Llangorse Crannog: The Excavation of an Early Medieval Royal Site in the Kingdom of Brycheiniog

By Alan Lane and Mark Redknap

The crannog on Llangorse Lake near Brecon in mid Wales was discovered in 1867 and first excavated in 1869 by two local antiquaries, Edgar and Henry Dumbleton, who published their findings over the next four years. In 1988 dendrochronological dates from submerged palisade planks established its construction in the ninth century, and a combined off- and on-shore investigation of the site was started as a joint project between Cardiff University and Amgueddfa Cymru - National Museum Wales. The subsequent surveys and excavation (1989-1994, 2004) resulted in the recovery of a remarkable time capsule of life in the late ninth and tenth century, on the only crannog yet identified in Wales.

This publication re-examines the early investigations, describes in detail the anatomy of the crannog mound and its construction, and the material culture found. The crannog’s treasures include early medieval secular and religious metalwork, evidence for manufacture, the largest depository of early medieval carpentry in Wales and a remarkable richly embroidered silk and linen textile which is fully analysed and placed in context. The crannog’s place in Welsh history is explored, as a royal llys (‘court’) within the kingdom of Brycheiniog. Historical record indicates the site was destroyed in 916 by Aethelflaed, the Mercian queen, in the course of the Viking wars of the early tenth century. The subsequent significance of the crannog in local traditions and its post-medieval occupation during a riotous dispute in the reign Elizabeth I are also discussed. Two logboats from the vicinity of the crannog are analysed, and a replica described. The cultural affinities of the crannog and its material culture is assessed, as are their relationship to origin myths for the kingdom, and to probable links with early medieval Ireland. The folk tales associated with the lake are explored, in a book that brings together archaeology, history, myths and legends, underwater and terrestrial archaeology.

• Language: English
• Publisher: Oxbow Books
• Release date: Feb 28, 2020
• ISBN: 9781789253078

Alan Lane

Dr Alan Lane, FSA, is a Senior Lecturer in Early Medieval Archaeology at Cardiff University. He specialises in the archaeology of the Celtic West and North and has carried out excavations and research in Wales and Scotland on high status settlement sites. He has worked on the Iron Age, post-Roman and Viking ceramic sequences of the Hebrides and their use as site identification markers.

Book preview

SECTION 1

Introduction and Setting

Fig. 1.1 Llangorse Lake lies to the east of the medieval town of Brecon, within the Brecon Beacons National Park and is surrounded by three mountain ranges: the Brecon Beacons to the south, the Black Mountains to the east and Mynydd Eppynt to the north.

CHAPTER 1

The Crannog, its Name and its Setting

1.1 SITE LOCATION

Llangorse crannog lies on the north side of Llangorse Lake/Llyn Syfaddan, about 40 metres from its present, modified, northern shore (OS Grid Ref. SO132 265). Llangorse is the largest natural lake in south Wales, situated to the east of Brecon the old county town for Brecknockshire (now part of the larger county of Powys). It lies within the Brecon Beacons National Park, an area of outstanding natural beauty (Figs 1.1, 1.3).

1.2 THE NAME OF THE LAKE AND THE NAME OF THE CRANNOG

by David N. Parsons & Alan Lane

Earlier name-forms of what is currently, in English, called Llangorse Lake are listed and discussed by Peter Powell 1986–87, Morgan & Powell 1999, 138, and Owen & Morgan 2007, 260. The earliest documented form is Brecenanmere, which appears in the ‘Mercian Register’, a tenthcentury collection of annals incorporated into two manuscripts of the Anglo-Saxon Chronicle, Manuscript B of the late tenth century and Manuscript C of the mid-eleventh century (see below; Taylor 1983, 50; O’Brien O’Keeffe 2001, 75–76; Sims-Williams 1993, 59–60). It appears under the year ‘916’ (sub anno 919) in an account of a Mercian raid into Wales which ‘destroyed Brecenanmere and captured the king’s wife and 33 other persons’. The name is Old English and means ‘mere [= lake] of Brycheiniog [or possibly ‘of Brychan’, the kingdom’s eponymous founder]’, although the actual object of the attack was presumably the royal seat on the crannog rather than the lake itself.

The Welsh name, Llyn Syfaddan, is first found, as linn Syuadon, in the Liber Landavensis, an early twelfth-century compilation based in part on earlier materials (Evans & Rhŷs 1893, 146). The lake is mentioned in the boundary-clause to a charter (no. 146) which is theoretically of early eighth-century date, but is evidently a later fabrication; it is unclear how much earlier than the date of the manuscript’s compilation the boundaries may be (Sims-Williams 1993, 51–53; Davies 1979, 98), although Coe (2004, 38–40) makes a good, if properly cautious, case for the early eleventh century (see below, Chapter 21). The name combines Welsh llyn ‘lake’ with an element of unknown meaning; Syfaddan may possibly be a personal name (Rhŷs 1901, i, 74; Powell 1986–87, 41; Owen & Morgan 2007, 260). Some alternative early speculations are detailed by Powell (1986–67, 40); a more recent proposal by Thomas (1994, 160 n. 40), that the name may be an ancient *Samo-ton(a), explained as a ‘divine female personification of the summer months … as the best fishing period’ is rejected by Morgan & Powell (1999, 138) on linguistic grounds. In fact, since the weight of early spellings tends to suggest Syf- rather than Saf-, then most detailed suggestions to date which have usually involved -a- in the first syllable are problematic.

These two early names, Welsh and English, enjoyed lasting currency. In the first half of the sixteenth century Leland notes Brechenauc mere … in Walche Llin Seuathan (Smith 1906, 10). The Welsh version is recorded regularly from the fifteenth century (eg llyn syuadon in the work of the poet Lewis Glyn Cothi; Jones 1953, 181) into the modern period (eg Llyn Safaddu on the 1832 OS 1″ map; Fig. 1.2). The English mere is found from the thirteenth century onwards, although more often on its own than qualified by Brecon or Brycheiniog (eg a fishery in la mere 1299 CIPM iii, 426). Other records contain mara, probably as an approximate Latinisation of mere rather than with the Latin sense of ‘marsh, swamp’, although since this is also one sense of Welsh cors, in Llangorse (see below), a deliberate translation is conceivable. Note also that Mara was sometimes used, between the twelfth and eighteenth centuries, as an alias for Llangorse, sometimes as ‘Mara Moto’, ‘Mara Blaen Llevenye’ (Morgan & Powell 1999, 106).

