Underwater excavations at Ederline Boathouse Crannog, Loch Awe, Argyll
Principal Investigator: Jon Henderson (University of Nottingham)
Underwater excavations have been in progress on the submerged crannog at Ederline Boathouse in Loch Awe since 2004. Despite a previous 4th century BC radiocarbon determination from a structural timber from the surface of the site, these new excavations have also uncovered deposits dating from the fourth to sixth centuries AD. The results from the first intrusive investigations into a Loch Awe crannog demonstrate that crannog sites are as complex as they were long-lived.
Introduction and background
Fig 01: The crannog appears as a massive boulder
mound overlying this reef at its south western end
The crannog at Ederline Boathouse in Loch Awe was examined by McArdle and McArdle (1973) as part of the first complete underwater crannog survey of any loch in Scotland, and was further described by Morrison (1982). The site itself is located at the south-western end of Loch Awe and is visible in the summer months above water as a low grass covered stony mound, approximately 4m in diameter.
This dry area represents only a small fraction of the true extent of the site, as the main mound is submerged and measures 37m by 27m and rises over 2.5m from base to top. Ederline is unusual in that it is one of a minority of crannogs which utilise a natural bedrock outcrop as a foundation; at Ederline this consists of an elongated reef running south-west to north-east for approximately 80m in 3m to 4m of water.
There are numerous structural timbers (horizontals and piles) protruding from the site, principally on the surface of the crannog in less than 1m of water. At the time of the survey by McArdle and McArdle in 1972 the water level of Loch Awe was extremely low due to a dry summer allowing the inspection of the flat surface of the site above water. A large number of oak structural timbers were recorded at this time, as well as two saddle querns and a rotary quern, all of which were left in situ on the site (McArdle & McArdle 1973, 9). In the early 1980's the site was revisited and one of the oak vertical piles from the surface of the site was sampled and provided a radiocarbon determination of 370 bc ± 45 (UB-2415), calibrating at 400-190 BC (Morrison 1982; Dixon pers. comm.).
In 2004 a two week trial underwater excavation was carried out to gather much-needed taphonomic data for a boulder mound crannog and to identify whether significant organic deposits were preserved on the site (Cavers & Henderson 2005). A trench, measuring some 3x5m, was opened on the northern side of the site, in approximately 3m of water, and excavated to loch bed level. Around 1m of well preserved organic deposits was encountered, along with numerous structural timbers and a large animal bone assemblage. Despite the radiocarbon determination from the 4th to the 2nd century BC obtained from the timber on the top of the site, two sherds of E-ware were recovered from a sealed organic context on the side of the site suggesting an Early Historic date for this deposit.
Fig 02: A trench, measuring some 3x5m, was opened on the northern side of the site, in approximately 3m of water, and excavated to loch bed level.
Five radiocarbon dates were recovered from bone and timber samples from the two clear organic deposits (Contexts 103 and 105) excavated in 2004 which were clearly separated by an inorganic silt layer (Context 104;). Three samples from Context 103, which produced the E-ware, provided dates calibrating to the fourth to sixth century AD (1505 ± 35 BP, SUERC-11462; 1575 ± 35 BP, SUERC-11465; 1535 ± 35 BP, SUERC-11469). The lower organic level, Context 105, produced two earlier dates calibrating to 171 BC - 21 AD and 405-209 BC respectively (2055 ± 35 BP, SUERC-11463; 2285 ± 35 BP, SUERC-11464). Archaeobotanical analysis of these deposits suggested that a good deal of the plant material was structural and/or fuel derived in origin (Bogaard & Hall 2005). Some crop material was also recovered suggesting that both processing (chaff) and consumption (wheat/rye bran) was taking place on the site.
The site appears as a boulder mound measuring c.37m by 27m and rises over 2.5m from base to top (see Figure 1). The majority of the site is submerged throughout the year - in the summer months a low stony mound, approximately 4m in diameter, is visible on which there is a small (less than 2 m high) tree growing. The deepest part of the site lies at 3m and is located at the natural bedrock outcrop on the northern margins of the site.
A boulder layer overlies the entire site, mainly consisting of large (c.1m) and medium (0.3-1m) stones, generally rounded and sub-angular and on average 3-4 stones deep usually equating to c. 1m depth of covering on deposits excavated. The boulder capping is referred to as Context 100 in Trench 1 and 200 in Trench 2.
