Pass.: 120–2). But the list of lordships presented in Balliol’s 1293 Sheriffdom of Lorn confirms that only Ardscotnish (with the Cowal lordship of Strathlachlan) had been retained by the wider kindred from the previously coherent Knapdale/Glassary supra-lordship block by the last decade of the century, and the inclusion of the ‘terra comitis de Meneteth de Knapedal’ within this administration confirms that the Stewarts of Menteith also now held the wider lordship, while ‘James the Steward’ was awarded the sheriffdom of Kintyre the following day (RPS: 1293/2/17, 1293/2/18). Indeed, another Walter Stewart would reportedly accompany his father-in-law Robert I on his legendary portage across the Tarbert isthmus in 1315 and, although he doesn’t appear as witness to the charters issued at this time (RRS V: no. 69 reported in POMS: H1/53/77; HP II: 129–34), Michael Brown (2004: 264) has suggested that he may have been confirmed as ‘lord of Kintyre’ during this visit. As in Glassary (which Master Ralph of Dundee now held), regime change in Knapdale had favoured this more eastern Scottish lordship and, although it is possible that Tarbert had already been annexed by Alexander III, Menteith or wider Stewart control would be more consistent with the crown’s long-favoured 13th-century policy of devolving administration in the north and west to powerful regional magnates.

Following a state-sanctioned attack on the Norwegian-held Isle of Skye by the Earl of Ross (Barrow 2003: 141), it is no coincidence that Hakon’s fleet arrived off the Argyll coast in this period of rapid late-13th-century administrative change. Our understanding of what this episode reveals about Tarbert, however, has been somewhat obscured by late-20th-century interpretations that the ‘lord of Kintyre’ who joined Hakon’s fleet with Angus Mòr of Islay was a brother by the name of Murchadh MacDonald (Duncan & Brown 1957; Cowan 1990: 119), and more recent scholarly consensus identifying this high-status chieftain as Murchadh MacSween (Sellar 1971: 21; 2000: 206; Barrow 2003: 142) demands some important changes of perspective on this narrative. Indeed, the recent dispossessions suffered by MacSween’s kin-network not only lend a more emotive edge to Edward Cowan’s (1990: 119) suggestion that the attacks on the earldoms of Lennox and Menteith were part of a deliberate strategy targeting the estates of the Stewarts and their allies, but also provides a more convincing context for the earlier harrying of Tarbert by a similarly sized 50-boat sub-fleet. Documentary evidence indicating that the neighbouring estates of Skipness and Killislate had been taken into Stewart control in the previous year (see above) raises questions regarding the identity of those under attack, although, frustratingly for current scholarship, Hakon reportedly recalled this task force before they reached the main Tarbert settlements (ESSH: 619–20; Johnstone 1882: 33). It may not have been possible for Hakon to drag his specially built oak ship across to Loch Fyne,9 but he must have understood the symbolism of this retreat, and that the sub-fleet did not recreate King Magnús’s legendary portage is inconsistent with suggestions that it was held by Angus Mòr (see above). Unlike the Loch Lomond expedition, however, in this case there was an alternative route available and, accepting the capture of the recently re-fortified royal castle at Dunaverty10 and raids in Caithness, Kintyre, and Lennox, Hakon does appear to have kept his military and diplomatic focus on regaining control of his island estates. This included granting (or re-granting) the Isle of Arran to MacSween, although it is unlikely that this was at all meaningful, and the death of Murchadh mac Suibhne is reported in the annals only four years later, having been captured off Ireland and held in prison by the Earl of Ulster until he expired from starvation (ALC: 1267.2; Nicholls 2007: 92; Simms 2007: 107; McWhannell 2014a: 4).

