EILEAN DÒMHNUILL

Eilean Dòmhnuill has produced the largest assemblage of Hebridean Neolithic pottery so far excavated and is therefore central to our understandings of both the insular Hebridean Early/Middle Neolithic ceramic traditions and the artificial loch islets more generally. Eilean Dòmhnuill is a sub-circular artificial islet approximately 20 m across at its widest point lying close to the southern shore of Loch Olabhat in North Uist (Fig. 2). The islet is primarily made up of accumulated occupation material, most notably ash, and the remains of various structures including buildings, perimeter walls, and stone surfaces. It remains uncertain whether these overlie a natural core, although the water around the islet is shallow enough to permit access by wading.

Fig. 2 Eilean Dòmhnuill a’Spionnaidh, Loch Olabhat, North Uist (photo: Mike Copper)

Excavations in 1986–92 revealed 11 principal phases of occupation at the site dating from the 2nd quarter of the 4th to the early 3rd millennium bc (Armit Reference Armit1986; Reference Armit1987; Reference Armit1988; Reference Armit1990; in prep.). It is likely that just one or two stone footed, and probably turf walled, structures stood on the islet during any particular phase. Hearths within the structures varied from a substantial 2 m long kerbed, paved, and subdivided hearth in Structure 8.1 (see Fig. 7) to smaller examples probably used for shorter periods. The stone kerbs of several of these hearths appeared to have been deliberately flattened, notably those found within the early Structure 9.1, while others, including those within Structures 7.1 and 6.1, were built on top of gradually accumulating mounds of ash. While these also date to the site’s earlier period of occupation, erosion may well have removed evidence for similar practices during the later periods. Large quantities of ash were found both outside and inside the structures, occasionally forming sizeable hearth mounds. Indeed, the fabric of the islet as it appears above water is substantially comprised of this material.

Throughout its occupation, Eilean Dòmhnuill was connected to the loch shore by a causeway, initially of wood, but rebuilt in stone at some point probably in the early 3rd millennium bc. From an early stage the islet was entered via an orthostat lined and stone paved passageway (Fig. 3). The stones on either side of the passage curved outwards to form a low wall that probably constituted the footing for a wattle fence or screen that may have encircled the islet and which would have obscured the interior and channelled access. As a response to rising loch levels this façade and the associated entrance hornworks were repeatedly dismantled and rebuilt further back from the shore.

Fig. 3 Entrance passage at Eilean Dòmhnuill (photo: Ian Armit)

It is of interest that, beyond the grinding of grain and the cooking of food, little evidence for ‘domestic’ activities was found at Eilean Dòmhnuill, and the often shallow and finely stratified deposits do not appear to have been significantly disturbed by trampling, as might be expected if animals had been kept on the islet (Armit Reference Armit2003). The apparent absence of fresh dung, as indicated by beetle fauna, adds further weight to this conclusion (Mills et al. Reference Mills, Armit, Edwards, Grinter and Mulder2004, 892).

Throughout its occupation Eilean Dòmhnuill was subjected to periodic inundations, made worse by a long-term rise in average water level in Loch Olabhat. This situation was almost certainly exacerbated by deforestation and soil erosion from the early 4th millennium onwards (Mills et al. Reference Mills, Armit, Edwards, Grinter and Mulder2004). Indeed, environmental degradation, probably linked to turf stripping, cultivation, and overgrazing by cattle and sheep/goats, appears to have been a problem from the earliest stage at the site (ibid., 894).

A significant and extended inundation occurred between the formation of Levels 5 and 4 at some point in the early 3rd millennium bc, leading to the temporary abandonment of the islet. The latest stratigraphic block within Level 5 produced four radiocarbon dates. While the earlier three of these overlapped between 3496 and 3103 cal bc, the final date – on birch charcoal – of 2909–2638 cal bc (at 95% probability, or 2897–2702 cal bc at 68%, OxA-9080, 4215±45 bp) suggests that the Level 5 occupation probably extended into the early 3rd millennium bc. This is corroborated by a date of 3081–2666 cal bc (at 95% probability, or 3006–2704 cal bc at 68%, OxA-9159, 4265±60 bp) on a barley grain from Level 8 that is likely to have been redeposited from Level 5. Two radiocarbon dates were produced from material closely associated with the post-inundation deposits: 2836–2356 cal bc (at 95%, or 2573–2476 at 68%, OxA-9082 bp determination needed) on birch charcoal from Level 4, and 3021–2702 cal bc (at 95%, or 3002–2780 at 68%, OxA-9083, 4275±45 bp) on birch charcoal from Level 3 (Copper Reference Copper2015, appx 1). While these dates imply that the inundation of Eilean Dòmhnuill could conceivably have lasted for up to 200 years, with an extended post-inundation occupation, a shorter period of flooding and abandonment is suggested by the similarity between the pottery from the pre- and post-inundation occupations and the lack of Grooved Ware at the site.

