An Early-medieval Monastic Cemetery at
Llandough, Glamorgan: Excavations in
1994
By NEIL HOLBROOK and ALAN THOMAS
EXCAVATIONS ADJACENT to the church of St Dochdwy in Llandough, near CardiV,
examined part of a large monastic cemetery. In all 1,026 inhumation burials were examined, by
far the largest Early-medieval burial population so far recovered from Wales. Sherds of imported
Bii amphorae were found in the backfill of five graves indicating activity at Llandough in the late
5th or 6th century. Radiocarbon dates indicate that burial had commenced by the mid-7th century
at latest, and it appears that the cemetery continued in use until the demise of the monastery in the
late 10th or early 11th century. A Romano-British villa lay to the south of the church, and it is
conceivable, although unproven, that there was continuity of settlement from the Roman Period
onwards.
The village of Llandough lies to the north of Penarth and about 3.5 km from the
centre of Cardiff (Fig. 1). It occupies gently sloping ground near the crest of an
escarpment, which overlooks the estuary of the River Ely. The present church of
St Dochdwy largely dates to the mid-19th century, but has long been considered to
overlie the site of one of the major Early-medieval monasteries of Glamorgan
recorded in the Llandaff charters. A fine late 10th-or early 11th-century carved
cross that still stands in the churchyard provides further evidence of the antiquity
of the site.
In 1990 an area immediately to the north of the churchyard was proposed for
residential development (NGR ST 16817331). The Vale of Glamorgan Borough
Council granted outline planning permission for development with a condition
requiring archaeological evaluation given its proximity to the presumed site of the
monastery. In 1990 Glamorgan-Gwent Archaeological Trust (GGAT) excavated
eight trenches across the proposed development area. The results suggested to the
excavators that large parts of the site had been disturbed by post-medieval
quarrying and levelling. Nearer the churchyard wall disarticulated human skeletal
remains, which appeared to have been disturbed from their original context, were
located. This part of the site was designated by GGAT as being of medium-high
archaeological potential.1 In March 1992 full planning permission was obtained
1 Glamorgan-Gwent Archaeological Trust (GGAT), Land at Great House Farm, Llandough: An Assessment of
the Archaeological Implications of Redevelopment at the Site of Great House Farm, Llandough (unpub. rep.,
1990); ‘Llandough’, Archaeol. Wales, 30 (1990), 66.
1
DOI: 10.1179/007660905x54044
2 neil holbrook and alan thomas
fig. 1
Site location.
a monastic cemetery at llandough 3
for the residential development, with no condition requiring further archaeological
work. Generously the developers, Ideal Homes Wales Ltd (now part of Persimmon
Homes), voluntarily undertook to sponsor an excavation of that part of the site
which was considered to be of medium-high archaeological potential. The total
area to be examined was approximately 0.1 ha, and in March 1994 Cotswold
Archaeological Trust (CAT; now Cotswold Archaeology) was commissioned to
undertake the excavation.
At an early stage it became clear that the site contained numerous burials that
continued beyond the bounds of the agreed excavation area. Consequently the
excavation was extended in order to establish the extent of the cemetery within the
development site. In the event an area of 0.22 ha containing a total of 1,026 burials
was investigated and work finally ceased on site in September 1994. A preliminary
typescript report outlining the principal excavation results was issued later that
year.2 It was readily appreciated that the costs of full stratigraphic and osteoarchae-
ological analyses of the excavated evidence would be considerable, and that this
was not a cost which the developers could legitimately be asked to bear (Ideal
Homes Wales had already increased their financial contribution more than five-
fold from the original project budget to cover the costs of excavating the burials).
