Archaeological survey and Excavation at Lan Fawr Enclosure, Llangynidr, Brecon, SO 1707 1838

Interim Report Oct 2008

Excavations at Lan Fawr, facing north (photo: Alan Baines)

Dr Jemma Bezant Department of Archaeology and Anthropology University of , Lampeter

Contents

Contents ...... 1 List of Figures ...... 2 Project metadata ...... 3 Summary ...... 5 Scope ...... 5 Acknowledgements ...... 5 Lan Fawr: Introduction ...... 8 Aims ...... 10 Geophysical Survey ...... 10 Excavation...... 12 Trench A ...... 13 Trench B...... 15 Discussion ...... 17 Future research questions ...... 19 References ...... 20 Appendix 1 ...... 21 CPAT Regional Historic Environment Record...... 21 Appendix 2 ...... 24 Section drawings, trenches A and ...... 24 Appendix 3 ...... 26 Ceramic sherds ...... 26 Appendix 4 ...... 27 Excavation site record ...... 27

1 List of Figures

Figure 1. Location Map pp7

Figure 2. Geophysical Survey pp11

Figure 3. Trench location map and aerial survey mapping pp13

Figure 4. Trench A. The stone rampart (facing north, scale 2m, photo J Bezant). pp14

Figure 5. Trench A. The stone rampart. Roughly coursed, red sandstone slabs (Facing north, scale 20 cm, photo J Bezant). pp14

Figure 6. Trench B. Former land surface and ditch (Facing southeast, scales 2m, photo J Bezant). pp16

2

Project metadata

Project Name Geophysical Survey and Excavation at Lan Fawr upland enclosure, Llangynidr, Brecon, SO 17071838 Survey Purpose To conduct geophysical survey and small- scale excavation of a bivallate enclosure. Spatial Coverage Circa. 2.5 acres centred on S) 1707 1838 Country Wales, UK Duration August 2008 Weather Wet Soil condition Well drained Land use Pasture

Monument type Upland bivallate enclosure Monument period Late prehistoric? Survey Director Dr. Jemma Bezant Client Llangynidr Local History Society

Landowner Mr William Gibbs Geophysics Survey Type Magnetometer Geophysics Instrumentation Bartington magnetic gradiometer 601/2 Geophysics Area surveyed 2.5 acres Geophysics Method of coverage 30 x 30 metre grids walked in zigzag format. Survey starts in bottom left corner, direction south. Geophysics Traverse interval 1m Geophysics Sample interval 0.125m Geophysics Spatial accuracy Some variation of grid layout occurred across each survey area due to unevenness of ground and operator error, this is estimated to be + 1m. The estimated accuracy within grids is

3 estimated to be within 10 cm.

Geophysics Sample accuracy Automatic trigger whilst walking, 0.1 nT sensitivity Geophysics Data Files Raw data deposited with client. Each grid as a DAT file is accompanied with metadata HDR file. Geophysics Digital treatment Downloaded as xyz format via Grad601 v.3. Each grid imported into Archeosurveyor and clipped to + 6 nT. Zero mean traverse applied to reduce striping. Composites exported as Tiff files. Geophysics Geolocation Survey grid positions recorded using Garmin Legend GPS. Post-processed bitmaps then geolocated to OS Landline map base via ArcMap v9.2 Excavation Data Paper record comprising context record sheets and indexes have been deposited with client (app 4) along with photo record on CD. Site drawings have been scanned and recorded as Tiff files in line with AHDS recommended archiving standards. Finds to be retained by client. Soil samples, GIS map data and shapefile layers to be retained by J Bezant for future research.

4

Summary

As part of the annual fieldwork project conducted by the members of Llangynidr Local History Society, a three week long project of geophysical survey and excavation was directed by Dr Jemma Bezant of the Archaeology and Anthropology Department, University of Wales, Lampeter. A geophysical survey was conducted prior to excavation where two long, narrow trenches were opened across the inner and outer enclosure banks. The inner bank was found to be a substantial rampart wall constructed of red sandstone quarried from a nearby outcrop. The outer bank comprised an earthen bank and outer ditch and had been constructed to emphasize slightly rising ground that formed the interior of the enclosure on the edge of the plateau. Dating remains difficult at this stage but a very abraded, possibly late prehistoric to medieval, ceramic sherd was recovered from the upper ditch fill.