In this respect it is significant that Grooved Ware assemblages are known from the Western Isles at the Udal in North Uist and at An Doirlinn in South Uist. At An Doirlinn Grooved Ware use is modelled to have started at 2800–2620 cal bc (at 95% probability, and probably 2750–2650 cal bc at 68%, Garrow & Sturt Reference Garrow and Sturt2017, 201–2 & fig. 4.37, firstGroovedWare), with the end of the use of Hebridean pottery modelled as 2840–2640 cal bc (at 95%, or 2830–2660 cal bc at 68%, Garrow & Sturt Reference Garrow and Sturt2017, 202, lastHebridean). Furthermore, there are formal and decorative similarities between the Hebridean Grooved Ware and vessels from the Stones of Stenness, Balfarg, Knowth, and Barnhouse (Schulting et al. Reference Schulting, Sheridan, Crozier and Murphy2010; Richards et al. Reference Richards, Meirion Jones, MacSween, Sheridan, Dunbar, Reimer, Bayliss, Griffiths and Whittle2016), and between pottery from Barnhouse and Calanais in Lewis, which all probably date to the 1st century of the 3rd millennium cal bc or earlier (Sheridan Reference Sheridan2016a, 592–5; Richards et al. Reference Richards, Meirion Jones, MacSween, Sheridan, Dunbar, Reimer, Bayliss, Griffiths and Whittle2016). This implies that the An Doirlinn Grooved Ware is unlikely to be the earliest in the Outer Hebrides. Significantly, there is presently no convincing evidence that Grooved Ware overlapped in use with earlier styles at An Doirlinn or at any other site in the Western Isles (Copper Reference Copper2017b, 169). As such, the dating of the earliest use of Grooved Ware at An Doirlinn provides a terminus ante quem for the presence of Grooved Ware in the Outer Hebrides. The lack of evidence for an overlap in use of Grooved Ware and pre-existent Neolithic ceramic styles at any Hebridean site implies, furthermore, that the appearance of this style of pottery was associated with the demise of the indigenous Early/Middle Neolithic styles. Its absence from Eilean Dòmhnuill therefore argues against an extended inundation and subsequent re-occupation extending into the mid-3rd millennium bc.

Grooved Ware therefore appears to represent a significant break with the Hebridean Early/Middle Neolithic ceramic traditions that probably began in the first couple of centuries of the 3rd millennium. Its absence from Eilean Dòmhnuill suggests either that the settlement was abandoned before Grooved Ware replaced the insular pottery traditions, or that for some reason Grooved Ware was not adopted by the inhabitants even though they were aware of its existence. Either way, an extended period of inundation would suggest that the later phases at Eilean Dòmhnuill would probably have extended into the mid-3rd millennium bc, a scenario that appears unlikely in light of the absence of Grooved Ware at the site despite its presence at An Doirlinn.

HEBRIDEAN HOMOLOGIES

Structurally, Eilean Dòmhnuill represents a compound, enclosed by a wall/palisade, approached by a causeway and containing (in most phases) probably just a single building. This has led several researchers to draw attention to similarities between Eilean Dòmhnuill and Hebridean passage tombs (eg, Armit Reference Armit2003, 98; Mills et al. Reference Mills, Armit, Edwards, Grinter and Mulder2004, 889; Cummings & Richards Reference Cummings and Richards2013, 198). Where the tombs are entered along a narrow passage, the islet was approached via a narrow causeway; where the tombs are hidden within cairns, Eilean Dòmhnuill was at least partially hidden by a palisade; elaborately decorated pottery is found at Eilean Dòmhnuill as well as in the chambered cairns. In their own ways both Eilean Dòmhnuill and the cairns are places apart.

Eilean Dòmhnuill is not alone in this respect. Other sites on artificial islets or natural loch islands that have produced similar elaborately decorated pottery to that found at Eilean Dòmhnuill occur elsewhere in the Outer Hebrides, including at Eilean an Tighe in North Uist (Scott Reference Scott1951; Squair Reference Squair1998a, 303–59); Loch a’Choire in South Uist (Henley Reference Henley2012); and Pigmies Isle and a series of recently discovered artificial islet sites in freshwater lochs in Lewis (MacKenzie 1904–Reference MacKenzie5; Stevenson Reference Stevenson1945–6; Sheridan et al. Reference Sheridan, Murray, Garrow and Anderson-Whymark2014; Garrow et al. Reference Garrow, Sturt and Copper2017b). To this might be added a small group of sherds of distinctively Hebridean Early/Middle Neolithic form from the remote island of Hirta in St Kilda (Fleming & Edmonds Reference Fleming and Edmonds1999, 152–3; Gannon & Geddes Reference Gannon and Geddes2015, 27–9; Copper Reference Copper2017a). Recent dates from the Lewis lochs (on charred pot residues and preserved Salix and Alnus wood) and Eilean an Tighe (on charred pot residues) range from 3640–3380 to 3510–3340 cal bc (at 95% probability: OxA-28951, 4749±30 bp, and OxA-28954, 4610±29 bp) (Garrow & Sturt Reference Garrow and Sturt2017, 204; Garrow et al. Reference Garrow, Griffiths, Anderson-Whymark and Sturt2017a, appx 1; 2017b). These dates are broadly contemporary with the early and middle periods at Eilean Dòmhnuill. Significantly, as at Eilean Dòmhnuill, no characteristically Late Neolithic or Early Bronze Age material has yet been recovered from any of these islets.