Given the evident importance of the discoveries, the desirability of full post-
excavation analysis was discussed by the Ancient Monuments Board for Wales,
and in January 1998 Cadw: Welsh Historic Monuments generously provided
funding for the stratigraphic analysis of the site, and the National Museums &
Galleries of Wales agreed to produce and co-ordinate reports on the artefacts. In
October 1998 Louise Loe was awarded a Ph.D. scholarship from the University of
Bristol to undertake the osteological examination of the skeletal assemblage, for
which a doctorate was awarded in 2004.
previous archaeological work in llandough
Although the potential antiquity of the site occupied by St Dochdwy’s church
has long been recognised, no systematic investigation occurred in the village until
an area to the south of the church was excavated by the Barry and Vale
Archaeological Group in 1963 (Fig. 2).3 In 1979 further excavations were
conducted in this area by GGAT when the earliest structure found was a possible
late Iron-age roundhouse, perhaps part of a farmstead.4 In the 2nd century a.d. a
Roman villa was constructed, which developed considerably in the early 3rd
century and was occupied until the early 4th century. The fact that some of the
walls of the villa were reused as foundations for a 12th- or 13th-century barn was
the only evidence identified that occupation may have continued on the villa site
into the post-Roman period.5
2 A. Thomas and N. Holbrook, Excavations at Great House Farm, Llandough, Cardiff, South Glamorgan: A
Preliminary Report (unpub. rep., 1994).
3 E. J. Beare, Llandough, Penarth (unpub. MS., Dept. Archaeology, National Museum and Gallery, Cardiff,
1963).
4 H. Owen-John, ‘Llandough: the rescue excavation of a multi-period site near Cardiff, South Glamorgan’,
123–77 in D. M. Robinson (ed.), Biglis, Caldicot and Llandough: Three Late Iron Age and Romano-British Sites in South-East
Wales. Excavations 1977–79 (BAR Brit. Ser., 188, Oxford, 1988).
5 Ibid., 152.
4 neil holbrook and alan thomas
fig. 2
Location of the 1994 and earlier excavations.
a monastic cemetery at llandough 5
Building operations outside of the main excavation area disturbed a number
of burials, probably at least 30, and two of these were recovered through salvage
recording. They were dated by radiocarbon determination to the period spanning
the late 8th to late 10th century (Tab. 8). A further burial within the excavation
area was dug through a demolished wall of the villa and produced a radiocarbon
date of 999–1284 cal a.d.
At the close of the 11th century the village of Llandough was granted to
Tewkesbury Abbey, and a grange was subsequently established here. This
remained in Tewkesbury’s ownership until the Dissolution when it was confiscated
by the Crown and sold.6 A dovecote and barn, possibly of late 12th-century date,
overlay the villa site and may be part of the grange. These were associated with
pits, crude areas of paving, and sections of walling. The corner of a substantial
12th- or 13th-century house had previously been uncovered in 1963, when a gully
containing three human skeletons of possible 13th-century date was also recorded.7
Further structures were revealed during the 1979 watching brief in the same area.8
Jeremy Knight discusses the historical evidence relating to the monastery at
Llandough in a separate article in this journal, and that relating to the post-
Conquest period is discussed elsewhere.9
THE EXCAVATION SITE
The present churchyard of St Dochdwy’s occupies a raised position, with the
escarpment overlooking the Ely estuary to the north, and a combe occupied by a
stream to the south. The 1994 excavation site lay immediately north of the
churchyard wall, and extended as far as the edge of the escarpment (Fig. 2). The
surface geology of the escarpment is made up of Lower Lias overlying rocks of the
Rhaetic series and Red Marls. Within the excavation area the lias took the form of
thin alternating bands of rock and yellow clay, sloping gently to the south and west.
The site lay at approximately 43 m above sea level. Prior to the start of the
excavation this area was waste land covered with rough vegetation. It had
previously been occupied by the farmhouse and outbuildings of Great House
Farm, which had been comprehensively demolished in 1988. The farmhouse dated
largely to the 17th century, although it may have incorporated some earlier fabric.
