Given that settlement at Bagh nam Feadag spans a vast timescale, the text below offers an outline of some of the major events in the development of the Western Isles and those that relate to the site presented here. This is by no means an exhaustive account of such a complex story which archaeologically speaking, with some sixty years of investigation, is very much in its infancy.
Wheelhouses...
Aerial view of Grimsay Wheelhouse showing typical 'wheel spokes'.
A wheelhouse can be described as a building with a distinctive ground plan, having a circular outer wall that encloses a variable number of regularly spaced cells that open onto a central space. The term wheelhouse derives from the similarities of the structure in plan to a wheel with radial spokes. Although wheelhouses are generally referred to as a standard building form no identical examples are known when the structural features are examined in detail. For example, the overall diameters, internal space and number of bays, along with additional features such as entrance passages and additional bays, vary from one site to another (Crawford 2002, 230 table 3 pp; McKenzie 2003, 8, 24-41). Wheelhouses are found in two types of location; machair and moorland. Given that only five out of the currently identified thirty-one Western Isles wheelhouses are located on moorland, archaeologists have argued that this indicates the machair as the preferred location (e.g. Armit 1996, 84). Although this may be the case, the lack of survey on the moorland in recent years and deliberate concentration on the machair strip by archaeologists, notably the Sheffield Environmental and Archaeological Research Campaign, has compounded the distinction that I would argue is not as significant as currently thought.
The number of mounds visible on the moors, like those seen at Grimsay and particularly amongst the hills of South Uist, suggest that the ratio between each type of location needs to be revised. This would of course impact upon the way that we currently view moorland wheelhouses - i.e. as something unusual, unique or specialised. Another main distinction concerning wheelhouses is that some structures have piers connecting with the outer wall, and others have free standing piers which create an aisle between the wall and pier, giving rise to the name aisled roundhouse. Armit (1990, 61) would claim that this latter distinction is an unnecessary typological division, however it remains possible that such a division may be indicative of the internal organisation of space and may be helpful in addressing problems concerning the development of the structural phases once sufficient reliable dating evidence is accumulated. Previously, wheelhouses have been dated by the analysis of pottery, which has since been shown to be a particularly unreliable method; from the presence of imported items such as glass beads and more recently by radiocarbon dating.
Ian Armit has suggested a period of before the first century BC based on the evidence from Hornish Point (1992, 68-9), whereas Ewan Campbell on the basis of radiocarbon dates and Roman items at Sollas B, suggests a much later second century AD date (1991, 139). Whether the construction of wheelhouses can be pinpointed to a specific date or general time period remains to be shown as much of the data is contradictory or hindered by sub-standard excavations in the early 20th century.
The number of mounds visible on the moors, like those seen at Grimsay and particularly amongst the hills of South Uist, suggest that the ratio between each type of location needs to be revised. This would of course impact upon the way that we currently view moorland wheelhouses - i.e. as something unusual, unique or specialised. Another main distinction concerning wheelhouses is that some structures have piers connecting with the outer wall, and others have free standing piers which create an aisle between the wall and pier, giving rise to the name aisled roundhouse. Armit (1990, 61) would claim that this latter distinction is an unnecessary typological division, however it remains possible that such a division may be indicative of the internal organisation of space and may be helpful in addressing problems concerning the development of the structural phases once sufficient reliable dating evidence is accumulated. Previously, wheelhouses have been dated by the analysis of pottery, which has since been shown to be a particularly unreliable method; from the presence of imported items such as glass beads and more recently by radiocarbon dating.
Ian Armit has suggested a period of before the first century BC based on the evidence from Hornish Point (1992, 68-9), whereas Ewan Campbell on the basis of radiocarbon dates and Roman items at Sollas B, suggests a much later second century AD date (1991, 139). Whether the construction of wheelhouses can be pinpointed to a specific date or general time period remains to be shown as much of the data is contradictory or hindered by sub-standard excavations in the early 20th century.
Prehistoric and Early Medieval Settlement...