The Lewis islets hold a particular significance in respect of the interpretation of sites such as Eilean Dòmhnuill. Recent work on a small islet in Loch Langabhat, measuring just 127m², that had been significantly enhanced by the deposition of a large number of boulders around a natural core, failed to find any evidence for occupation during the Neolithic and no evidence for a causeway could be identified (Duncan Garrow pers. comm.). Despite this, the large quantity of well made and elaborately decorated Early/Middle Neolithic pottery recovered from the loch bed around the islet as well as from the island itself, together with the 4th millennium bc dates from the structural timbers associated with the nearby and structurally similar artificial islet in Loch Bhorgastail, are strongly indicative of an Early/Middle Neolithic origin for this site. Pottery from both sites has produced residue dates falling within the later 4th millennium bc (Garrow et al. Reference Garrow, Sturt and Copper2017b; Duncan Garrow pers. comm.). Construction of the islet in Loch Langabhat would have been a major undertaking, yet it was evidently not a settlement, which raises the question of why it – and the other islet sites, including Eilean Dòmhnuill – served as foci for the deposition of such large quantities of fine pottery.

THE EARLIEST POTTERY IN THE OUTER HEBRIDES

Before discussing the nature and possible roles of Eilean Dòmhnuill and the other Hebridean loch islets, it is necessary, given its potential significance at these sites, first to consider in more detail the character of Hebridean Early/Middle Neolithic pottery. Eilean Dòmhnuill has produced the earliest Neolithic dates from the Outer Hebrides. A date of 3792–3537 cal bc (at 95% probability, or 3711–3639 at 68%, OxA-9085, 4895±50 bp) was obtained from heather charcoal from a stratigraphically early layer of organic material (possibly flooring) containing pottery. In addition, a date of 3704–3521 cal bc (at 95%, or 3659–3532 at 68%, OxA-9079, 4830±45 bp) was obtained from a barley grain from a layer overlying the primary hearth of Structure 9.1, the earliest positively identified structure on the islet (Copper Reference Copper2015, appx 1). As pottery is reliably associated with all of the phases of occupation at Eilean Dòmhnuill, these dates also currently represent the earliest dates for pottery in the Western Isles. At the time of writing, a Bayesian model for the site is in preparation, although preliminary indications are that occupation probably began there in 3720–3510 cal bc (at 95%, or 3660–3560 cal bc at 68%; Garrow et al. Reference Garrow, Griffiths, Anderson-Whymark and Sturt2017a, 115, Start Eilean Dòmhnuill). It is important to note that while these dates derive from the earliest excavated layers, they by no means represent the base of the islet; therefore, occupation must have begun somewhat earlier.

With regard to Hebridean pottery more generally, an updated version of a recently published (Ashmore Reference Ashmore2016, 590) Bayesian model suggests a date of 3845–3590 cal bc (at 95%, and probably 3745–3645 cal bc at 68%) for the start of the Hebridean Neolithic pottery traditions (Copper Reference Copper2015, 391–3, start_neolithic_ceramic). A more recent model, utilising data from 13 Outer Hebridean sites – but significantly, though unavoidably, excluding monuments due to a lack of available dates (Garrow et al. Reference Garrow, Griffiths, Anderson-Whymark and Sturt2017a, 115–7) – suggests that Neolithic practices (including pottery manufacture) arrived in the Western Isles between 3800 and 3700 cal bc. This was probably not much later than their first appearance on the Scottish mainland.

The pottery from several chambered cairns of the Western Isles provides close comparanda with material from Clyde cairns in Arran, Argyll, and Bute, which has, over the years, led to much discussion of the nature of interaction between these areas at the start of the Neolithic (Piggott Reference Piggott1954, 231–2; Scott Reference Scott1964, 150–8; Henshall Reference Henshall1972; Sheridan Reference Sheridan2000, 9–11; Copper Reference Copper2015). It is therefore unfortunate that none of the Hebridean cairns has been radiocarbon dated. Reliable dates for Clyde cairns outwith the Western Isles would be significant given the likelihood that the earliest Hebridean cairns were modelled after them. However, secure dating of Clyde cairns is complicated by the small number of dependable radiocarbon dates available. Whittle and colleagues suggest that Clyde cairns began to be constructed between 4295–3495 cal bc (at 95% confidence, and probably 3800–3650 cal bc at 68%; Whittle et al. Reference Whittle, Bayliss and Healy2011, 829–30 & fig. 14.169, start Scottish chambered cairns), while Schulting and colleagues propose a start date of 3700–3570 cal bc for the use of the closely related court cairns of Ireland, with their initial construction perhaps dating slightly earlier (Schulting et al. Reference Schulting, Murphy, Jones and Warren2012, 27, 30 & fig. 9). These further support the idea that the earliest Hebridean pottery dates to some point after 3800 cal bc, and probably to the 2nd quarter of the 4th millennium bc.

It is possible that some of the vessels from Hebridean chambered cairns pre-date the earliest pottery from Eilean Dòmhnuill. Unfortunately, the difficulty of attributing individual vessels to stages in the use of a tomb ultimately restricts the chronological value of the chambered cairn assemblages in terms of developing a typological sequence. This is highlighted by the assemblages from the Point of Cott in Orkney and Blasthill in Kintyre (Barber Reference Barber1997, 69; Cummings & Robinson Reference Cummings and Robinson2015, 6–9). Given its already-developed, distinctively insular character, the earliest securely dated Hebridean pottery is unlikely to be the earliest pottery in the Western Isles. However, while formal similarities with putatively earlier pottery from chambered cairns around the Firth of Clyde could imply that the pottery from Hebridean chambered cairns is amongst the earliest in the Western Isles, presently known dates do not fully support this conclusion.