The present wall of the churchyard matches the masonry in the church, which
was rebuilt in the mid-19th century.10 The chancel arch is said to be Norman.11
Diane Brook has charted the development of the churchyard in an unpublished
M.A. thesis: the current boundary was in existence by 1878, but had developed
from a churchyard of more rectilinear form depicted on maps of 1786, 1824 and
1846 (Fig. 3).12
6 H. J. Thomas, ‘Historical Notes’, 177 in Owen-John, op. cit. in note 4.
7 Beare, op. cit. in note 3.
8 Owen-John, op. cit. in note 4, 155.
9 Thomas, op. cit. in note 6.
10 C. J. O. Evans, Glamorgan: Its History and Topography (Cardiff, 1938), 282.
11 S. Glynne, ‘Notes on the older churches in the four Welsh dioceses’, Archaeol. Cambrensis, 6 ser., 1 (1901),
245–78, at pp. 258–9.
12 D. L. Brook, Early Ecclesiastical Sites in South-East Wales (unpub. M.A. Thesis, University College, Cardiff,
1981).
6 neil holbrook and alan thomas
fig. 3
Cartographic representations of the boundary of St Dochdwy’s churchyard. Sources: 1786 and 1824, Estate
Maps. 1846, Tithe Map. 1878, 1st Edition O.S. The maps appear to show that the churchyard formerly had a
more rectangular form. After D. L. Brook, Early Ecclesiastical Sites in South-East Wales (unpub. M.A. Thesis,
University College, Cardiff, 1981).
excavation and post-excavation methodology
During removal of the topsoil and overburden by a mechanical excavator
equipped with a toothless grading bucket, it became clear that numerous
articulated burials lay within, and outside, of the area of initial excavation. As a
result it was agreed with the developers that, while hand excavation of the original
area was continuing, machining should continue to the west, north and east in an
attempt to identify the extent of the cemetery within the development site.
Following the completion of this process it was agreed that the excavation should
be phased with a view to allowing the building contractors access to those parts of
the site where archaeological work was complete. As a result excavation resources
were initially transferred to the line of a road, which was to be built through the
site. Following the removal of the burials along its route excavation then
concentrated on the burials to the south of the road, and lastly the area to the north
of the road.
a monastic cemetery at llandough 7
All of the burials were excavated in accordance with the guidelines prepared
for the Institute of Field Archaeologists.13 The burials were recorded on pro forma
sheets following standard CAT procedures. Every skeleton was recorded by vertical
black-and-white and colour-slide photography. The drawings of the burials in this
report have been produced from the photographic archive. Features such as pits
and postholes were half-sectioned and ditches sampled at appropriate points along
their length. In January 1998, when the post-excavation programme commenced,
details of every grave were entered on to a computerised database using Microsoft
Access software. For each individual grave a single sheet has been compiled
detailing the contextual data; stratigraphic relationships; grave-type; orientation;
posture; arm, leg and skull positions; and finds information. A fully computerised
plan of the cemetery has also been compiled using Microstation software. The plan
and database were then linked using MapInfo software which allows full
distribution plots of the information on the sheets and finds to be cross-referenced
and analysed.
Washing and processing of the skeletons took place in 1996–2001. Initially
different parts of each skeleton had been bagged separately in accordance with the
guidelines, but during the processing of the bones the osteological advisor
recommended that the bones from each individual be amalgamated into one bag.
Osteological recording of each skeleton was undertaken in the Rheumatology Unit
of Bristol University by Louise Loe as part of her doctoral research between 1998
and 2001.
purpose of this report and associated publications
This report is one of series of linked publications on differing aspects of the
cemetery. It is designed to be read as a free-standing account of the excavations,
and contains full finds reports. It is considered important, however, that a detailed
catalogue of all burials should be made widely available to facilitate future analysis
and review of the conclusions presented here. A digital archive has therefore been
prepared and this is hosted by the Archaeology Data Service. It can be freely
accessed at http://ads.ahds.ac.uk/catalogue/resources.html?llandough_cadw_2004 The
digital archive comprises a full listing of each grave, including stratigraphic
relationships, age and sex of the skeleton, data on body posture and arm positions,
and full listings of any artefacts contained within the grave-fill. The database is
linked to a digital site plan which allows details of a particular grave, and its
location within the cemetery, to be quickly found. The archive also possesses a
search facility which allows users to generate their own distribution plots of specific
attributes, either singly or in combination. This report contains a short summary
account of the human remains which focuses mainly on the demographic
characteristics of the cemetery population. A full report on the palaeopathology of
the Llandough cemetery will be published elsewhere.14 Finally a wide-ranging
13 J. I. McKinley and C. Roberts, Excavation and Post-Excavation Treatment of Cremated and Inhumed Human Remains
(Institute of Field Archaeologists Technical Paper, 13, 1993).