View from Usinish wheelhouse looking towards Beinn Mhor
Our understanding of prehistoric settlement on the machair is somewhat fragmentary as intensive surveys have failed to identify middle Bronze Age settlements (Parker Pearson et al forthcoming). The suggestion has been made that middle Bronze Age settlements could be found directly under later Bronze Age settlements, however, only the wheelhouse at Cill Donnain (Zvelebil 1991) has supported this claim to an extent (Parker Pearson et al forthcoming). Further, the problems in locating late Bronze Age and early Iron Age settlements prior to recent surveys caused great difficulties in understanding the development of broch and wheelhouse architecture, lending weight to migration theories (Scott 1948; MacKie 1965). In part, the problem was that brochs and wheelhouses appeared to have made a sudden arrival, and any continuity from earlier periods could not be investigated fully as these earlier sites had not been discovered or studied adequately. Currently, this question of origins regarding the internal or external development of new building traditions is benefiting from examination against a more complete archaeological record, although problems persist with both invasion and continuity theses. The Pictish period did not begin in the Western Isles until the seventh century AD, one to two centuries later than eastern and northern Scotland. This apparent later date has been given as no examples of Pictish material culture have been found in the Western Isles before a seventh century AD date (Parker Pearson et al 2004, 106). Therefore, in the Western Isles the fourth-sixth centuries AD are referred to as the Late Iron Age I with the seventh–eighth centuries called the Late Iron Age II or Pictish period. The pre-Pictish Late Iron Age I in the Western Isles is unlikely to be an indication of the archaeological visibility of any Pictish material, especially when acknowledging the extensive excavation and survey that has occurred in this region of Scotland, but more a reflection of the political and social relationships between the Western Isles and their neighbours during this period.
Re-use of earlier settlements during the Late Iron Age I and later has been shown through excavation at Dun Vulan (Parker Pearson et al 2004, 106). The re-use of earlier buildings during this period can also be detected at other sites such as Dun Bharabhat, Clettraval and Bagh nam Feadag. This trend is something that is now recognised and examined in depth, in contrast to many of the wheelhouse sites in the Western Isles where the recognition of such later occupation has not previously been made: ‘The inhabitants did not disappear; rather, they reused the villages and houses of their predecessors, and are therefore difficult for us to identify’ (ibid 109). With a number of wheelhouses showing evidence for later occupation it should be considered that the majority of middle Iron Age buildings were re-used, including those yet to be excavated, such as remote island duns and currently uninhabited areas such as Usinish and Ronay.
Re-use of earlier settlements during the Late Iron Age I and later has been shown through excavation at Dun Vulan (Parker Pearson et al 2004, 106). The re-use of earlier buildings during this period can also be detected at other sites such as Dun Bharabhat, Clettraval and Bagh nam Feadag. This trend is something that is now recognised and examined in depth, in contrast to many of the wheelhouse sites in the Western Isles where the recognition of such later occupation has not previously been made: ‘The inhabitants did not disappear; rather, they reused the villages and houses of their predecessors, and are therefore difficult for us to identify’ (ibid 109). With a number of wheelhouses showing evidence for later occupation it should be considered that the majority of middle Iron Age buildings were re-used, including those yet to be excavated, such as remote island duns and currently uninhabited areas such as Usinish and Ronay.
Norse and Medieval Settlement...
Norse settlement in the Western Isles is currently less understood than many other periods, primarily because so few examples from this period have been found or studied. What is clear from those sites that are known is that they occur either immediately adjacent to or on top of earlier settlements, with only a couple of exceptions (e.g. Cille Pheadair) being located in apparent isolation. Such relationships have led archaeologists to view these relationships as strong evidence for continuity from the Pictish period (see Parker Pearson et al 2004, 130). With continuity taking place at many earlier sites in the Norse period, another development has been discovered with the excavation of substantial mounds on the Bornais machair. The Norse settlement here has developed on top of a Pictish settlement which is dated to the seventh and eighth centuries (ibid 133). The buildings here began with a longhouse-like building utilising large timber posts, a building technique not used since the Bronze Age in the Western Isles (ibid 133) and very different from that which had gone immediately before – often subterranean, revetted and dry-stone in nature. Excavation at Bornais has shown how the Norse period settlement had expanded to the point where it became potentially the largest settlement in the region that retained its importance some time into the period of Scottish rule, operating in some form until the fifteenth century (Sharples & Parker Pearson 1999, 41-62).