In addition to vessel forms linked to those found within the Clyde tombs of Argyll, Bute, and Arran, several hundred ‘Unstan bowls’ – small, shallow round-bottomed bowls with vertical collars (see Fig. 4.2) – have been found at sites in the Western Isles. Named after the eponymous chambered cairn in Orkney, their presence in the Western Isles would at first sight seem to suggest that an additional source for some elements of the Hebridean ceramic assemblage, and arguably for the Hebridean Neolithic as a whole, may have lain in Orkney where such vessels have been found at a number of Early Neolithic chambered cairns (Davidson & Henshall Reference Davidson and Henshall1989, 64–84) and settlement sites (Ritchie Reference Ritchie1983; Richards & Jones Reference Richards and Jones2016). As such, the relative dating of the advent of Neolithic practices in Orkney and the Western isles is of some significance.

Fig. 4 Characteristic Hebridean vessel forms from Eilean Dòmhnuill: 1) Multiple ridged baggy jar; 2) Unstan-type bowl; 3) Shouldered bowl; 4), Necked bowl (after Armit in prep.). Scale: 50 mm

Currently, the earliest dates for Neolithic activity in Orkney – from willow charcoal closely associated with wheat, barley, and other grains beneath a later mound at Varme Dale – are 3766–3536 cal bc (at 95% probability, or 3699–3639 at 68%, AA-53158/GU-10629, 4875±45 bp) and 3767–3635 cal bc (at 95%, or 3698–3646 cal bc at 68%, AA-53157/GU-10628, 4890±40 bp; Griffiths Reference Griffiths2016, 296–7). Kerns’ (Reference Kerns2016) Bayesian model suggests a start to the Orkney-Cromarty cairn sequence in Orkney of 3783–3384 cal bc (at 95% probability, or 3596–3419 cal bc at 68%, Start Orkney Cromarty Chambered Cairns), although his modelled dates for interments begin much later (Kerns Reference Kerns2016, 20–1 & table 2.1). Kerns’ model for Orcadian settlements suggests a beginning at 3794–3385 cal bc (at 95% probability, or 3611–3429 cal bc at 68%; Kerns Reference Kerns2016, 37 & table 2.2, start Orcadian Neolithic Settlement Sequence).

Griffiths (Reference Griffiths2016, 292) models the earliest datable activity at an Orcadian chambered cairn as 3640–3440 cal bc (at 95% probability, or 3570–3470 cal bc at 68%, FirstOrkney Cromarty). Based on data from timber and stone built stalled houses and chambered cairns, she suggests a start to the Orcadian Neolithic of 3730–3480 cal bc (at 95% probability; Griffiths Reference Griffiths2016, 272, Start Orkney Neolithic). She notes, however, that the currently available evidence may well under-sample and under-estimate the timing of this activity (ibid., 301). More recently, it has been argued that the first reliably dated Orcadian houses were in use from 3560–3360 cal bc (at 95% probability, and probably from 3445–3370 cal bc at 68%; Bayliss et al. Reference Bayliss, Marshall, Richards and Whittle2017, 1181, start_timber_houses).

The similarity between what are currently the earliest Neolithic dates from the Western Isles and Orkney means that there is at present no good reason to assume that Unstan bowls necessarily represent an introduction of a new style of pot from Orkney. Though present dating does not entirely rule out this possibility, it currently leaves open other scenarios, such as the emergence of this vessel form on the mainland prior to its adoption in Orkney and the Hebrides, perhaps at more or less the same time. Indeed, as one of us (Copper Reference Copper2015, 413–4) has previously pointed out, it is ironic that although Unstan bowls were originally named after large vessels from Orcadian chambered cairns, ‘the vast majority… are not large, are not from Orkney, and do not come from chambered cairns’. Hereafter, the term Unstan-type bowl will be used to refer specifically to the homogeneous small bowls of this type to distinguish them from the larger vessels found in some Orcadian chambered cairns. Use of the term ‘Unstan ware’ should, in the authors’ opinion, be discontinued, as it falsely implies that the bowls form a component of a specific and recurrent set of vessels in all areas where they are found.

POTTERY AT EILEAN DÒMHNUILL

Analysis of the very large pottery assemblage from Eilean Dòmhnuill has provided an overview of the nature and development of Hebridean Neolithic pottery at the site throughout the 4th and into the 3rd millennium bc. The assemblage has highlighted important differences between the elaborately decorated pottery from this and other sites, including loch islets, and the notably less elaborate assemblages excavated from a small number of sites of the same date elsewhere in the Western Isles. The Eilean Dòmhnuill assemblage comprises 22,281 sherds, probably representing around 1900 vessels. It is likely that less than 20% of the vessels were left undecorated, with almost all plain vessels being of simple bowl or cup forms. The pottery at Eilean Dòmhnuill is both highly eclectic (in that there are lots of vessel forms) and also conservative (as all of them were in use throughout the life of the site). The large number of vessel forms in use throughout its occupation means that defining exclusive vessel categories is difficult, yet certain types stand out as notably more coherent (that is, bear a strong ‘family resemblance’, Clarke Reference Clarke1978, 231–3). Of these, the most significant are:

• ∙ Multiple ridged baggy jars: round-bottomed baggy or ovoid jars decorated with parallel raised or applied horizontal cordons that usually separate bands of sloping incised lines often forming a herringbone motif (Fig. 4:1).