14 L. K. Loe, Health and Socio-Economic Status in Early Medieval Wales (BAR Brit. Ser., forthcoming).
8 neil holbrook and alan thomas
fig. 4
General view of burials in the northern
part of Area III. © Cotswold Archaeology.
discussion of the cemetery and its context can be found in Jeremy Knight’s paper
which follows this in this journal.
ANALYSIS OF THE CEMETERY
preservation/scale of disturbance
The original ground level of the cemetery had been truncated at some stage
following its abandonment. Over much of the excavation area, particularly the
north and west parts, articulated skeletons were found a matter of only a few
centimetres below modern ground level (Fig. 4). Indeed although no burials were
found in the north-east corner of the excavation, and only a few in the north-west
part of the site, it is almost certain that they had once existed in these areas but had
been completely removed by later truncation. The plan as produced (Fig. 5) cannot
therefore be regarded as a true reflection of the original distribution of burials
within the excavated area. Much of the levelling was presumably associated with
the construction of Great House Farm, of which some walls survived within the
a monastic cemetery at llandough 9
excavation area. Evidence was also recovered of post-medieval quarrying (p. 39).
One deep quarry bisected the excavation and others were exposed to the west of
the excavation area during the construction of two houses. In the southern part of
the excavation area nearest to the churchyard wall a dark brown loam 007
overlay the natural clay, possibly the remnants of a former cultivation soil.
In addition to the subsequent disturbance of the site, the intensive use of the
cemetery led to a large degree of intercutting which further reduced the level of
preservation. Despite these difficulties a total of 814 inhumations was excavated
and recorded, along with a further 212 disarticulated bone groups. Prior to the
processing of the skeletons a rapid assessment on their state of preservation was
carried out (Tab. 1). The assessment was based on the approximate percentage of
bones per skeleton remaining, and shows that most of the burials were either well
preserved (category A) or poorly preserved (categories D and E). The latter two
categories mostly comprise burials that had been disturbed by later interments.
The disarticulated bone groups were mostly found within the fills of later graves.
Of the 814 inhumations recorded on site, 801 were identified as discrete skeletons
during detailed osteological examination (see below).
Table 1
PRESERVATION OF THE SKELETAL
ASSEMBLAGE BY PROPORTION OF
BODY SURVIVING (FROM SITE
RECORDS)
Category Total
A (70–100%) 425
B (50–70%) 77
C (30–50%) 73
D (10–30%) 126
E (under 10%) 113
Bone groups (disarticulated) 212
Total 1,026
introduction to the analysis
N.B. In the following sections references to burials are prefixed by the letter B. The symbol
* indicates that a burial is illustrated. The letters and numbers in italics refer to the sex and
age of the burial, where M = male; F = female; U = uncertain sex; YC = young child
(0–<6); C = child (>6–<18); YA = young adult (18–25). Table 9 presents a full
classification of the various age-groups to which the skeletons have been ascribed.