The name Llangors, denoting the settlement lying 0.5 km to the north, is used as a qualifier for the lake from at least as early as 1689: Mara langors Glamorgan Archives MS CL I/1473 (ex. information Richard Morgan). Llangors Pool then appears on various eighteenth- and nineteenthcentury maps. Theophilus Jones marks Llangorse Mere on his 1805 map, which may be the first use of the nineteenth- and twentieth-century spelling Llangorse. Dumbleton and the Woolhope authors refer to Llangorse Lake or the Lake of Llangorse (Dumbleton, E. N. 1870b; Lloyd 1870). It may be noted that Brycheiniog and Llangors were not the only local place-names to be used as qualifiers for the lake: le Mere of Blaynleveny 1352 (CIPM viii, 521) makes reference to Blaenllynfi, a medieval castle some 3 km south of the lake.

The name Llangors itself contains a further term for the lake or its surroundings (Morgan & Powell 1999, 106; Owen & Morgan 2007, 260). In early Welsh cors meant both ‘reeds’ and the terrain in which they are generally found, ‘swamp, bog’. Either of these may have been the original sense in Llangors, the ‘church-enclosure’ by the cors, another name first recorded in the Liber Landavensis, this time in a more reliable document of apparent early tenth-century date (see below; charter no. 237b; Davies, W. 1979, 124; Sims-Williams 1993, 61–62). The characteristic reeds are probably alluded to also by Gerald of Wales at the end of the twelfth century. As transmitted, his text reads (Dimock 1868, 33): lacus ille de Brecheneiauc… quem et Clamosum dicunt, ‘that lake of Brycheiniog … which they also call Clamosus’. Latin Clamosus should mean ‘noisy’, giving rise to explanations about the sound of the ice groaning on the frozen lake; but it is likely that Gerald actually wrote Calamosus, ‘reedy, full of reeds’ (Jones 1940; Morgan & Powell 1999, 138). On this interpretation of Gerald’s evidence it is perhaps possible that the lake bore an alternative early Welsh name derived from cors, and that it is in fact the lake-name which enters into the place-name Llangors.

Fig. 1.2 Detail showing Llangorse and the crannog from the first series Ordnance Survey 1" map of 1832. (© Crown Copyright: Ordnance Survey)

Fig. 1.3 Aerial view from the north-west showing the lake with the crannog close to the northern shore in the centre foreground. (© Crown Copyright: RCAHMW)

There seems to be no record of a name for the crannog itself before the sixteenth-century court records cited by Richardson (2007, 115–21), which refer to it as twmp ‘mound’ (see below, Chapter 23). This may be compared with Bwlc, which appears on the Ordnance Survey 1" map of 1832’ (Fig. 1.2). This is a Welsh borrowing from English bulk, recorded in the sense ‘heap’ (Geiriadur Prifysgol Cymru cyfrol 1, 352), which is perhaps the intention here. Such a late English loan-word is hardly an ancient name, however. One earlier record, a fifteenth-century gloss to linn Syuadon in the Liber Landavensis, incorporates ynys ‘island’ (Evans & Rhŷs 1893, 340); however, aqua vocetur llinn ynys yvavon ‘water called the lake of Ynys Yvavon’ almost certainly represents a garbling of Llyn Syfaddan rather than an alternative early name for the lake based on a lost name of the island or crannog.

The earlier publications of our work on the crannog used the twentieth-century English spelling Llangorse which appeared on Ordnance Survey maps, in the village and on road signs. Some subsequent publications used Llangors. Current Ordnance Survey usage is Llangors for the village and Llangorse Lake for the lake. Several modern authorities prefer Llan-gors (Richards 1969; Powell 1986–87, 40; Davies, W. 1989; Owen & Morgan 2007, 260), the hyphen indicating that the stress falls on the second element of the name. There is of course a Welsh name for the lake – Llyn Syfaddan – so it might be appropriate to adopt a two language usage for the site and call it the Llangorse Lake/Llyn Syfaddan crannog. However, as Llyn Syfaddan does not seem to be a significant surviving local usage, we have adopted the English usage Llangorse for the lake and for the crannog and the Welsh spelling Llan-gors for the village.

1.3 THE CRANNOG TODAY

The modern appearance of the crannog is as a small wooded island (SO 12892690), measuring about 40 metres by 30 metres in size (Fig. 1.4). In summer the dense bush and tree vegetation (mainly alder) conceals the surface of the island, but when the vegetation dies down or is cleared, it is apparent that the island is a stony mound rising to about 0.8m above summer water level, made up of medium-sized and smaller stones and occasional boulders. Prior to the 2004 conservation measures, the edge of the island was clear of vegetation to the south, south-east and west, and eroding timber and stone deposits were apparent. To the northeast silt deposits obscured the edge of the island which gently sloped into the lake, while to the north a reed swamp makes definition of an edge fairly arbitrary. When the current work began on the island in 1988, oak timbers were visible round the southern and western edges of the island in two partial lines, some timbers rising out of the water in dry summers and submerged in wet winters (Fig. 1.5). Local recording since 2006 indicates that the water level fluctuates considerably with as much as a 1.50m difference between summer and winter water levels (Garnet Davies, pers. comm.; Fig. 1.6). Occasional larger roundwood timbers were also visible as well as eroding wattle deposits in some places. The mound is sharply eroded on the west and south sides, and shelves into about 1.2m of water (low summer lake level). To the north the reed swamp gives way to a shallow channel, 40 metres wide, between it and the modern jetties on the north shore of the lake (Fig. 1.7). This shoreline has been heavily landscaped for outdoor watersports activities with a children’s adventure camp being constructed in the mid-1960s (Cragg et al. 1980, 188). The original shoreline, which may have been much more heavily reeded and more graduated in depth in the past, was originally set further to the north.

Fig. 1.4 The crannog in summer 2004, overgrown with trees and bushes.

Fig. 1.5 Oak palisade planks 602–609 protruding above water level on the south side of the crannog (summer 1990), in front of platform rubble.

Fig. 1.6 The flooded crannog in the winter (November 1988).

1.4 LOCATION, ENVIRONMENT AND LAND-USE

The lake is situated in a low-lying basin between the catchments of the rivers Usk and Wye, two of the major rivers which drain the hills of mid-Wales (Fig. 1.1). The Usk flows from its source in Carmarthenshire eastwards between the Brecon Beacons and Black Mountains. Its valley provides a major east-west routeway into Wales with the hill massif of the Brecon Beacons rising over 800m beyond it to the south. Both Roman roads and modern highways emphasise the natural importance of this route into Wales and through it to the west coast of Britain. To the east the Black Mountains (over 700m in height) separate the Usk valley from the Wye basin while low hills to the north of the lake allow an easy passage along the river Llynfi down to the Wye valley at Talgarth and Glasbury. To the north-west lower hills rise up into the Eppynt Plateau (over 400m) and beyond to the Cambrian Mountains of mid-Wales (RCAHMW 1997, 1–4).