Scope

This report details the methods and results of a geophysical survey and excavation. An interim interpretation of the results is provided but a full analysis remains to be met by future research.

Acknowledgements

Special thanks must be made to the landowner William Gibbs whose generous grant made this project possible. To all of the members of the Llangynidr history society (especially those of the excavation team), including Deb Fisher, David Stephenson, Deb Vulliamy, Peter Seaman, Peter Hodges, Kerry Graves, Douglas Piercy, Jan Bailey, Stuart Busson, Richard Fisher, Charlene Hutchison, Rosemary Evans, Angus Kings, Ann Jessop, Sue Ware, Michael Eustace and Wendy Toomey. I am especially indebted to the kind

5 hospitality and guidance of Mike and Hilary Scott-Archer and to the kind help supplied by Alan Haines and Geoff Williams.

6 Fig. 1. Location Map.

7

Lan Fawr: Introduction

Although this project is aimed primarily at the activities and training relating to a volunteer community group it is recognized that the project addresses some of the recommendations made in the recently published Research Framework for the Archaeology of Wales (http://www.archaeoleg.org.uk) and the Middle Usk valley in southeast Wales is one of the areas defined by the RFAW as a target for further archaeological investigation vis a vis the late prehistoric period. The site (see fig 1 for location) also lies immediately to the south east of the Pencelli-Talybont character area as defined by CPAT as part of their Middle Usk Valley, Brecon and Llangorse Register of Landscapes of Historic Interest in Wales (http://www.cpat.org.uk/projects/longer/histland/usk/muintr.htm#Themes).

The Llangynidr area some 10 miles southeast of Brecon in southeast Wales lies south of the which occupies a wide level valley floor of large, irregularly enclosed, hedged fields. The slopes are mostly lightly wooded and are enclosed onto the fringes of Mynydd Llangynidr. Lan Fawr lies just within the uppermost extent of enclosed land before it gives way to open common.

These enclosed areas are predominantly improved pasture with unimproved grazing on the upland limestone which is crisscrossed by upland routes and former quarrying activity. The Roman road is presumed to run east-west along the route of the current A40. A Roman auxiliary fort is set on a detached knoll just east of the pass at , lying roughly midway between the military settlements at Brecon Gaer (NPRN 92001) and (NPRN 306295).

Lan Fawr lies at 300m above ordnance datum and overlooks the middle Usk valley to the north from an elevated but sheltered north-facing ridge, directly above Llangynidr village and the Myarth Forest. This ridge comprises Old Red sandstone and, where it outcrops, has provided suitable building stone for the

8 extensive common enclosure walls and farmsteads. It is below one of these heavily quarried outcrops that Lan Fawr enclosure is sited on the steep edge of a level shoulder along the ridge. The oval enclosure measures 80 metres east-west by 30 metres north-south and is described in some detail in an entry in CPAT’s Historic Environment Record (App. 1) which describes it as an Iron Age defended enclosure. A topographical survey is also reproduced by Browne and Percival (1989) and the earthwork has also been mapped as part of an aerial mapping exercise undertaken by Tom Pert of the RCAHMW (see fig. 3). The earthwork is named on the 1st edition OS 1:10560 map as Cloddau Duon which is a name known from other earthwork sites in Wales deriving from cloddi – ‘to dig’.

The inner bank comprises a bank of grass-grown rubble, along the top of which is a spread of small to medium sized subangular sandstone boulders. There is no apparent structure or entrance although robbing appears to have occurred at several points. The outer bank comprises a low but wide ridge with some evidence for an external ditch. A later field boundary comprising a low earth bank topped with uncoursed rubble stone radiates out over both the inner and outer banks and the whole site is enclosed by a commons wall, being the extremity of enclosed land on this upland stretch.

The site has panoramic views to the east, north and west, over the middle Usk Valley and the Black Mountains beyond and a number of other significant Iron Age enclosures would have been visible from this location. Whilst contemporaneity with any of the sites cannot be established they are discussed further below.

9 Aims

The aims of the project are two-fold:

 To ascertain the character of the archaeological deposits on the site though a geophysical survey and small-scale evaluatory excavation. Dateable material in any quantity is not expected but the potential for 14c sampling will be explored for future sampling.