The medieval settlement landscape of the Outer Hebrides differs vastly to that which had gone before. At some point around the fourteenth century the machair was abandoned and the focus of settlements turned to the transitional zone between the machair and moorland and further inland to the east (Parker Pearson et al forthcoming). Supporting evidence can be seen at Bornais and Cille Pheadair for this abandonment as occupation appears to cease sometime in the thirteenth or fourteenth century AD (Parker Pearson & Sharples 1999; Sharples 1999, 30). Another indication of machair abandonment is the lack of finer vessels with stabbed decoration, a common type found dating to after the Norse period (Parker Pearson et al forthcoming). Of the sites on the machair that have produced Norse ceramics, only the settlement at Udal has so far produced this type of vessel (Crawford 1986) and would appear to be an exception (Parker Pearson et al forthcoming). In 1266 the Western Isles were subjugated by the Scottish Crown through the Treaty of Perth. This treaty specified that those within the territory at this time were free to leave or if they remained would become subject to Scottish rule. It is not known to a great extent how people reacted to this development and settlement changes specifically associated with this transfer of power have not been forthcoming (Parker Pearson et al 2004, 145-7). For example, excavations at Bornais, a settlement spanning the periods before and after the Treaty of Perth, do not show any decisive changes around 1266. Three houses were built on the mounds at Bornais in the fourteenth century while new settlement began 600m away to the east (ibid 161). By the time Pont and Blau’s maps were made (1590s and 1664) settlements appear to be situated east of the machair zone in South Uist. Two possible explanations for this shift in settlement patterns have been offered by Parker Pearson and Sharples, the first being that 'the dislocation was caused by the climatic deterioration known to occur during this period' (Parker Pearson & Sharples 2003). The suggestion argues that the succession of wet summers in this period led to a reduction in cereal production which prompted the cultivation of larger areas, something that is thought to cause instability and expose the machair to catastrophic sand movements. Storms, from later periods (1690s) are known to have moved large quantities of sand, engulfing settlements (Armit 1996, 229). Thus, settlement moved to the transitional zone where settlements could be constructed directly on the underlying bedrock and avoid the susceptibility of the changing machair landscape.
Developments, such as catastrophic sand blows, should be visible in the archaeological record at sites such as Bornais had they occurred, however, this is not the case. To complicate matters further, Armit has argued that the opposite occurred in North Uist – with settlement moving from the moorland to the machair around this time. What is clear is that further research into the circumstances of these settlement movements is desirable. A regional approach to this problem should also be considered as such sudden events as sand blows would presumably have been more than local in their influence and impact (Armit 1996, 229). However, we should bear in mind that the response of human settlement to environmental conditions can not be oversimplified as, clearly in some cases such as the exposed wheelhouse at Clettraval, other factors were more important. The second explanation for such a transition could be attributed to the changing political circumstances of the period. Over four centuries, the ownership of the Uists passed from the kings of Norway to the kings of Man, the Clan Ruairi, the Lord of the Isles and finally to the Clan Ranald. It is unclear the extent to which these political movements affected settlement during these turbulent centuries. As stated by Parker Pearson amongst others, ‘as archaeologists, we still know more about prehistoric life on Uist than we do about medieval houses and villages, a situation that can only begin to be rectified by large scale excavation of a medieval settlement’ (2004, 148).
In South Uist archaeologists have noted that settlement in the moorland tends to be smaller than their predecessors on the machair when the shift occurred (Parker Pearson et al 2004, 12). A change to beef farming that began during the Norse period could be viewed as a catalyst for this development, with the machair-moorland junction enabling access to both rough grazing and arable land. It should be noted that these settlements along the machair-moorland boundary can only exist where the geography permits, as in the case of Grimsay and the settlement at Bagh nam Feadag in particular, no machair exits (to any great extent) yet settlement spanned in some form from the Iron Age to the post medieval period. What we may be seeing in the Western Isles is a mixture of long established settlements with little dispersion alongside a more organised system of bailtean creation. The splitting and reorganisation of existing townships can be seen on Grimsay and elsewhere in the Western Isles, with placenames often containing a generic place name element of Gearraidh meaning home pasture or shieling (e.g. Gearraidh Dubh). Gearraidh Dubh, which is close to Bagh nam Feadag is similar to the Gearraidh townships in South Uist in that it contains no machair or any settlement before the medieval period, yet has earlier settlement nearby.