We should not assume that the wider Donnsléibhe kindred operated a politically coherent administration in this period, and whether the MacGillechrists retained control of Tarbert and Glassary as well as Ardscotnish throughout these events is unknown. Moreover, with no buildings yet discovered beneath the late-13th- to early-14th-century outer bailey features described above, the architectural character and location of the complex inherited by later administrators of the lordship is also currently obscure. It is possible that some masonry fabric from this period survives in the curtain walls surrounding the large L-shaped enclosure, and this might be supported by the contrasts in fabric survival noted above, but there is a dearth of comparative upstanding masonry of similar architectural form in the wider region which can be confidently ascribed to the early 13th century. Indeed, within a discourse focused on the importance of maritime transport during tensions between the Scottish and Norwegian crowns, we might expect the four castles held by Ewan MacDougall from King Hakon in 1249 to present particularly relevant comparanda (EHHS: 556; Caldwell & Ruckley 2005: 98), but only three of these sites can be identified with any confidence11 and of these only Dùn Chonaill is currently associated with upstanding curtain wall fabric ascribed to the medieval period (RCAHMS 1984: no. 402, 265–8). These fortresses are generally situated on naturally defensible islands where surrounding sea-cliffs have been augmented by masonry only where strictly necessary, and a similar situation pertains at the very well-documented royal castle of Dunaverty (Kintyre), which is now largely reduced to the fragmentary remains of a lime-bonded masonry curtain wall that may only ‘tentatively be ascribed to the medieval period’ (RCAHMS 1971: no. 309, 157). To the north of Glassary, the large McIver-Campbell enclosure at Dùn an Garbh-sroine (Craignish) presents closer morphological parallels with Tarbert,12 including a main entrance situated close to a re-entrant angle in one of the curtain walls and a large summit area of ‘irregular undulating character’ (RCAHMS 1992: no. 124, 263; Campbell & Sandeman 1961/2: no. 319, 46). But the dry-stone techniques reportedly displayed by the upstanding remains on this site present a curious contrast and would appear to preclude closer archaeological dating without excavation.

This lack of masonry evidence also presents a notable contrast with the network of lime-bonded masonry castles surviving in Knapdale and Glassary, highlighting that the island fortresses most widely associated with the Norwegian crown did not serve as administrative centres, and often lacked the sheltered anchorages required to effectively perform this function (cf Martin 2017). Further work is required to establish how the large enclosure, burgh and west pele at Tarbert mediated the movement of materials and people across the isthmus, but the medieval portage legends and the castle itself do appear to symbolise political authority, commercial exploitation and resource protection more clearly than military might (cf McNeill 2017), and the need for the crown to protect Scottish interests against Norwegian invasion through castle construction here and elsewhere should not be overstated. From this perspective Urquhart Castle does indeed provide an interesting comparator for future work.

As a strategically important administrative centre, we would expect Tarbert Castle to be associated with stone and lime structures from an early period, but it is also likely the site contained timber-framed buildings and this allows an indirect link to the c 1250 MacGillechrist grant and the 1326 exchequer accounts. Tradition reports that many Highland chieftains (as in Gaelic Ireland) occupied thatch-roofed, timber-framed, wicker and/or turf-walled buildings throughout the later medieval period (Thacker in prep), and this provides a convincing context for an early-16th-century report suggesting that the bishops’ palace at Dunkeld was constructed ‘in highland style’ before 1400 (Vit. Dun.: 16; Oram 2008), for the early-13th-century ‘great house of wattles’ constructed at Dumbarton reported in the Paisley Registrum (Reg. Pass.: 166; Fairhurst & Dunbar 1971: 242), and perhaps also for the ‘cutting and carriage of branches of birch (bulorum) for repairing the hall and chambers’ included in the 1326 accounts relating to Tarbert (OPS II.I: 35).13 Indeed, these latter sources also appear to describe the raising of a timber-framed, thatch-roofed hall on the site, with ‘piers’ supported on lime-bonded masonry underbuilding, and ‘clay and sand’ (mud-walled or perhaps daubed) walls (OPS II.I: 34). These descriptions suggest that the early-14th-century castle showcased a wide variety of traditional constructional techniques, some of which were of longstanding symbolic importance in well-wooded Highland districts but leave no significant upstanding remains. It may not be until this same early-14th-century period that evidence for a castle constable, royal burgh and (in the same year that Robert I died) a deer park (Gilbert 1975) emerges into the surviving documentary record. The MacGillechrist charter suggests that resources from Tarbert were already controlled by an estate ‘sergeant’ in c 1250 (Appendix 1), however, and with accelerating trade in fish and beef, and merchants to the bishop of Glasgow reportedly travelling to and from Argyll by 1275 (McDonald 2017: 21–2; McNeill 2017; Oram 2017: 259–60), timber from the Firth of Clyde and beyond was likely to have become a valuable commodity from an early period – for communities on both sides of the portage.

Regarded as a standalone building for a century and a half, previous interpretations of the courtyarded quadrangular structure surviving on the summit of Tarbert Castle had suggested that this is the earliest upstanding building on the site, citing architectural parallels with royal castles of enclosure constructed in eastern Scotland to support an early-13th-century constructional date which coincides with increased crown authority across Argyll during the reign of Alexander II. This paper has presented a consistent suite of buildings archaeological evidence to suggest that the cross-wall is a multiphase late-13th-century or later building that post-dates the walls of the larger L-shaped enclosure, and documentary evidence to suggest that this feature and the associated inner bailey ranges were constructed in 1325/6.