• ∙ Unstan-type bowls: shallow bowls with vertical or slightly concave decorated collars (Fig. 4:2).

• ∙ Shouldered bowls: decorated, sharply carinated bowls of closed form (Fig. 4:3). Closely related vessels from Argyll and around the Firth of Clyde have been termed Beacharra bowls (Childe Reference Childe1935, 66; Piggott Reference Piggott1954, 170–3; Scott Reference Scott1964, 150–8).

A series of other less coherent vessels forms, including a range of necked vessels, also characterises Hebridean Neolithic assemblages (Fig. 4:4).

Over 95% of Unstan-type bowls and many shouldered bowls bear a motif of horizontal grooves over sloping or vertically incised lines. It could be proposed that the sharing in the Western Isles of distinctive motifs on Unstan-type and shouldered bowls, whose distributions overlap only in this one region, may imply that these types of pot were understood as regional variants of the same vessel category. At Eilean Dòmhnuill the two forms are often hard to tell apart, and the distinction may indeed not be valid in this area. It has been argued that the earliest Hebridean pottery resulted from a blending and reworking of elements of potting traditions found in north-east Scotland and further south along the Atlantic façade (Copper Reference Copper2015, 397–436). Such a situation may parallel that noted for chambered cairns, which show a similar combination of features found separately elsewhere (Scott Reference Scott1942; Henley Reference Henley2004). More speculatively, it is conceivable that there could be an indigenous element reflected in insular vessel types such as the multiple ridged baggy jars, although local innovation could account for this equally well.

An important aspect of the Eilean Dòmhnuill pottery is that all of the key vessel forms were in use throughout the life of the site. Although the low number of vessels of unambiguous form dating to the period after the major inundation of the early 3rd millennium means that this conclusion must be treated with some caution for the later part of the occupation, it is nonetheless clear that the pottery from the later levels does not alter substantially in character from that associated with the pre-inundation occupation. This strongly suggests that, as they are in place from the earliest phases at Eilean Dòmhnuill (ie before 3500 cal bc, and probably before 3650 cal bc), the distinctively Hebridean vessel types, including multiple ridged baggy jars and shouldered bowls, must have developed before that date. It also suggests that the remarkably homogeneous Unstan-type bowls, also present from the earliest phase, were in use in the Hebrides from, or very shortly after, the very start of the Neolithic in this region. Indeed, Unstan-type bowls are the most common vessel type recognised at Eilean Dòmhnuill, a situation also noted at Northton in Harris (Johnson Reference Johnson2006, 64), and are very common at Eilean an Tighe, both in the Western Isles. Remarkably, the only notable change in the nature of the Eilean Dòmhnuill assemblage is a gradual but statistically significant decline in the proportion of collared rims (R²= 0.79919, p-level=0.00275) and a corresponding rise in the proportion of everted rims (R²=0.77497, p-level=0.00391) between Level 11 (early) and Level 1 (late) (Fig. 5).

Fig. 5 Diachronic variation in rim types at Eilean Dòmhnuill as percentages of all vessel rims recorded within each level. At no point is any rim form entirely absent. The only statistically significant change is a proportional decline in collared rims and an increase in everted rims (adapted from Copper Reference Copper2015, fig. 5.1)

Fabric groups represent variations on a theme and cross-cut all vessel types, with the size of added opening agents correlating closely with vessel size. Opening agents were derived from igneous gneiss, suggesting that all of the vessels were manufactured locally. Techniques of manufacture and decoration do not alter throughout the life of the site, while in all phases decoration is dominated by motifs formed from incised vertical, horizontal, and sloping lines. Where decorative elaboration occurs (Fig. 6) it is almost invariably the result of the multiplication of elements found separately on other vessels rather than by the use of novel motifs (Copper Reference Copper2015, 303–14). Overall, the picture arising from the statistical analysis is of a strikingly conservative tradition in which vessel forms and decorative motifs were reproduced over a considerable period of time.

Fig. 6 Decorative elaboration through the multiplication of motifs on a Hebridean shouldered bowl (after Copper Reference Copper2015, fig. 6.10). © National Museums Scotland, reproduced by permission

Elsewhere in the Western Isles a few assemblages differ somewhat in character from the elaborate pottery found at Eilean Dòmhnuill, although there is no evidence that they were any less conservative in respect of the reproduction of standard vessel forms and decoration. At An Doirlinn, decorated pots and elaborate vessel forms are rare (Copper Reference Copper2017b), while at both An Doirlinn and Screvan Quarry (Squair Reference Squair1998b), multiple ridged jars and Unstan-type bowls were absent. The existence of relatively plain assemblages at some sites and assemblages characterised by elaborately decorated vessels and high numbers of Unstan-type bowls and multiple ridged jars at others (including Eilean Dòmhnuill and the other Hebridean loch islets) raises the question of the nature of the sites from which they come and the activities that may have prevailed at such locales.