 To train and engage the members and volunteers of the community in archaeological techniques and research.

Geophysical Survey

A total of 2.5 acres was surveyed encompassing both the enclosure banks and south, to the base of steeply rising ground. A base line was established along the longest axis of the field, roughly WNW - SES and this was marked at both ends by a permanent peg in order to allow for any subsequent reconstruction of the survey grid. The base line was laid out to encompass the maximum extent of the site and at a diagonal to any expected archaeological alignments to allow survey across, rather than along, any linear features. A series of 30 metre squares were then established perpendicular to this base line using hand tapes, and the corners of each grid square were marked with temporary wooden canes. The position of the baselines and grid squares were then recorded using a handheld Garmin Legend GPS to allow accurate input to GIS. Handheld GPSs of this kind typically have a precision level of between 8-30ft but are routinely used to locate geophysical prospection surveys of this kind. With any type of field survey there is always a tension between the level of precision required and the time and resource spent achieving it. As such, the survey location maps only provide an approximate guide to the exact location of features with a maximum error tolerance of 8- 30ft over each individual grid. The method of ‘walking’ survey employed by

10 this kind of geophysical technique itself is only as accurate as the operator’s judgment along each 30 metre traverse and can typically generate errors

Fig. 2. Geophysical survey shown with positions of trenches. along each traverse of up to 2 metres in the traverse direction. It is usually recommended therefore that excavation of any archaeological features is preceded by a higher resolution geophysical survey carried out using the excavation grid to allow for the utmost accuracy. In practice this is usually forgone and a degree of tolerance is accepted.

The survey was carried out using a Bartington magnetic gradiometer 601 in the dual configuration to allow for rapid and accurate survey. This technique is

11 ideal for the prospection of ‘noisy’ archaeological areas that might contain built structures with both domestic occupation and industrial activity. It is also capable of detecting more subtle features such as ditches and former river channels.

Each 30m grid was surveyed at a traverse interval of 1m with a sampling interval of 8 readings per metre in a zigzag format in order to give high resolution results. The data was downloaded in xyz format and processed using Geoplot v.3. The data was subsequently adjusted to reduce striping and are reproduced here as greyscale bitmaps of readings clipped to a range of +6 nT (nanotesla).

The survey shows both the inner and outer banks as a dark curving feature. A number of point anomalies are also shown as small black and white patches. This area is known to be heavily littered with rifle shells and this undoubtedly accounts for much of the ‘noise’. Central to the inner enclosure are two larger point anomalies measuring up to 3m across and some 17m apart. The nature of these readings suggests burning and it is tempting to speculate that these may be hearths as they are situate on a level area within the enclosure, (though later burning through agricultural clearance cannot of course be ruled out). Even at such a high resolution it would be fortunate to detect smaller post holes but the likelihood of roundhouse structures within this enclosure remains high.

Excavation

The farm is subject to a Tir Gofal agreement (E/10/0343 , CPAT HE1) and the site sits within an area of species-rich grassland which is essentially unimproved pasture. In order to preserve this habitat, topsoil was retained for replacement and the trenches were backfilled at the end of the project.

Two trenches were opened c 2 x 8 metres. One was positioned over the inner rampart (A) and another (B) over the outer bank and ditch (FIG). This was in

12 order to investigate the nature and construction method of the two enclosing banks and to investigate any corresponding ditches.

Trench A

Trench A (App. 2) was excavated to a maximum depth of 70 cm. Topsoil (001) up to a maximum 25 cm deep was indistinguishable from the ploughsoil (002) beneath which varied from loose to soft comprising a well-sorted medium to fine sand. Few find were recovered from these contexts. A brass

A

B

Fig 3. Trench location map (overlain onto the earthwork which has been mapped through aerial survey by Tom Pert of RCAHMW) rifle shell is apparently typical of this region that continues to see military training and many such finds have been reported from metal detectorists in

13 Fig. 4 Trench A. The stone rampart (facing north, scale 2m, photo J Bezant).

this field. A small amount of flint flake and debitage pieces are probably residual and are likely to have been imported with field lime from elsewhere

Fig. 5. Trench A. The stone rampart. Roughly coursed, red sandstone slabs (Facing north, scale 20 cm, photo J Bezant).