Trench 1
The 3x5 m trench laid out in 2004 was extended 2 m into the crannog mound. The primary aim of Trench 1 was to uncover the original foundations of the crannog to provide information on how the surviving mound was formed. The boulder layer (100) was found to overlie a loosely compacted layer of naturally sorted small stones (101) and a layer of white grey inorganic sand (102) - all three of these contexts covered the whole trench.
Fig 03: The section exposed in 2004 was recovered and worked back into the crannog mound.
The section exposed in 2004 was recovered and worked back into the crannog mound (see Figure 2 and 3 above). The rich Early Historic and Iron Age organic levels (Contexts 103 and 105) were found to run through the entire trench (see Figure 2; Plate 1) with the layer of inorganic white-grey silt (104) separating them throughout. Context 105 is interpreted as a discard deposit from the main mound - it consists of moderately compact red-brown organic silt and contains wood-chips, bracken, hazel nuts, twigs, bone and charcoal. The deposits averaged 15-20 cm in thickness throughout the trench. Context 104 is interpreted as a possible abandonment phase between contexts 103 and 105 as it is inorganic silt similar to context 102. Context 103 is similar to 105 - it consists of moderately compact red-brown organic silt and contains wood-chips, well-preserved bracken, hazel nuts, twigs, and large amounts of charcoal and animal bone - and is also interpreted as a discard deposit. It ranged from 20-30 cm in thickness. Both 103 and 105 contained numerous pieces of worked wood, ranging from large timbers to worked points and woodchips. There were significant amounts of charcoal and burnt bone recovered from both deposits and several timbers were seen to be superficially charred and had clearly been close to fire prior to their deposition. In general, Context 103 produced more bone, charcoal and burnt timber than 105 - it was this context that produced the two sherds of E-ware in 2004 and the bottom stone of a rotary quern during the 2007 season (see Plates 2 and 3).
Although the organic deposits became thicker as they approached the main crannog mound, no in-situ structural timbers or deposits which could be said to indicate the basal construction levels of the crannog mound were identified. In particular, the lack of any upright piles in this trench is in marked contrast to the evidence from Oakbank Crannog and Milton Loch where surrounding piles have been identified, and variously interpreted as supports for a platform or palisade structure. In the eastern half of the trench closer to the main mound, there was considerable evidence for burning. Context 107 is a dump deposit, some 50 cm thick, consisting entirely of charcoal pieces ranging in size from 2-15cm. There were spaces between the charcoal fragments suggesting they had been deposited in the water in one event - perhaps after a conflagration on the main mound. This deposit was sealed by Contexts 102 and 101. Two further contexts were identified: Contexts 108 and 109 appeared as grey inorganic deposits containing small stones and gravel.
Trench 2
A trench measuring 7x2 m was opened on the surface if the mound in an effort to identify in situ occupation deposits (Plate 6). Under a 70 to 90cm capping of boulders measuring 0.3 to 1m in diameter (Context 200) and naturally sorted angular stones (Context 201), a mixed deposit (202) consisting of inorganic grey sandy silt, gravel and small stones was encountered throughout the whole trench which overlay over 30 in situ horizontal alder timbers and three oak uprights (Plate 4 and 5).
Fig 04: The timber layer was more fully exposed in the north-eastern end of the trench.
There were no finds from Context 202 but it contained small flecks (less than 0.5 cm in diameter) of charcoal and burnt bone throughout, suggesting the deposit had been heavily re-worked and mixed over time presumably by water. Only two timbers displayed evidence of cut marks and none could be considered to represent floor or building divisions. The timber layer was more fully exposed in the north-eastern end of the trench (Figure 4) and consisted of horizontal alder timbers laid in layers at roughly right angles to each other (Plate 5). These are interpreted as the structural layers of a packwerk mound. Two discrete deposits of comminuted plant material, twigs and bracken were identified in the trench (Contexts 203 and 204) and may represent the fragmentary remains of flooring material - both were sampled in their entirety for archaeobotanical analysis.