The medieval settlement landscape of the Outer Hebrides differs vastly to that which had gone before. At some point around the fourteenth century the machair was abandoned and the focus of settlements turned to the transitional zone between the machair and moorland and further inland to the east (Parker Pearson et al forthcoming). Supporting evidence can be seen at Bornais and Cille Pheadair for this abandonment as occupation appears to cease sometime in the thirteenth or fourteenth century AD (Parker Pearson & Sharples 1999; Sharples 1999, 30). Another indication of machair abandonment is the lack of finer vessels with stabbed decoration, a common type found dating to after the Norse period (Parker Pearson et al forthcoming). Of the sites on the machair that have produced Norse ceramics, only the settlement at Udal has so far produced this type of vessel (Crawford 1986) and would appear to be an exception (Parker Pearson et al forthcoming). In 1266 the Western Isles were subjugated by the Scottish Crown through the Treaty of Perth. This treaty specified that those within the territory at this time were free to leave or if they remained would become subject to Scottish rule. It is not known to a great extent how people reacted to this development and settlement changes specifically associated with this transfer of power have not been forthcoming (Parker Pearson et al 2004, 145-7). For example, excavations at Bornais, a settlement spanning the periods before and after the Treaty of Perth, do not show any decisive changes around 1266. Three houses were built on the mounds at Bornais in the fourteenth century while new settlement began 600m away to the east (ibid 161). By the time Pont and Blau’s maps were made (1590s and 1664) settlements appear to be situated east of the machair zone in South Uist. Two possible explanations for this shift in settlement patterns have been offered by Parker Pearson and Sharples, the first being that 'the dislocation was caused by the climatic deterioration known to occur during this period' (Parker Pearson & Sharples 2003). The suggestion argues that the succession of wet summers in this period led to a reduction in cereal production which prompted the cultivation of larger areas, something that is thought to cause instability and expose the machair to catastrophic sand movements. Storms, from later periods (1690s) are known to have moved large quantities of sand, engulfing settlements (Armit 1996, 229). Thus, settlement moved to the transitional zone where settlements could be constructed directly on the underlying bedrock and avoid the susceptibility of the changing machair landscape.
Developments, such as catastrophic sand blows, should be visible in the archaeological record at sites such as Bornais had they occurred, however, this is not the case. To complicate matters further, Armit has argued that the opposite occurred in North Uist – with settlement moving from the moorland to the machair around this time. What is clear is that further research into the circumstances of these settlement movements is desirable. A regional approach to this problem should also be considered as such sudden events as sand blows would presumably have been more than local in their influence and impact (Armit 1996, 229). However, we should bear in mind that the response of human settlement to environmental conditions can not be oversimplified as, clearly in some cases such as the exposed wheelhouse at Clettraval, other factors were more important. The second explanation for such a transition could be attributed to the changing political circumstances of the period. Over four centuries, the ownership of the Uists passed from the kings of Norway to the kings of Man, the Clan Ruairi, the Lord of the Isles and finally to the Clan Ranald. It is unclear the extent to which these political movements affected settlement during these turbulent centuries. As stated by Parker Pearson amongst others, ‘as archaeologists, we still know more about prehistoric life on Uist than we do about medieval houses and villages, a situation that can only begin to be rectified by large scale excavation of a medieval settlement’ (2004, 148).
In South Uist archaeologists have noted that settlement in the moorland tends to be smaller than their predecessors on the machair when the shift occurred (Parker Pearson et al 2004, 12). A change to beef farming that began during the Norse period could be viewed as a catalyst for this development, with the machair-moorland junction enabling access to both rough grazing and arable land. It should be noted that these settlements along the machair-moorland boundary can only exist where the geography permits, as in the case of Grimsay and the settlement at Bagh nam Feadag in particular, no machair exits (to any great extent) yet settlement spanned in some form from the Iron Age to the post medieval period. What we may be seeing in the Western Isles is a mixture of long established settlements with little dispersion alongside a more organised system of bailtean creation. The splitting and reorganisation of existing townships can be seen on Grimsay and elsewhere in the Western Isles, with placenames often containing a generic place name element of Gearraidh meaning home pasture or shieling (e.g. Gearraidh Dubh). Gearraidh Dubh, which is close to Bagh nam Feadag is similar to the Gearraidh townships in South Uist in that it contains no machair or any settlement before the medieval period, yet has earlier settlement nearby.