Fig. 1.7 Aerial view of the crannog July 1990, showing island, reed beds and modern developments on the north shore of the lake. (© Crown Copyright: RCAHMW)

Fig. 1.8 Llangorse Lake topography, showing the lake, crannog and rivers.

The lake is fed principally by the Afon Llynfi, which enters the lake from the south, and drains to the north (Fig. 1.8). The lake is shallow and nutrient-rich, with a surface area of roughly 139 ha. and a maximum depth of about 7–9m (Chambers 1999). At an altitude of 153m it is defined as a lowland context, although set within a wider upland environment (Chambers 1999, 344). The lake is overlooked by the sandstone scarps of the Brecon Beacons which rise to 872m (2,893ft), and the Black Mountains, which rise to 719m (2,363ft). Those hills in the immediate catchment of the lake rise to 380m and 508m. The landscape around Llangorse is dominated by sedimentary rocks. The Black Mountains and Brecon Beacons to the south are formed from a thick stack of variably coloured, and durable sandstones, whereas Llangorse and the country to the immediate north is more subdued. Softer marls underlying this area are more easily eroded, with more resilient interspersed sandstone layers producing a gently stepped terrain. These rocks are of Devonian age, formed around 400 million years ago, but have been modified to form the current landscape by Devensian glacial and fluvioglacial processes. Only the higher ground to the southeast and south-west of the lake escaped ice cover. Llangorse Lake is underlain by glacial till, overlain by glaciolacustrine deposits and flanked by alluvial fan scree.

Fig 1.9 Llangorse Lake (dark blue) showing depth and earlier, larger glacial lake.

The environment of the lake

Llangorse was first designated a Site of Special Scientific Interest (SSSI) in 1954. The SSSI citation focusses on its nutrient-rich character, high biodiversity and productivity (Duigan et al. 1999, 331). These lowland lake characteristics are extremely rare in a Welsh context. The fringing vegetation of the lake consists mainly of common reed (Phragmites communis), bulrush (Typha latifolia) and reed swamp associates such as fringed waterlily (Nymphoides peltata). The fish population is a typical eutrophic fauna including roach, perch, pike and eels (Duigan et al. 1999, 331).

As noted above the lake is a relatively recent feature of the landscape, a result of Late Quaternary glaciation. According to Chambers (1999) it is a relic of a much larger proglacial lake – ‘Lake Llangors’ which would have had a water level some 37m higher than the modern lake (Fig. 1.9). Palaeolimnological studies indicate the lake basin has a sediment sequence from basal rock to lacustrine clays below deposits of lake marl, further clay deposits, muds and further marls. Higher deposits of red-brown silty clay are thought to represent increased sedimentation related to Iron Age or later agriculture in the catchment area. The depth of the lake has altered in postglacial times with about 4m of silty clay deposited in some areas in the last 2000 years and infill since the end of glaciation of 10–12m in places. Modern infill has accelerated with 65cm apparently deposited in the last 150 years (Chambers 1999). The lake is fed from the south by the Afon Llynfi and five small seasonal streams on its northern and eastern shores. It is drained from the north side of the lake by the Llynfi feeding into the Wye. Much of the lake is fairly shallow – less than 3m – but with two deeper troughs where water exceeds 7m. Detailed interpretation of the glacial history of the lake has been disputed but the outline of its glacial origin seems clear (Chambers 1999, 348; Raikes 1986–87; Palmer et al. 2008).

Two cores were taken in conjunction with the excavation of the crannog, Llangorse I from within the deposits of Trench A and Llangorse II from beyond the crannog deposits. The two sequences were matched, but only Llangorse I contained the upper deposits including the peat surface on which the crannog was constructed. Unfortunately Llangorse I was not analysed for its pollen data as the researchers’ focus was on molluscs and ostracods. No radiocarbon dates were processed and the two sequences were dated by the pollen in Llangorse II. This indicated that the marl had been deposited between c. 10,000 and 8,000 BP and that the beginning of peat growth post-dates that horizon (Walker et al. 1993).

Pollen data from cores in the lake and near the crannog show the post-glacial vegetational history of the area with the colonisation of tree species from c. 9,500 or 9,000 BP giving rise to stable soils and close-canopy woodland. Human impact on the woodland may be traceable from c. 5000 BP and a major change in pollen and sediments is loosely dated to the Iron Age/Roman period with increased tree clearance and intensified agriculture in the catchment (Chambers 1999, 349–56; RCAHMW 1997, 5–7).

Local soils and land-use

As noted already Llangorse lies in a basin between higher upland areas. Areas of fine silt soils that are subject to seasonal waterlogging extend for around 200 metres to the north and to the south of the lake. The wider area of lowlands surrounding the lake is comprised almost entirely of well drained fine loam. The higher ground of Allt yr Esgair to the west and Mynydd Llangorse to the east is covered by well drained coarse loam soils (Ordnance Survey 1983).

These lowland loam soils are currently classed as amenable to high quality pasture or cereal cultivation, in contrast to the lowland silt soils and upland loam that are sufficient only to support poor quality grazing (Ordnance Survey 1983). Although the lowland soils may have arable potential, the majority of this region is under permanent pasture. Comparison of observed present day land use with Dudley Stamp’s 1931– 1935 Land Use Survey maps shows that, although the lower slopes of Allt yr Esgair and the occasional field amongst the lowland loam were under the plough in the 1930s (Land Utilisation Survey of Britain 1945), the practice of using the majority of the land in this region as pasture predates the influence of chemical fertilisers. Nevertheless both Gerald of Wales writing in the late twelfth century (Thorpe 1978, 93) and John Leland in the midsixteenth century (Smith 1906, 104) refer to rich arable productivity in Brycheiniog.

1.5 COMMUNICATIONS

There is no early medieval information about roads and trackways in the area. Consequently we can only speculate about routeways based on the local topography and the probability that Roman roads remained significant means of transport throughout the first millennium. Fortunately the impact of topography is fairly clear. The Usk valley provides an obvious avenue of access through the mountains of Brycheiniog from Abergavenny west through Brecon, to Llandovery in the Tywi Valley and then through Llandeilo to Carmarthen and beyond to the Irish Sea (cf. Figs 1.1 & 1.10). This follows the Roman roads and remains the route of the modern A40, the main trunk road from England through south Wales until the building of the M4 motorway. From Llangorse a modern minor road follows the Llynfi north-east and skirts the outliers of the Black Mountains to drop into the Wye Valley, which opens north-east into Herefordshire and England. Another route may be surmised following the A479 from Tretower to Talgarth and continuing northwest up the Wye and the A470 to Builth in mid-Wales.

Fig. 1.10 Roman road network in south Wales, showing location of significant Roman sites.