14 as part of an improvement regime. The soil here is fertile. This overlay a stony spread (003) at the northern end of the trench. This comprised the tumble from a dry-stone constructed stone rampart which remained up to 1m high and 5m thick in places. This red sandstone material acts as excellent building material and fractures naturally into useful flat slabs.

These were recovered up to 80 cm in length where they had been laid lengthwise in rough courses to effectively form a faced wall which was filled by smaller rubble pieces. Substantial quarrying on an outcrop immediately south of, and above, the enclosure has clearly been the source for the post- medieval extensive commons walling and undoubtedly provided the material for this rampart. The rampart, subsequent tumble and the ploughsoil contexts are stratified above a compacted surface that is likely to represent the contemporary land surface. This was not excavated.

Trench B

Trench B (App. 2) was excavated to a maximum depth of 1.4 metres where the base of the ditch was reached. Topsoil (001, see app. 4 for context record sheets) and ploughsoil (002) had the same characteristics as trench A but the ploughsoil was much deeper here up to a maximum depth of 70 cm. Again, several pieces of field lime, burnt stone material and flint flake were recovered. The boundary with the top of the ditch fill (007) was indistinct and this boundary produced a small ceramic sherd (App. 3) of uncertain date. The ditch (006) was cut into the former land surface (005) and had sloping sides with a ‘U’ profile and a roughly level but uneven base. A ditch with a similar profile was excavated at Arddleen in where a double-ditched enclosure contained evidence for roundhouses dating to the later Iron Age (Grant 2004).

The length of the trench was not sufficient to allow the entire ditch profile to be recovered but it was 90 cm deep and estimated to have been c.8m wide. Any boundaries between stratified ditch deposits were difficult to discern but rubble stone (008) at the base may be the primary deposits. This stony

15 material sat in a wetter, silty matrix that has been sampled for possible environmental analysis at a later date and the potential for survival of organic material has been noted. An earthen bank had been thrown up towards the interior of the enclosure but this has been substantially spread and eroded, surviving to a height of c. 50cm and a width of up to 4 m.

Fig. 6 Trench B. Former land surface and ditch (Facing southeast, scales 2m, photo J Bezant).

The relationship between the two enclosing structures remains unknown but the differing construction techniques make it not unreasonable to suggest that they represent different phases of use, with the outer earthen bank and ditch preceding the construction of the stone rampart.

16 Discussion

Lan Fawr is of a type described by the RCAHMW (1997ii:19) as a multivallate enclosure with wide-spaced ramparts. Directly to the north of Lan Fawr lies the prominent valley floor ridge known as the Myarth (SO 17302070) rising some 297 above OD. A partly bivallate enclosure on the eastern edge of the scarp is shown on the 1st edition 1” OS map but any remains of this are difficult to locate now and the site has been heavily re-forested (ibid. 83). This would have been one of the largest such sites in the county measuring 354 by 210m and its proximity to Lan Fawr, and its command of the valley routeways exploited into the Roman period portrays its significance . Also prominently visible to the east is a heavily defended bivallate spur known as Crug Hywel (SO 22552065) which lends its name to the town of Crickhowell below (ibid. 119-123). The site measures 162 by 59m and was enclosed by a stone wall but is also partly defended by a rock-cut ditch and steep stony scarp.

The hilforts and other enclosures of this region generally comprise a varied but poorly dated group and include enclosures that, although defended, would have had a domestic rather than a purely military function. These smaller enclosures may have been built by single families or larger groups of extended kin. Domestic, undefended valley floor settlements exploiting fertile valley soils can be inferred though they remain to be discovered. This type of ‘homestead’ may have had its origins around the late Bronze Age and continued into the Roman Period. Though they remain poorly dated as yet, this area of southeast Wales was home to the Silurian tribe and contemporary accounts relate a sizeable population (Burnham 1995:77). The ramparts of at Aberllynfi near Hay on Wye, were slighted during the campaign of Julius Frointinus around AD 75 and the three Roman forts in Breconshire all relate to this campaign. Although there is little evidence for civilian settlement during the late Iron Age through into the early centuries AD, expanded arable cultivation and tree clearance in the immediate post-Roman period is attested by pollen cores taken from (RCAHMW 1986: 1).