Mid-13th- to early-14th-century documentary evidence has also been highlighted to suggest that the site’s transition from private to royal control was a more protracted process than previously recognised. These sources imply that a power centre of some kind was located at Tarbert in this period and, whether continuously or not, geohistorical considerations suggest that this is likely to have been the case since the early medieval period at least. But the large L-shaped enclosure which now represents the earliest visible structure on the site displays architectural features which were probably constructed late in the 13th- to 14th-century period under discussion here, and no convincing evidence for a 12th- or early-13th-century caput has yet been recognised. Indeed, it remains possible that all surviving masonry structures on the site were built by the crown after 1315, as the earliest documentary-led interpretations had previously suggested (OPS II.I: 33; RSRS I: lxxi; see above).

In parallel with the nationalist perspectives relating to the legendary 11th- and 14th-century royal portages with which this paper began, recent discourse has continued to present the consolidation or extension of crown authority into 13th-century Argyll as part of a wider cultural struggle (Oram 2014). In these narratives, West Highland and Hebridean resistance to the aggressively expansionist policies of the Scottish crown parallels Irish resistance to Anglo-Norman colonisation in this same period (see Cowan 1990; McDonald 1997; Duffy 2007), continuing a process which eventually led to the loss of local power centres, reduced regional autonomy, and assimilation into the Lowland-centric state of Scotland (Hunter 1999: 13). We can no longer draw on the buildings surviving at Tarbert Castle as early evidence for this colonial historical narrative, however, and it seems likely a more nuanced process took place. Indeed, the evidence for architecturally sophisticated masonry buildings elsewhere across Kintyre and Knapdale suggests that established kin-groups to the west of the Firth of Clyde were already actively negotiating their identities with respect to these progressive forces by the late 12th century, while the Stewarts to the east would come to cultivate a Gaelic ‘persona’ through the adoption of Gaelic epithets, commissioning West Highland grave-slabs, and promoting the cult of St Brendon of Clonfert (Boardman 2007). Such cultural hybridities legitimised the power held by these different administrations and, by supporting the incremental expansion of selected groups, successive 13th-century Scottish kings were thereby able to consolidate royal authority across the region. The evidence presented above to suggest that the MacSweens and MacGillechrists extended their administrative reach into north Kintyre during the early- to mid-13th century provides an interesting example of this strategy and, with space also precluding discussion of the Lamonts, MacEwens and MacNeills here, the kin-network descended from Donnsléibhe was clearly more powerful during Alexander II’s reign than MacSorley-focused scholarship has often acknowledged. Grants made to the monastery at Paisley suggest that these Argyll chieftains were already in the shadow of the Stewarts by the mid-13th century, but their eventual loss of power provides a striking example of how West Highland clan fortunes might also rapidly fall, depending on crown policy and favour.

Against a research background which has continued to investigate how Scottish medieval masonry buildings relate to wider administrative and environmental changes, more precise relative and calendrical dating of the upstanding and buried masonry features surviving on the Tarbert Castle site would be advantageous, and publication of the wider evidence from recent community excavations is eagerly awaited.

Supplementary material: available online at https://doi.org/10.9750/PSAS.151.1348

To all the sons of holy mother church now and to be (or present and future) Donenald Makgilcriste, Lord of Tarbard, sends eternal greeting in the Lord. Know that I with a devout mind have granted, and by this my present charter, for the salvation of the souls of my ancestors and for the salvation of my own soul, have given in pure and perpetual alms on behalf of me and my heirs to God, St. James and St. Mirin of Paisley, and the monks there serving God and to serve him for ever, the free right of cutting, taking, and carrying away all kinds of timber pleasing to those ‘religious’, for the building and maintaining of their monastery and house of Paisley, within all the woods of my whole land. I give also and grant on behalf of me and my heirs to the same ‘religious’ and their men for ever as free entry and exit with all kinds of timber cut, or about to be cut and carried away as the free right of cutting, taking and carrying away without oversight of the sergeant. In testimony of which thing I have affixed my seal to my present charter. These being witnesses, Sir Hugh of Parlciner, perpetual vicar of Kylmacolme; Sir Nicholas, chaplain; Sir Malcolm, chaplain (Mitchell 1886: 9–10).