The Roman roads in the area have recently been reviewed and while the existence of roads between forts is not doubted the uncertainties about their precise lines have been emphasised. RR62a is known to run from Abergavenny west to Brecon and the Ordnance Survey mark it as visible at Bwlch, just south of Llangorse Lake. However Sylvester and Toller cite three or four possible routes, none of which are confirmed by archaeological remains (2010, 96, fig 4.3; Evans et al. 2010, 317). The Royal Commission dismiss two routes to the immediate south and east of the Lake (RCAHMW 1986, 167–69; cf. Margary 1973, 333–34).

CHAPTER 2

Discovery of the Crannog and Early Investigations

2.1 THE DISCOVERY OF THE CRANNOG

The crannog was discovered in 1867 and first excavated in 1869 by two local antiquaries, Edgar and Henry Dumbleton, who presented their discoveries to the British Association for the Advancement of Science which met at Exeter in that year (Dumbleton, E. N. 1870a, 130). The crannog was first recognised because of a reduction in water level, possibly caused by the cutting of a new channel for the river Llynfi (Lloyd 1870, 95); Edgar Dumbleton says that the lake level had been lowered by about 18 inches about seven years earlier (i.e. in the early 1860s), and it is clearly shown on the map of the lake shores by Isaac Davies dated July 1866 (Fig. 2.1). There may have been fluctuations in the lake level over the previous decades as the island, named Bwlc, is shown on the earliest (1832) One Inch Ordnance Survey map (Fig. 1.2) as well as the 1840 tithe map of the parish of ‘Llangasty Tally Llyn’. The surveying for the 1832 map may have been done before 1813 or 1814 although revisions were made in 1828–30 (Harley & Oliver 1992, x–xiii). The size and general location of the island on the 1832, 1840 and 1866 maps correspond relatively well with the modern shape of the island without its reed beds and so may indicate a similar water level, although the 1866 depiction seems either to have been turned by 90° clockwise or is depicting erosion on the western side (Fig. 2.2d). The earliest depiction of the island on an estate map of 1810 shows a much larger island, perhaps five times larger, and located 500m to the south-west (Fig. 2.3). The varied and schematic map depictions of the island contrast markedly with the accuracy of the field boundary depictions and clearly reflect the professional focus of the map surveyors. However although the island was clearly visible early in the nineteenth century there is no reason to disbelieve the Dumbleton report to the British Association for the Advancement of Science that ‘until two years ago no idea was entertained that this [island] had anything remarkable about it’ (Dumbleton, E. N. 1870a, 130).

Fig. 2.1 1866 Estate map of Llangorse Lake with crannog marked as island.

Fig. 2.2 Early depictions of the crannog: (a) 1810 estate map (b) 1832 OS 1" map (c) 1840 tithe map (d) 1866 estate map. (© (L–R): Powys Archives, Ordnance Survey)

Henry Dumbleton states that the summer of 1868 was particularly dry and, as the lake level was very low, oak piles were observed ‘close beneath the surface of the water’ (1870, 102). The lake remains the subject of large seasonal fluctuations in water level (up to 1.5m; see Chapter 1.3). However, the illustration of the crannog published in Archaeologia Cambrensis in 1870 (Fig. 2.4) show the tops of planks in the southern palisade extension exposed above water (about 100mm, partially hidden within reeds), whereas the same palisade timbers were always submerged by a minimum of about 10cm at lowest summer water levels during the 1989–93 excavations. This may indicate the very exceptional nature of the 1868 water level, although artistic licence in the crannog depictions to indicate the position of submerged timbers seems quite likely (see below). Edgar Dumbleton says that until the water level was lowered the island was much smaller – ‘not half its present size’ (1870b, 196). The depiction of the island with a higher central oval area, in both of the published 1870 reports, would seem to represent that small earlier island – the summit area being an eroded shore edge (Fig. 2.5).

Evidence for a significant variation in past lake levels may be indicated by Henry Dumbleton’s record that trees were observable under water on the east side of the lake and elsewhere (1870, 101–106). One such tree, an alder, is noted on the schematic profile of the island edge and adjacent lake shore (Fig. 2.6).

The Dumbletons recognised the island as artificial because its rocky composition was quite unlike the rest of the lake shores (Dumbleton, E. N. 1870a, 130; 1870b, 193). Having recognised oak piles surrounding the site, Henry Dumbleton had some removed and confirmed they were artificial (Fig. 2.7). By 1869 the Dumbletons had carried out several ‘perpendicular openings’ or ‘holes’, including one noted as the principal one at the centre or ‘highest point’ of the island, then 4.5 feet above the water level (Dumbleton, H. 1870, 102).

By July 1870 two somewhat varied plans and other site illustrations were available, as well as more detailed information about structures and finds. This may mean further excavation had taken place. The Woolhope Naturalists’ Field Club, a Hereford-based society, visited the lake in July 1870 and Henry Dumbleton presented an account of the site to the visitors. The scenic sketch view (Fig. 2.8) accompanying the plan in this report may be a primary on-site sketch as it is dated 22nd July 1870. The accompanying plan has the same draughtman’s monogram but is presumably the result of longer term surveying (Fig. 2.9). The Woolhope sketch shows a considerable amount of plant growth and at least one young tree on the crannog, and the sketch plan clearly plots the location of at least three bushes or trees. This engraving is in some respects different from the same perspective published in Archaeologia Cambrensis (Fig. 2.4). However, the vegetation growth on what may be the primary on-site illustration may well be correct, as photographs from 20–30 years later show more mature trees in the same locations; the earlier 1840 map also shows plant growth.

Fig. 2.3 ‘A Plan of the Division of the Brecknockshire Estate’ by H. A. Biedermann, 1810. The crannog was clearly visible, but incorrectly positioned within Llangorse Lake. (B/D/BM/M/1/9; © Powys Archives)

Fig. 2.4 View of the crannog from the south, as reproduced in Archaeologia Cambrensis 1870, Fig. 2. (By permission of the Cambrian Archaeological Association)

Fig. 2.5 Plan of the crannog, reproduced from Archaeologia Cambrensis 1870, Fig 1. (By permission of the Cambrian Archaeological Association)

Fig. 2.6 Schematic section through the crannog reproduced from Archaeologia Cambrensis 1870. (By permission of the Cambrian Archaeological Association)

Fig. 2.7 Details of plank and pile, reproduced from Archaeologia Cambrensis 1870, Fig. 5 & 6. (By permission of the Cambrian Archaeological Association)

Fig. 2.8 View of the crannog from the south, reproduced from Transactions of the Woolhope Naturalists’ Field Club 1870, opposite page 101.