DISCUSSION: THE HEBRIDEAN PARTY SCENE

A number of factors suggest that the significance of Eilean Dòmhnuill extended beyond that of a straightforward domestic settlement. These include the flood-prone nature of the islet that would have necessitated periodic (and probably seasonal) abandonment; the presence of substantial hearths (Fig. 7) and large quantities of ash despite the lack of evidence for activities beyond the later stages of food preparation; the large quantities of elaborately decorated pottery that are absent from contemporaneously occupied sites such as An Doirlinn; and the site’s potentially homologous relationship with passage tombs. This observation of significance beyond straightforward domestic settlement is even more pronounced for the smaller artificial islets such as Loch Bhorgastail and, particularly, Loch Langabhat where no evidence for occupation was found. This is despite the effort that had gone into the construction of the islet and the presence of elaborately decorated pottery, including multiple ridged jars and Unstan-type bowls, in the surrounding waters (Garrow et al. Reference Garrow, Sturt and Copper2017b). Domesticity alone would therefore appear insufficient to explain the nature of most of the Hebridean islet sites, including Eilean Dòmhnuill. Taking into account the features noted above it is suggested here that one possible explanation for the unusual nature of the Hebridean islets is that such locales were associated with formal gatherings in which the preparation and consumption of food – feasting – held a particular significance. While feasting is here understood as communal consumption that differs from everyday commensality, it should be stressed that this need not imply particularly large numbers of people. Nevertheless, the importance of feasting in the mediation of inter-communal relations within many recent and contemporary small-scale societies implies that it may well have played a significant role in the Hebridean Neolithic, a possibility that would at the very least be congruent with the unusual nature of Eilean Dòmhnuill and the other islet locales and which necessitates a brief consideration of the nature and materiality of feasting more generally.

Fig. 7 Large two-part hearth within Structure 8.1, Eilean Dòmhnuill (photo: Ian Armit)

Following Dietler and Hayden (Reference Dietler and Hayden2001b, 3) the term ‘feast’ may be minimally defined as the ritualised communal consumption of food in a way that differs from everyday meals. It is also helpful to distinguish specific feasts from feasting as a process. In this respect, identifying certain recurrent features across similar sites, including distinctive ceramic forms or purpose-built structures, strengthens the argument for their association with feasting (Twiss Reference Twiss2008; see also McVeigh Reference McVeigh2016, 71–8 for a recent discussion of feasting in the British and Irish Neolithic). Hayden (Reference Hayden2001, 24–7) has proposed that feasts are a form of ‘social technology’ or ‘political ecology’ in which food surpluses are converted into goods and services, while Dietler (Reference Dietler2001, 66–7) argues that feasts constitute a distinctive kind of inherently political, symbolic, ritual practice in which relationships are negotiated, social and economic goals pursued, and power and ideology contested. Twiss (Reference Twiss2008, 436) has noted that feasts are ‘… inherently polysemic, arenas of simultaneous competition and integration … doubly suited to early agricultural societies’. During feasts symbolic references to the past create a sense of continuity and naturalise social structures. Through the providing and/or attending of feasts reciprocal obligations may be established and relations of superiority and inferiority created and strategically manipulated (Dietler Reference Dietler2001, 73–4). Feasts are such important loci of social, economic, and political action (Friedman Reference Friedman1975, 170–1; Hayden Reference Hayden1990) that, it has been argued, they could function almost as a form of currency (Russell Reference Russell1998, 46), facilitating the creation of social and economic debt (Mauss Reference Mauss1954). In addition, commensal gatherings provide important locales in which commonly understood behaviours may emerge and be subsequently reinforced or contested (Bourdieu Reference Bourdieu1977; Giddens Reference Giddens1984). In this way, feasting, at many scales, is a key component in the creation and negotiation of social capital within small-scale societies.

It has been argued that hunter-gatherers often have a strong ethic of sharing that precludes gift-giving from being understood as a generous act requiring reciprocation (Winterhalder Reference Winterhalder1996; Russell Reference Russell1998, 48–9; Kelly Reference Kelly2007, 161–81). However, contact with intrusive agricultural groups could well have created pressures to adopt new economic practices in order to provide a surplus that could be used to maintain social capital through the provision of feasts. In this respect, it is notable that Thomas (Reference Thomas2013, 124–5) has argued for the centrality of feasting across the Mesolithic–Neolithic transition, particularly in relation to cattle (a related aspect of such a process could have been the necessity to provide bridewealth (Russell Reference Russell1998, 51)). Such practices would mean that a significant proportion of a community’s resources would have needed to be given over to providing feasts if effective socio-economic relations with its neighbours were to be maintained. In many parts of Europe this is something that may well have been all but impossible without the adoption of an agricultural economy (though significant exceptions are likely to have existed in areas such as southern Scandinavia where highly productive natural environments could have supported dense populations). Whether the need to provide provisions for feasting would have encouraged the adoption of farming or whether the adoption of farming provided the opportunity to engage in feasting is difficult to say, and the two are not necessarily mutually exclusive. If the former, then the arrival of even small numbers of agriculturalists could have been sufficient to act as a catalyst for the adoption of social practices centred around ritualised commensality. Recent research now raises the possibility, however, that a pre-existent political economy of this nature could have been transplanted more or less wholesale onto the Scottish Atlantic façade as a result of large-scale migration (Cassidy et al. Reference Cassidy, Martiniano, Murphy, Teasdale, Mallory, Hartwell and Bradley2016; Brace et al. Reference Brace, Diekmann, Booth, Faltyskova, Rohland, Mallick, Ferry, Michel, Oppenheimer, Broomandkhoshbacht, Stewardson, Walsh, Kayser, Schulting, Craig, Sheridan, Parker Pearson, Stringer, Reich, Thomas and Barnes2018). While the new aDNA data may mask subtle regional variations, the scale of population replacement at a macro-level now seems undeniable and needs to be taken into account in any explanation of the Hebridean Neolithic.