Discussion
Trench 1 did not hit deposits relating to the original construction of the mound and it must be concluded that these lie further into the crannog mound. The deposits which were present most likely represent discard from occupation taking place higher up on the main mound. Despite the environmental richness of the deposits and the recovery of significant amounts of animal bone, these deposits are remarkably devoid of other finds - although the small size of the evaluatory trench must be kept in mind. Woodchips, hazelnuts, a rotary quern stone, and the two sherds of E-ware and morticed timber recovered in 2004, aside, there have been no other finds of note from the organic deposits in Trench 1.
Trench 2 on the upper surface of the mound failed to produce definite in situ floor deposits or timber architectural features. The deposits encountered, consisting of horizontal timbers within an organic matrix, are consistent with the expected characteristics of 'packwerk' foundation deposits, where timbers are laid in layers to create an artificial mound (see Crone et al 2001). As a result these deposits are likely to contain material related to construction of the site, and not to its domestic occupation. It therefore remains a possibility that the occupation on top of the mound has been eroded away, leaving only the upper log platform structural layers of a 'packwerk' type construction.
The recovery of very small pieces of burnt bone and charcoal throughout the matrix of Context 202, which overlay the 'packwerk' deposits, supports this view. It is possible that that this mixed deposit represents the eroded remains of domestic occupation on the site. However, two intact deposits of organic material (Contexts 203 and 204) were recovered which may relate to the original floor levels of the crannog and will be subject to archaeobotanical analysis. The small size of the evaluatory trench must once again be kept in mind as occupation deposits may exists elsewhere on the mound. Equally, excavation in Trench 2 did not progress beyond the layer of timbers encountered while the layout of the timbers and three piles identified in the trench may be more easily understood with the excavation of a larger area.
Future work
The evidence recovered from both trenches demonstrate that there is a high degree of organic preservation on the site and that it is therefore likely that occupation deposits, or at least richer deposits relating to occupation, survive on the site. The identification of 'occupation' versus 'construction' deposits in crannog excavations remains a problem in interpretation. Samples were taken of all contexts encountered during the excavation and it is recommended that these be subject to archaeobotanical analysis in an effort to determine their mode of deposition and whether or not they can be said to relate to occupation or construction. In particular the analysis of the possible floor level material from Trench 2 (Contexts 203 and 204) compared to material from the 'packwerk' matrix of the mound will be of interest. With the exception of Buiston (Crone 2000) few crannogs have undergone systematic studies of their stratified organic deposits.
Significant bone assemblages were recovered from Contexts 103 and 105 and it is recommended these be subject to a specialist study. Substantial bone assemblages of later prehistoric and Early Historic date are relatively rare and Ederline provides the opportunity to study changes in diet and meat supply over this period.
The recovery in the 1980's of an early Iron Age date, calibrated to 400-190 BC, from the top of the mound may indicate that Early Historic occupation has been completely eroded from the surface of the mound and now only survives along the basal margins of the site in the form of deliberate discard or material deposited naturally as part of the erosive process. Timbers have been sampled from Trench 2 which can be submitted for radiocarbon dating to test this theory. The dates recovered will also be an indication as to whether domestic occupation is likely to survive on the surface of the mound.
The trial excavations in 2006 and 2007 have not penetrated the organic deposits at the site to any great extent. Even so the information obtained so far has been extremely informative in the context of crannog studies and challenges the validity of dating these sites from just one radiocarbon determination obtained from one part of the site. To date no crannog that has been subject to any level of excavation has provided evidence of single phase occupation. Further excavation at the site clearly has the potential to be extremely informative, not just in terms of identifying the processes of crannog construction and taphonomy, but also in the way in which crannogs developed as settlements through the Iron Age and Early Historic periods. More importantly the probable existence of Early Historic activity at Ederline has a wider significance for Scottish archaeology. The lack of domestic settlement in Argyll relating to the Dál Riadic period has been commented by other authors (Campbell 2001).
Given the evidence from Ederline and Loch Glashan (Crone & Campbell 2005) it is possible that this evidence may be present in the ubiquitous but largely unexplored crannogs of Argyll which would not only fill out the picture of the Early Historic period in western Scotland but could begin to answer the extent of Irish influence in the postulated Dál Riadic incursions. Crannogs offer the closest known parallels to Irish settlement forms in Scotland but this link has never been fully investigated. Such investigations would have obvious implications on current perceptions of Scottish identity especially as regards the impact of projected Dál Riadic incursions on indigenous populations.