The illustrations published in Archaeologia Cambrensis for July 1870 show the island with little more than a few bushes against an unwooded background (Fig. 2.4). The plan shows the island as having a distinct oval central area (Fig. 2.5) and marks some additional timbers not shown on the Woolhope plan. There is also a stylised section of the island and lake bed (Fig. 2.10) (similar to that published in the Woolhope Naturalists’ Field Club transactions; Fig. 2.11). The Archaeologia Cambrensis engravings are by nature clearer, and professionally produced from the original drawings. The two views of the crannog are similar with exaggerated hills and varying vegetation, but both show a perspective that appears to combine a view of the crannog from the south, probably from a boat, with one of the hills to the east of the lake perhaps sketched from the shore. However, it is clear that vegetation and hedges on the surrounding shores have been artistically stripped away by the artist or engraver, J. H. Le Keux, in order to emphasise the grandeur of the terrain.

The mound and timbers recognised by the Dumbletons were described as follows: an island 90 yards in circumference, squarish in shape with rounded corners, at its highest 5 feet above water level. This was made of stone except at the northeast angle (Dumbleton, E. N. 1870b, 193–94). The Archaeologia Cambrensis plan (Fig. 2.5) seems more complete and has timbers which are not shown on the Woolhope copy. It seems probable that the engraver also reduced the amount of vegetation on the crannog, to make the depiction clearer, and that differences in the rubble spread are less reliably shown in the Archaeologia Cambrensis plan.

Fig. 2.9 Plan of the crannog reproduced from Transactions of the Woolhope Naturalists’ Field Club 1870, opposite page 101.

More than 60 oak slab piles up to 6.5 feet long are shown in two palisade lines with material in between them. Only one clear line is shown on the Woolhope plan (Fig. 2.9), although some elements of an inner palisade line are clearly visible on the Archaeologia Cambrensis plan. Eighteen roundwood piles are recognisable on the Woolhope plan, mostly in the water outside of the oak palisade lines. The Dumbletons thought these might be supports for buildings projecting into the water, although no clear building plan was recognisable, and this interpretation was queried by Lee in the discussion published in the Woolhope Transactions (Lee 1870, 106). Sixteen roundwood timbers were shown lying radially to the centre of the island in four groups. Some of these at the south-east corner ran under the stones of the mound on the inside. One group of four were edged by an oak horizontal and some ran against the inner oak palisade by the water’s edge. A number of other timbers are visible on both plans, both in the water and on the island. Four other horizontal logs (three soft-wood and one oak) were noted ten yards from the north-east of the crannog in the channel between the island and the shore, and were thought might represent traces of a gangway between the crannog and the shore (Dumbleton, H. 1870, 104). These timbers were fairly close to the remains of a logboat discovered in fragments in 1990 (see Chapter 16) and it is possible that this is what has been represented. Indeed, the Dumbletons describe one timber as notched oak exhibiting marks of a ‘heavy cutting instrument’ (Fig. 2.5; Dumbleton, E. N. 1870b, 198; Dumbleton, H. 1870, 104).

The Dumbletons give three descriptions of the sequence of deposits, illustrated by two section drawings. The 1869 Exeter conference report records from the bottom: 1 – faggot wood and reeds; 2 – loose mould with a few stones and a considerable layer of charcoal about 1 foot above the lake level; 3 – stone more frequent nearer the surface; maximum depth of deposit five feet above the water (Dumbleton, E. N. 1870a). The Woolhope account is more detailed and notes variations in different parts of the island. Firstly, it describes an apparent cutting in the centre of the mound with a maximum depth of about 4.5 feet above the water level. The sequence from the bottom is then: 1 – shell marl; 2 – reeds and small wood apparently bound together like faggots; 3 – layer of tolerably compact peat; 4 – three feet of large loose stones and earth. It states that these proportions were reduced near the water’s edge, but implies a similar sequence of deposits in other cuttings. Secondly it refers to the north-east angle (where Henry Dumbleton had noted the absence of stones): 1 – peat and reeds; 2 – vegetable mould with much charcoal; 3 – a deep deposit of vegetable mould. The sketched section of the east side of the island (Fig. 2.11) shows from the bottom: 1 – shell marl; 2 – peat, reeds and wood; 3 – stone; 4 – a lens of charcoal towards the edge of the mound; 5 – more stone; 6 – surface vegetation (Dumbleton, H. 1870, 100). This does raise the question as to whether the claimed presence of peat above timber in the main cutting is a mistake. The third account in Archaeologia Cambrensis gives a similar description of several perpendicular openings: 1 – shell marl; 2 – reeds and faggot-like small wood (observed all round the edges of the island immediately above the soft marl); 3 – black compact peat 18 inches above the marl; 4 – large loose stones with vegetable mould and sand. As before, the northeast sequence is different: 1 – peat, small wood or reeds, one foot in thickness; 2 – six inches of earth with charcoal; 3 – 18 inches of vegetable mould with very few stones. The illustrated east side section (Fig. 2.10) is similar to the Woolhope section but, like the plans, has been tidied up for publication (Dumbleton, E. N. 1870b, fig. 4). Both sections show a major lens of charcoal with the stone of the mound above and below it, and also overlying the vertical piles on the edge of the island. Both also indicate that the deposits in the north-east angle of the island were different with fewer stones, deep deposits of vegetable mould, lots of bone and a distinct layer of charcoal. Edgar Dumbleton also refers to sand in among the top layer of stones and on the north-east side.

Fig. 2.10 Schematic section through the crannog reproduced from Archaeologia Cambrensis 1870, Fig. 4. (By permission of the Cambrian Archaeological Association)

Fig. 2.11 Schematic section through the crannog reproduced from Transactions of the Woolhope Naturalists’ Field Club 1870.

Both the Woolhope Naturalists’ and Archaeologia Cambrensis accounts claim that the sequence of deposits in the centre found peat above deposits of wood in contrast to the north-east corner. In contrast the section drawings label a single horizon containing peat, reeds and wood. This has quite important implications for the interpretation of the site (see Chapter 4).

Both reports comment on the plentiful bones in the north-east corner of the crannog – ‘nearly all the larger bones were found on the low shallow side’. However, bones were reported to be plentiful everywhere and at all levels in the excavation of the interior except in the middle of the mound. Apart from the bones the finds were few: two bone awls (possibly natural), a large piece of leather with perforated edges containing remains of leather lacing, three or four scraps of pottery and one oval stone with ‘marks of grinding’. One piece of bronze ‘apparently a portion of some small hollow utensil’ decorated with concentric lines was noted as probably Roman, although the possibility it was modern was recognised. Many of the bones were split or cracked. The bones were commented on by three specialists. Professor Rolleston of Oxford noted large and small horses, pig, cow and sheep. Boyd Dawkins reported a later assemblage with red deer, wild boar and bos longifrons (Dumbleton, H. 1870, 103–04). Finally a Professor Owen examined a series of bones and identified hog, bos longifrons, ass or a small equine (Dumbleton, H. 1872, 147–48). It would seem from the references to the different animal bone collections that there was some debate about whether the greater percentage of non-domesticated species might indicate greater antiquity for some of the bones. The supposed presence of bos longifrons and wild species had initially been taken to indicate a ‘remote period’. However Henry Dumbleton concluded his short 1872 note by concluding that the cattle bones need not be very ancient and that the animal bones rendered ‘very uncertain the age of the deposit on the island’.