The semiotic potential of material culture used in feasts includes the marking of social boundaries such as age and sex, and the separating of insiders from outsiders (though the relationship between ceramic style and social identity is far from straightforward: cf. David et al. Reference David, Sterner and Gavua1988; Gosselain Reference Gosselain2011). Furthermore, feasts may be differentiated from everyday food consumption events by various framing devices, including the use of special vessels and settings (Miller Reference Miller1985, 181–3). The visibility and semantic potential of pottery means that it has a particular salience within commensal ritual. Jordan (Reference Jordan2015, 35–6) notes that the tight integration of dining activities and artefacts means that they may persist as a ‘coherent combination of specific cultural traits and attributes’, while Miller (Reference Miller1985, 157–83) has argued that ceramic variability may form a ‘semiotic code’ intimately bound within social practices. To paraphrase Wittgenstein (Reference Wittgenstein1953, pl. 43), the meaning of a pot is its use in practice. However, the communal nature of the practices within which pots are deployed may work to constrain understandings of their significance and meaning. In this respect the specific understanding of the nature of a vessel category will be to a greater or lesser extent normative. Furthermore, the existence, longevity, and broad spatial extent of a highly coherent vessel form – that is, where tokens bear a strong family resemblance, such as the Unstan-type bowl – particularly when associated with a specific type of locale, implies that such vessels may have been understood in very similar ways across space and through time.

We have argued that large assemblages of elaborately decorated Hebridean vessels and more widespread, highly coherent vessel categories such as Unstan-type bowls, being associated with distinctive and (often literally) isolated locales such as Eilean Dòmhnuill (itself producing querns, hearths, and large quantities of ash but little evidence for other ‘domestic’ activities) implies that we are looking at the residue of iterated feasting events. If this is the case, then the conservatism of the assemblage may well have resulted from the necessity of maintaining commonly understood cultural practices and artefacts within choreographed settings that mediated social interaction within and between communities. The persistence and broad geographical spread of certain highly coherent vessel forms, most notably the Unstan-type bowl, implies that the significance carried by such vessels inhibited formal and decorative change. Display appears to have been an important element of the elaborately decorated Hebridean assemblages, and the importance of reproducing commonly understood vessel forms and decorative schemes suggests that compliance with social norms within the environment in which such vessels were employed was of some significance throughout the later 4th and earliest 3rd millennia bc. In this respect, it is of interest that while divergence from the standard range of forms and/or decoration does occur at Eilean Dòmhnuill (Fig. 8), it is notably rare.

Fig. 8 A rare example of curvilinear decoration on a necked jar from Eilean Dòmhnuill (after Copper Reference Copper2015, fig. 6.46) © National Museums Scotland, reproduced by permission

Although the pots found on the loch beds around the smaller islets in Lewis must have been used elsewhere (although perhaps nearby), the nature of their deposition, and of the small islet sites themselves, which have produced little evidence to suggest that they were permanent settlements, implies a significance that goes beyond the mundane and may well have been ceremonial in nature. Indeed, it is possible that there was some equivalence between deposition in lochs in the Western Isles and in pits or at causewayed enclosures elsewhere. Conversely, the less elaborated assemblages from sites such as An Doirlinn and Screvan Quarry may represent the opposite situation. Here decorative embellishment and the production of specific vessel forms was less important than the potential of the pots as storage or cooking devices; display does not seem to have been such an important consideration at these latter sites.

It could be argued that feasting was complementary to agriculture in the Neolithic, and that the environmental degradation around Loch Olabhat in the 4th millennium bc may well have resulted in large part from the pressures to provide resources for feasts to maintain social capital. At the same time, over-intensive agricultural regimes, such as that which may have led to environmental damage around Loch Olabhat (Mills et al. Reference Mills, Armit, Edwards, Grinter and Mulder2004), may have created inter-communal tensions that were resolved through practices structured – ironically – through commensal ritual. If so, then the conservatism in social practice and technology that is evident in the nature and context of the Hebridean ceramic assemblages, and which is highlighted in particular by the pottery from Eilean Dòmhnuill, exemplifies the importance of iterated inter-communal commensal ritual in maintaining social relations in the Western Isles at this time. Furthermore, the presence of Unstan-type bowls, but not other Hebridean vessel forms, in Orkney and on the Scottish mainland suggests that these connections and the associated shared understandings are likely to have extended well beyond the Western Isles themselves.