17 During the late Bronze Age through the early Iron Age there is evidence for a climatic downturn to wetter, cooler conditions and this ameliorated from the 5th century BC until the post Roman period. A shorter grass-growing season would have meant pressure on grazing regimes, a lowering of the tree line and the reduced viability of crops. Pasture quality would have also lowered due to podsolisation and peat formation in the uplands (ibid.6). Most of the Breconshire Iron Age sites are located in the Usk-Llynfi Basin within easy reach of reasonable quality land and water. Sites on hillslopes generally seem to be sited at the transition between valley slopes and the open upland and this certainly seems true at Lan Fawr, indicating a mixed economic strategy, perhaps with transhumant practice, grazing cattle on the open uplands.

The nearest comparable site to Lan Fawr is at Penfawyddog some 1.5 miles due east along the ridge at SO 19571845. A small univallate enclosure it is located on the edge of a north-facing slope and with rising ground to the south like Lan Fawr. The site measures 84 by 67 m and is enclosed by a stone and earth bank (ibid. 82). Just under 1.5 miles south of this lies a larger bivallate enclosure at Coed Pen-twyn SO 19351622 (ibid. 78). This enclosure comprises mainly a stone-built rampart and measures 140 by 103m on the edge of a steeply defended scarp.

A small, subrectangular univallate enclosure SO 11172095 stands on the north-facing slope some miles south of Talbont on Usk, nearly 4 miles northwest of Lan Fawr along the same ridge. Pantywenallt measures 53 by 41m (ibid. 84). Quarter of a mile north of this stands a larger oval bivallate enclosure on the summit of a prominent hill at Tump Wood SO 11272149 measuring 120 by 69m and enclosed by earthen, stone-faced banks (ibid. 85). Cross Oak fort lies just north of Talybont overlooking the Usk at SO 11052355 and is univallate lying on a small knoll measuring 105 by 95m. Coed y Gaer (SO 17622402) lies due north of Lan Fawr on the opposite side of the valley and is sited on the end of an elevated spur overlooking Cwmdu (ibid.88). It is a small univallate site but on a well-defended highly visible knoll. West of this is Allt yr Esgair (SO 12312420) which is a small univallate enclosure sited on the edge of a steep southwest facing slope (ibid.90). On the summit of this

18 ridge a very large multivallate fort guards the ridge between the Usk and the Llynfi measuring 566 by 114m at SO 1270 2430. Lan Fawr also has distant views of the large, complex Dinas Fawr to the north at SO 1789 3008 (96).

Future research questions

The late Iron Age period in south Wales was undoubtedly a productive landscape populated by numerous groups or small communities. The precise nature of the relationships between these different communities and their resources remains opaque but following the recommendations made by the RFAW, the following research questions can be addressed for future investigations at Lan Fawr:

 How were these farming communities organised? Although well- defended and highly visible, it is likely that the Lan Fawr represented the focus of an agricultural community situated at the transition between the upland-lowland.

 Is this site typical of the material-poor south-east Wales community? Special attention must be made to the recovery of dateable material.

 What can the morphologies of such sites within their landscape settings tell us about their context? The highly visible nature of Lan Fawr lends itself well to line of sight studies for example, and the analysis and comparison with other sites within GIS is potentially important.

19

References

Browne, D M & Percival, D, 1988, Lan Fawr, Llangynidr, Archaeology in Wales Vol 28, 45.

Burnham, H. 1995 A Guide to Ancient and Historic Wales; Clwyd and Powys. London: HMSO for Cadw.

Grant, I. 2004 The Excavation of a double-ditched enclosure at Arddleen, Powys, 2002-03. The Montgomeryshire Collections Vol 92:1-32.

RCAHMW , 1997 , An Inventory of the Ancient Monuments in Brecknock (Brycheiniog); The Prehistoric and Roman Monuments, Part II.

20 Appendix 1

CPAT Regional Historic Environment Record

PRN 17470 - Lan Fawr Enclosure NGR :- SO17071838 (SO11NE) Unitary authority :- Powys Community :- Llangynidr Prefered site type :- Iron Age - Defended enclosure (Structure - Near destroyed )

Spur-sited earthwork comprising annular inner bank and partly concentric outer bank and ditch. Sequence of later bounderies overlying earthwork. Inner bank comprises low bank 0.5m high. Enclosure measures 78m e-w x 30m. Possible remains of outer defensive system. Plan has general similarity to later prehistoric multivallate enclosures (BROWNE, D.M. & PERCIVAL, D., 1988, 45).