The Dumbletons were apparently, then, the first to recognise the island as artificial and to identify it as a crannog. It is quite clear they were fully aware of the nineteenth-century excavation of crannogs in Ireland and Scotland as well as Ferdinand Keller’s publication on Swiss lake villages (1866). Henry Dumbleton states: ‘It may be almost superfluous to mention before this audience that, within the last 20 years, many remains of habitations raised on piles, as well as upon artificial islands, have been found in several of the lakes of Switzerland and Italy. In England and Wales, so far as the writer is aware, no such remains have been hitherto found, but in Ireland there are many such artificial islands, bearing in construction and arrangement a striking similarity to the one under discussion’ (1870, 101). The Monmouthshire antiquary J. E. Lee, who had translated Keller’s Lake Dwellings work into English, visited the crannog with the Woolhope excursion and confirmed its close similarity ‘in almost every respect’ to the ‘crannoges’ found in Scotland and Ireland. As noted above, he advised the outing that they should not think the site was as old as the Swiss lake villages and indicated that there had been some scepticism about accepting Llangorse as a crannog: ‘some antiquaries of note were disposed to doubt the very existence of a Crannoge in this place; … those who thus doubted had never set foot on the island, and had merely contented themselves with a view of it from the mainland’ (Lee 1870, 106–07). However Lee advised against Dumbleton’s suggestion that the roundwood piles might be part of wooden buildings built out over the water ‘like the Swiss dwellings’ and pressed care in suggesting a date for the site unless datable finds were found in direct association with ancient animals (Dumbleton, H. 1870, 106–07).

In his Archaeologia Cambrensis paper Edgar Dumbleton suggests his intention to examine the crannog further (1870b, 198). If he did so there appears to be no record of the findings. Likewise we have, thus far, been unable to find any trace of their artefact finds or the animal bones, and there is no record of where they might have been deposited. Henry Dumbleton also suggested that another small island existed in the lake but without any evidence of timber piles (1870, 105). He does not give any indication of its location. The Archaeologia Cambrensis paper was reprinted with illustrations in Keller’s second edition of The Lake Dwellings of Switzerland and other parts of Europe, presumably at the instigation of Lee, the translator of this volume as well (1878, 660–65; plate cxlviii). This seems to have brought it to the attention of Robert Munro, who included a shortened version of Dumbleton’s account in his Ancient Scottish Lake-dwellings volume (1882, 296–98). Munro returned to comment on Llangorse in his 1890 volume on The Lake Dwellings of Europe. In his chapter on possible English lake-dwellings, in which he published entries from the Dumbleton report, Munro accepted Llangorse as genuine – ‘constructed after the manner of the Scottish and Irish crannogs’ (1890, 464); ‘the artificial island in Llangorse … indeed having all the appurtenances of the typical crannog’ (1890, 477). He commented critically that there was ‘no means of determining either the age of this singular lacustrine abode or the social condition of its inhabitants’; and that ‘It is to be regretted that no relics were found on this island, and I cannot help thinking that, in the circumstances, a more careful search would have furnished some scraps of the handiwork of its occupiers’ (1890, 465). Munro thought that Llangorse crannog was important and that it might be older than the majority of crannogs in Scotland and Ireland. He suggested that it, and the flimsier evidence from England, indicated that lake dwellings had been introduced into Britain from Europe by the Celts. These crannog-building Celts had been displaced by later invading Belgae and others, leaving Celts in Ireland and Scotland continuing the crannog-building tradition in later centuries and thus explaining the frequency of sites in those countries (Munro 1890, 489–90).

2.2. THE CRANNOG 1870–1980

There is no evidence of any discussion of the site in the decades following Munro’s observations in his 1890 The Lake-Dwellings of Europe, but in 1925 the crannog was again brought to professional attention when a logboat was discovered in the lake. The discovery was reported by Cyril Fox, the newly appointed Director of the National Museum of Wales, in a short note in Archaeologia Cambrensis for 1925 and a longer article in Antiquaries Journal (1926). The logboat and details of its discovery are dealt with in detail in Chapter 16. It was found by Tom Jenkins, a local carpenter, and initially reported as ‘some 200 yards east of the crannog’ (1925, 421). Fox indicates that Mr Jenkins had known about the logboat for some years but had only been able to recover it because of an unusually low lake level – it was still lying in 3.5 feet of water. In his longer paper Fox marks the location as about 400m to the east of the site and about 150 m offshore (Fig. 2.12; Fox 1926, 121, fig. 1). Although it was found some distance from the crannog, Fox suggested it should be associated with the island partly because ‘dugout boats are frequently associated with crannogs’ (1926, 126). He noted that: i. the nineteenth-century excavations had not recovered anything to establish the date of the crannog; ii. that, although many crannogs were of Iron Age date, logboats were in use both earlier and later. He also stated that ‘no remains, recognisably medieval or later, were found in the adjacent crannog’, and concluded that its occupation ceased before the thirteenth century at latest (Fox 1925, 422; 1926, 127, note 1), although there is no indication in his report that he had seen the finds. The remainder of Fox’s paper is concerned with other examples of ‘monoxylous craft’ and with a proposed typology and chronology of such vessels. On this basis he concluded that the Llangorse logboat (= Logboat 1; see Chapter 16) was likely to belong to the Roman period (1926, 146). This date was then transferred in some subsequent literature to the crannog.

Hubert Savory discussed the crannog in his article on prehistoric Brecknock and accepted that it was the only convincing example of such a site in Wales (1971b, 117–18). He noted the difficulty of assigning it a date, partly as a result of the loss of the finds and in particular the loss of the pottery found on the site, and speculated that the site might be Late Bronze Age, in comparison with the then dating of some Irish sites. However, he concluded that it was more likely to date to ‘some part of the true Iron Age’ and be contemporary with the logboat published by Fox, and so included it on his map of ‘Iron Age Brecknockshire (c. 400 BC – 100 AD)’ (Savory 1971b, 116, fig. 6).