Accounting for the end of the indigenous Hebridean ceramic traditions, probably between the 29th and 27th centuries bc (Copper Reference Copper2017b, 172), is fraught with difficulties and most likely cannot be separated from larger-scale social processes in operation at this time. Recently, Carlin (Reference Carlin2017) has argued that the seemingly marked disjuncture between the Middle and Late Neolithic in Britain and Ireland is in reality far from clear and that Grooved Ware was adopted outside of Orkney along with other distinctively Late Neolithic practices as part of more gradual processes of social change. Although focusing primarily on Ireland, Carlin’s argument could be extended to other areas between the Brú na Bóinne and Brodgar, notably the Scottish Atlantic façade including the Outer Hebrides. At present none of the Outer Hebridean islet sites has produced Grooved Ware, which raises the question of whether there was indeed a significant cultural shift in this region shortly after 3000 cal bc. Arguing for long-term continuity is the construction of a very late miniature bipartite chambered cairn inside the earlier stone ring at Calanais (Ashmore Reference Ashmore2016); the absence of Irish or Orcadian-style developed passage tombs in the Western Isles (though large Hebridean-style passage tombs certainly exist); and the deposition of a Grooved Ware pot in the simple passage tomb of Unival much as earlier styles were deposited (Scott Reference Scott1948). It is worth noting in this respect that Grooved Ware has also been found inside a Clyde cairn at Tormore on Arran (Henshall Reference Henshall1972, 305 & 371–2). Calanais itself has a long and complex history that may have included ceremonial activities that predated the stone ring and rows (Richards Reference Richards2013, 270–1), although the first archaeologically visible activity relating directly to the megalithic monument is likely to date to the 1st century or so after 3000 bc (Ashmore Reference Ashmore2016, 948–9). Whether this is taken to mark continuity or disjuncture, however, depends on which aspects of the monument’s history are emphasised. Significantly, Sheridan (Reference Sheridan2016a, 594) has suggested that the deposition of the single Grooved Ware pot at Calanais may have been associated with rituals related to the early use of the site.

Recently, Bayliss and colleagues (Bayliss et al. Reference Bayliss, Marshall, Richards and Whittle2017) have suggested that social developments in Orkney in the late 4th millennium bc, reflected in changing house styles, degrees of settlement nucleation, tomb architecture and use, and artefactual styles that included the development of Grooved Ware, resulted from a complex interplay of local and pan-regional processes and interconnections. Such developments would no doubt also have impacted upon communities in the Western Isles. The presence of Grooved Ware at An Doirlinn, with no apparent overlap in use with Hebridean styles, and at site RUX6 at the Udal, where there is no unambiguous evidence for pre-Grooved Ware activity (Ballin Smith Reference Ballin Smith2018), could strengthen the case for cultural discontinuity in the Western Isles. However, this needs to be placed within the context of a society in which fast-developing, long-distance connections may well have provided affordances for micro-regions or for certain sections of society that were not previously available – a process reflected more broadly across the whole of Britain and Ireland at this time (Ashmore Reference Ashmore2004; Sheridan Reference Sheridan2004; see also Helms Reference Helms1988). Although such a situation may go some way towards accounting for the appearance of Grooved Ware in the Outer Hebrides, and perhaps for the development of the Calanais complex, it would not explain why indigenous ceramic styles and their associated practices, including gatherings at, or near to, artificial loch islets, should be abandoned altogether.

One possible explanation for the changes occurring in the early 3rd millennium bc is that there was a general population collapse across Britain and Ireland in the Middle and/or Late Neolithic, possibly as a result of climatic changes, disease, or endogenous factors (Stevens & Fuller Reference Stevens and Fuller2012; Reference Stevens and Fuller2015; Shennan et al. Reference Shennan, Downey, Timpson, Edinborough, Colledge, Kerig, Manning and Thomas2013; Whitehouse et al. Reference Whitehouse, Schulting, McClatchie, Barratt, McLaughlin, Bogaard, Colledge, Marchant, Gaffrey and Bunting2014; Woodbridge et al. Reference Woodbridge, Fyfe, Roberts, Downey, Edinborough and Shennan2014; Bevan et al. Reference Bevan, Colledge, Fuller, Fyfe, Shennan and Stevens2017). It has been argued that such an event could have impacted upon the Scottish islands between c. 2900 and 2800 cal bc, possibly somewhat later than on the mainland (Stevens & Fuller Reference Stevens and Fuller2015, 865–8). One potential implication for the Western Isles is that significant population replacement could have followed as the result of movement from elsewhere into the largely depopulated landscape of the islands, resulting in the importation of new practices, monuments, and artefacts into the region.

The idea of a significant decline in population has, however, been subjected to criticism, notably on methodological grounds from Bishop (Reference Bishop2015a; Reference Bishop2015b), who drew particular attention to the use by Stevens and Fuller (Reference Stevens and Fuller2015) of summed radiocarbon probability distributions. Bishop argued that the sample size on which Stevens’ and Fuller’s research was based was insufficient to support their conclusions and that the dates available were biased by the nature of archaeological sampling strategies. Recently Bevan and colleagues (Bevan et al. Reference Bevan, Colledge, Fuller, Fyfe, Shennan and Stevens2017) have argued that they have been able to control for sampling bias and that a sufficient number of dates now exists to support the suggestion of a primarily climate-driven downturn in population between c. 3500 and 3000 cal bc, with a more gradual decline until the mid-3rd millennium bc.

Regardless of the outcome of such debates, however, there exist further problems with positing population replacement as a cause of cultural change in the Hebridean Late Neolithic. These include why movement should occur into a region experiencing potentially severe environmental stress, even if such stress was being felt more generally, and why such problems should have resulted in population movement in the 3rd millennium bc when there exists no evidence for movement in the latter 4th millennium bc. As such, the reasons behind the abandonment of the islet locales and the decline of the indigenous Hebridean Neolithic ceramic traditions must, for the time being at least, remain obscure.