(HF 65;fig.134a/b) On a spur bordering pastures to the ? of LAN FAWR at a height of 350 m above O.D. is an enclosure comprising an annular inner bank and probably a partly concentric outer bank and ditch. Overlying or immediately adjacent is a sequence of later boundaries, the latest of which are modern wire fences. A stone, partly ruined commons enclosure wall, lies to the S.W., standing up to 1.8 m high and constructed in a mixed build of coursed sandstone slabs and roughly coursed sub-angular boulders, with a 'soldiered' crown. Where there are gaps these have been closed by iron hurdles and post and wire fencing. There is an even earlier field system comprising ruined, structureless walls constructed of small sub-angular sandstone boulders, the best part of which runs along the crest of the inner bank. At the W. end of this wall there is an entrance through the gap created between it and a further wall running N.-S. There are two further gaps on the E. which may also have been entrances, and a more modern opening has been created by farm traffic on the S.E. The wall is extensively robbed, creating hollows, and elsewhere, the wall is merely a low bank of grassed rubble. A small undated horseshoe-shaped quarry is cut into the upper part of

21 the N. scarp slope of this spur. In the field to the E. of the site, close to and roughly parallel with the field wall shown on the plan, are two scarps which may be artificial and probably mark the limits of ploughing. The inner bank of the older earthwork consists of a broad, low, bank, of grass-grown rubble, up to about 0.5 m high, in places covered with bracken. The bank, indistinct from vehicular erosion, is punctuated by occasional small hollows caused by stone robbing. Mutilation is particularly pronounced on the W. The bank terminates on the N.E. and N.W., coinciding with a sharp steepening in the natural slope of the spur. The only possible entrance site is at the S.E. corner, where there appears to be a re-entrant in the bank's outer line. Downslope of the N.W. terminal is an accumulation of grass and bracken-covered rubble talus. There is a low, grass-covered rubble mound protruding from the bank near the S.W. corner, though its origin is unclear. Alongside the inner toe of the bank, on the S.W., is a very shallow hollow, passibly the last remaining indication of a quarry ditch. There is no sign of an outer ditch associated with the bank. A low linear swelling of the ground up to 0.45 m high, and an associated hollow to the S.W. between 8 m and 15 m in front of the latter bank are probably the remains of an outer defensive system. No interior features are visible. The dating and cultural affinities of the site are uncertain, but there is a general similarity in plan between later prehistoric, multivallate enclosures with concentric, wide-spaced ramparts. D.M.Browne and D.J.Percival, Arch.in Wales 28 (1988), p.45.

Sources:- Browne, D M & Percival, D survey , 1988 , , PLAN RCAHMW , 1995 , An Inventory of the Ancient Monuments in Brecknock (Brycheiniog); The Prehistoric and Roman Monuments, Part i - draft text , RCAHMW , 1997 , An Inventory of the Ancient Monuments in Brecknock (Brycheiniog); The Prehistoric and Roman Monuments, Part i , Waller, C M , 2000 , Tir Gofal Management Plan: Heritage Management Information (HE1). Dan-y-wern Farm - E/10/0343 , CPAT HE1 .

Events: Visit RCAHMW / 1988(01/09/1988) - record created 15/11/1999 Alex Gibson - copyright CPAT , last updated 29/08/2002

22 The above data are supplied by CPAT in partnership with its Local Authorities and the partners of END, © CPAT SMR partnership, 2007 (and in part © Crown, 2007 - as indicated)

CM - 05/06/2007 ( 15:02:53 ) - HTML file produced from CPAT's Regional HER Clwyd Powys Archaeological Trust, Curatorial Section, 7a Church Street, Welshpool, Powys SY21 7DL. tel (01938) 553670 , fax (01938) 552179, email [email protected] , website www.cpat.org.uk

23 Appendix 2

Section drawings, trenches A and

24 25 Appendix 3

Ceramic sherds

26 Appendix 4

Excavation site record

27