Fig. 2.12 Ordnance Survey map showing the crannog and the location of the 1925 logboat. (Reproduced from Fox 1926, fig. 1; base map Crown Copyright: Ordnance Survey)

Fig. 2.13 This photograph taken by J. K. St Joseph on 21 June 1960 shows extensive reed growth around the crannog. (Reproduced with permission of the Cambridge University Collection of Aerial Photography © reserved)

The National Monument Record has an aerial photograph of the crannog and adjacent shoreline taken by J. K. S. St Joseph (1912–94) on 21 June 1960 (AB I 72; Fig. 2.13), and cards indicate that it was visited by RCAHMW surveyors in the 1970s. The cards note that the site as ‘prob. IA/RB’ and that oak piles are visible protruding from the lake bed, but the surveyor (GB 9/11/73) expresses the view that the timbers are ‘doubtfully ancient’. This doubt about the age and nature of the site was subsequently revised in the early 1990s on the basis of evidence provided by the authors, and its identification as a crannog fully accepted, although the Commission entry for the Brecknock Inventory continued to assert that ‘How far such a structure (which differs fundamentally in design from the Irish and Scottish sites which give their name to the habitation type) might have been copies from them remains a matter of speculation’ (1997, 281). In 1978 Sean McGrail published a detailed corpus of known logboats in England and Wales which included a detailed discussion of the Llangorse find and published a radiocarbon date for it of 1136±60 quoted as ‘c ad 814’. McGrail and Switsur had previously expressed doubt about this date due to the fact that the early twentiethcentury conservation methods used on the boat might have distorted the date (1975, 195). McGrail subsequently accepted the date for the boat as Switsur concluded that the radiocarbon pre-treatment had removed any contamination. The boat was consequently dated as cal AD 1135± 60, calibrated as AD 870–965 or AD 770–1005 at two sigma (McGrail, in litt.; see Chapter 16).

A paper by Raikes on the geography of the lake cites the ‘AD 800’ date for the ‘canoe’ while quoting ‘tentative archaeological opinion’ (citing Chris Houlder of RCAHMW) that the crannog is late Iron Age or Romano-British in date (Raikes 1986–87, 30). McGrail’s date of c. AD 814 seems to have been adopted by the Brecon Beacons National Park Authority: the texts on their information notice-boards on the north shore landing stages in the late 1980s stated that the crannog had been built c. AD 800. This may explain why Raymond Williams’s fictional historical trilogy People of the Black Mountains, set in this part of Wales, includes a Viking period story in Book 2 (‘Dragon and Gentile’) which was set on the crannog (Williams 1990, 138–62). Williams’s story is set in AD 896 (coincidentally very close to the actual dendrochronological dates) and readers are told that the ‘cranog’ (sic) is much older and believed to have been the residence of a king ‘in the old times, before the loss of the sunken land’ (1990, 138).

Consequently when the current work was begun there was still uncertainty about the precise nature of the crannog, its cultural significance and its date. It seemed to be a unique site, and perhaps of Iron Age date, although an early medieval date and an Irish connection seemed possible (Lane 1988, 121–23). The uniqueness of the site and its poor dating meant that it had hardly figured in Welsh archaeological literature since its initial discovery. It could not be easily slotted into an intelligible context and its very uniqueness led to professional doubt about its identification and antiquity (Campbell & Lane 1989, 675).

2.3 THE REDISCOVERY OF THE CRANNOG

The crannog was first drawn to Alan Lane’s attention in the early 1970s by Leslie Alcock, then Professor of Archaeology in Glasgow University, as part of a special subject study on Dark Age Britain. Lane’s recollection is one of scepticism of the likelihood that the site was a genuine crannog, possibly exacerbated by his failing to find the Dumbletons’ plan in Archaeologia Cambrensis as it was bound separately from the short article (Dumbleton, E. N. 1870b). He did not visit the site until the early 1980s, by which time he had been appointed to teach early medieval archaeology in University College Cardiff (now Cardiff University). When he did visit the site, and rowed out to inspect it, he was absolutely flabbergasted to find substantial oak planks protruding out of the water (Fig. 1.5) and partially encircling the little stony island, just as the Dumbletons had reported. Being familiar with Scottish and Irish crannog sites there seemed no reason to doubt that this was a crannog and that the timbers were ancient.

In the mid-1980s Ewan Campbell and Alan Lane were working on the evaluation of possible early medieval settlements in south Wales, published as Early Medieval Settlements in Wales (Edwards & Lane 1988). As part of this process the site was revisited and the available published evidence regarding the site was evaluated in its local context. The known fifth- to seventh-century Irish connections of the area demonstrated by local occurrences of ogham inscriptions, the Irish associations of the origin legend of Brycheiniog, and the sheer numbers of early medieval crannogs in Ireland suggested that this might be a site with Irish associations, perhaps of early post-Roman date. In addition the local charter evidence published by Wendy Davies (1978; 1979) suggested a Welsh royal and ecclesiastical presence in the area in the early tenth century. The site was also clearly suffering from erosion on its southern side as a result of wave action backed by the prevailing winds. Vertical oak planks could be seen on the south and east sides of the stony mound of the island and soft wood wattle was eroding out in places. The resulting brief publication of the site (Lane 1988, 121–23, fig. 26) confirmed that they thought the site was a crannog, in spite of the scepticism of others, and acknowledged the difficulties of dating the site. However, the combination of erosion to a probable site of great importance, and its possible early medieval date, underlined the potential of a site at risk, and the need for further work.

Fig. 2.14 Contour survey of the crannog made in 1988, showing schematic depictions of the visible timbers and locations of the first batch of dendrochronologically dated timbers. (Reproduced from Antiquity 1988, fig. 2).

Fig. 2.15. Cardiff University student Rick Petersen breaking ice on the lake to access the oak palisade in the winter of 1988.

As the site has been designated an SSSI on environmental grounds, as well as a Scheduled Ancient Monument (SAM), there were constraints on what could initially be done. However, in 1988 a team from Cardiff University, led by Lane and Campbell, undertook surface survey of the island without cutting down vegetation. This was followed by the plotting of the visible oak timber lines and areas of eroding wattle and horizontal timber. The resultant plan published in Antiquity (Fig. 2.14; Campbell & Lane 1989, fig. 2) and Medieval Archaeology showed a sub-square mound quite similar to that published by the Dumbletons in the nineteenth century but with distinct evidence of erosion eating into the mound from the south. Three oak planks were cut below the water line to provide dendrochronological dates, or failing that radiocarbon dates, for the site (subsequently labelled timbers 519, 643, 732). Two samples provided successful growth sequences, matching the master tree ring sequence between AD 747 and 859 and AD 704–850. In view of the loss of sapwood on these timbers, which were protruding out of the water and subject to fluctuating lake levels, the date of cutting was estimated as after AD 860 and probably before AD 906 (Campbell & Lane 1989, 677–78).

This date was a surprise as it had been assumed that Irish influence might be more likely in the earlier sixth- to seventh-century period when many Irish crannogs were built. However, a late ninth-/early tenth-century date meant that the royal charter evidence was more relevant. Lane and Campbell then realised that an Anglo-Saxon Chronicle entry for AD 916 contained a direct reference to an attack on the lake resulting in the capture of the local queen. Given the charter evidence and the chronicle reference it seemed likely that this was a royal crannog of the late ninth /early tenth century and arguably a short-lived site. The possibility of an earlier origin could not, of course, be ruled out. However the prospect of a waterlogged site, with the potential for artefactual and environmental survival, for a period where virtually no settlement sites were known from Wales, made further exploration of the site a high priority (Campbell & Lane 1989). The appointment of Dr Mark Redknap, an experienced diver and underwater excavator, as Curator of Medieval & Later Archaeology in the National Museum of Wales (now Amgueddfa Cymru – National Museum Wales, AC–NMW), provided a new potential for combined off- and on-shore investigation of the site and the support and interest of the National Museum a new impetus. In consequence, the National Museum proposed a partnership project with Cardiff University, and this was launched in 1989.

SECTION 2

The Excavations

CHAPTER 3

The Survey and Excavation Strategy 1989–2004

3.1 TEAMS AND SEASONS

The site was initially investigated in 1989 and 1990 by two co-ordinated teams, one from AC–NMW led by Mark Redknap, the other from Cardiff University led by Alan Lane and Ewan Campbell. The National Museum undertook further survey and excavation in 1991 and 1992. In 1993, smallscale underwater work was undertaken on the south side of the crannog by a National Museum team in conjunction with filming content for a new Channel 4 television series called Time Team (Series 1), the Team Team undertaking some survey assessment of nearby landscapes. A final season of survey and limited excavation was undertaken in 2004 by Gifford and Partners Ltd and Marine Archaeological Research Consultants (MARC) prior to site conservation work. Post-excavation work, including the conservation and analysis of the carbonised textiles and lifted timbers, began in 1989 and continued intermittently until 2017. Although interim reports were produced after the excavation seasons and several synthetic papers have been published, the delay in completing this definitive report has been a consequence of the time required for some of the complex conservation and post-excavation analysis, fluctuating resources available and other commitments.

3.2 SURVEY

The initial 1988 investigation made a simple contour plan of the site without removing the surface vegetation (Campbell & Lane 1988, fig. 1). A slightly different plan was published with the Antiquity article of 1989 (Campbell & Lane 1989, fig. 2; see Fig. 2.14). A more detailed theodolite survey was undertaken in the first excavation season of 1989, after the removal of most surface vegetation. This involved the setting up of a permanent survey grid of 10m squares over the accessible dry areas of crannog, marked by metal surveying pins, which remained in use in successive seasons until 1993. A new GPS survey was undertaken in 2004 by Gifford and Partners Ltd using a survey grid aligned on magnetic north.

These different contour surveys were undertaken in conditions of varied vegetation cover and fluctuating water-levels, reflecting the time of year and changing rainfall in the Llangorse catchment. No clearance was attempted of reeds growing on the north side of the site, as the island is a Site of Special Scientific Interest (SSSI), and these reeds largely lay beyond the margins of the crannog. These variables have resulted in the creation of slightly different plans of the island and its shoreline. Figure 3.1 represents a compilation of these different surveys, although largely based upon the theodolite survey of 1989 after vegetation clearance of the main surface of the island. The reed swamp to the north was left largely untouched on nature conservation grounds.

A geophysical survey of the dry and cleared areas of the island was undertaken by Mike Hamilton of Cardiff University in 1989 using a resistivity meter. This did not reveal any useful or intelligible patterns, and so was not used as a guide to selection of the excavation areas.

The visible timbers on the edge of the island and those traceable underwater were plotted by the National Museum team using a combination of traditional land-archaeology techniques coupled with an approach to underwater survey known as the Direct Survey Method (DSM). This method had already been used to advantage on excavations of the wrecks of the Mary Rose, Sea Venture, Amsterdam and other underwater archaeological projects (for a fuller account, see Rule 1989). On these projects, tape measures were the main surveying tool, as poor underwater visibility made photography difficult, angles were difficult to read accurately and plumb lines problematic, and electronic measuring devices that would work well underwater had yet to become widely available. The DSM technique involved using direct (not necessarily horizontal) tape measurements from datum points of known three-dimensional cartesian co-ordinates with the quantification of errors and a computer programme to process the data and find best-fit solutions. The technique and computer programmes were developed by Nick Rule to provide a fast, simple, and accurate technique with quantifiable error, important on sites with poor visibility where plans are hard to verify by eye, and the project benefitted significantly by his contribution to site surveying during the 1989 and 1990 seasons.

Fig. 3.1 Contour survey (1989) of the island and adjacent part of Llangorse Lake.

Fig. 3.2 The Direct Survey Measurement (DSM) grid relied on a combination of reinforced steel pins on the dry crannog platform, and scaffold poles and tagged palisade timbers in the lake.

Fig. 3.3 Idealisation of DSM survey on submerged parts of the crannog in 1989. (Drawing M. Redknap & Tony Daly)

For the primary DSM survey stations, hooks were attached to a number of submerged palisade timbers, which were considered to be immovable and rigid, and sited for suitable use in successive seasons. Where no suitable timbers existed, scaffold poles were driven vertically into the lake bed, their above-water top ends providing locations for additional offshore datum points (P, Q, R, S, T; Fig. 3.2). Datum points were small metal hooks attached to timbers, each marked by a surface marker buoy. The distances were measured from every datum to every other for which there was line-of-sight (‘interdistances’) and drawn up as a sparse triangular matrix (Figs 3.3, 3.4). Divers used a minimum of three datum points for every point surveyed (for each palisade timber, usually both edges at lake bed level, and points on timber tops). Measurements were recorded on permatrace and transferred to diver log sheets and compiled for later processing. The computer programme would lookup the co-ordinates of each datum point in a database and calculate a position for the site point using an algorithm, and iterative adjustments. The solution was printed, with an estimate of error in the solution and estimate of error in each of the measurements. This established the proprieties for cross-checking measurements on site, in the case of large errors (the assumption being that, as on the Mary Rose, poor visibility and underwater conditions would lead to errors in reading or transcription). The technique worked well, providing high precision plots for finds and timbers, final residual errors ranging from 0–2.6%, average 0.1–0.2% (=the variation difference between observed distances measured by the diver and computed positions from datums).

Fig. 3.4 A: timber numbering system for palisade timbers (512–514, 579 in Trench B, 1989). B: hook and tag system used for