In seeking to analyse the cemetery data a series of traits have been defined, and
these are quantified in Table 2. The cemetery was provisionally divided into three
areas during the initial post-excavation programme (Fig. 6). This division was
necessarily somewhat arbitrary as no features such as boundary ditches or walls
contemporary with the use of the cemetery were found in the excavation. This
deficiency is exacerbated by the absence of vertical stratigraphy due to the
10 neil holbrook and alan thomas
fig. 6
Plan of the cemetery showing burials selected for radiocarbon dating and spatial divisions referred to in the
text.
truncation of the site. The division was therefore made purely on the basis of the
spatial distribution of the burials. During subsequent post-excavation work,
analysis of the artefacts and the results of a selective programme of radiocarbon
dating have permitted some conclusions to be drawn on the layout and
development of the cemetery. Area I comprised those burials contained within a
putative curvilinear boundary (no trace of the boundary survived but its existence
was suspected from a concentration of graves on a distinct NE.–SW. alignment;
see below p. 88). In the west part of the site rows of occasionally intercutting
inhumations were found, which for convenience have been termed Area II. To the
north of Area I there were further burials, and indeed they may continue to the
south-west to cut the original burials in Area II. These burials have been termed
Area III.
Types of Burial
All of the burials at Llandough were inhumations, and almost all were single
burials within individual grave cuts. There were, however, a small number of
exceptions. There was a single instance of an adult male burial (B18, M, 45+)
placed directly on top of a child (B24, C) in the same grave cut. There were also
two instances, one in Area I (B10, M, 40, and B11*, M, 40) and the other in Area
III (B206*, C, and B212*, C), where two male adults and two children were placed
side-by-side. The filling of the grave for B10/11* contained a fragment of 13th- or
Table 2
CONCORDANCE OF BURIAL TRAITS AND ARTEFACTS.
SKULL POSITION: U = FACING UPRIGHT; L = FACING LEFT, R = FACING RIGHT, AND INDET = INDETERMINATE
Skull position Orientation Pottery An. Cu/ Fe Fe Fe Coin Glass Glass Bone Pebble Flint Slag
Bone Pb Hob- other knive Ves- Bead Object
nail sel
Trait Total U L R indet E.–W. NE.– N.–S. NW.– IA/E. Rom Early Later Post
SW. SE. Rom med. med med
Arm A 95 5 19 28 43 40 42 1 12 1 10 1 2 1 42 0 0 10 1 0 1 0 0 6 0 5
Arm B 73 11 16 25 21 30 37 0 6 1 10 2 0 1 38 2 1 10 0 1 0 2 0 3 1 2
Arm C 55 3 17 9 26 32 18 0 5 1 4 1 0 0 24 0 0 5 0 1 0 1 0 4 0 4
Arm D 25 1 6 8 10 12 6 0 7 0 7 0 2 0 15 0 1 2 1 0 0 0 1 0 0 0
Arm E 4 1 2 1 0 3 1 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0
Arm F 10 1 2 3 4 5 3 0 2 1 2 0 0 0 5 1 0 1 0 0 0 0 0 0 0 0
Arm G 17 1 2 3 11 8 6 0 3 0 3 0 0 0 8 0 0 1 0 0 0 1 0 4 1 0
Arm H 6 2 2 0 2 3 3 0 0 0 0 0 0 0 4 0 0 1 0 0 0 0 0 0 0 0
Arm I 10 0 4 2 4 6 4 0 0 0 0 0 0 0 4 0 0 0 0 0 0 0 0 1 0 0
Arm J 2 0 1 0 1 0 2 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
Arm K 1 1 0 0 0 1 0 0 0 0 1 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0
Arm L 1 0 1 0 0 0 0 0 1 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 1
Arm M 1 0 0 1 0 0 0 0 1 0 0 0 0 0 0 0 0 1 0 0 1 0 0 0 0 1
Arm N 1 0 1 0 0 1 0 0 0 1 0 0 0 0 1 0 0 0 0 0 0 0 0 1 0 0
Arm O 1 0 0 0 1 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
Arm P 57 8 16 10 23 30 21 1 5 3 8 0 1 1 23 0 2 4 0 1 1 0 0 6 0 0
Arm Q 17 3 1 7 6 6 9 1 1 1 5 0 0 0 7 0 1 1 0 0 0 0 0 1 0 2
Stone 16 3 5 4 4 8 6 0 2 1 5 0 0 0 10 0 1 5 0 0 0 0 1 1 0 2
packing
Ankles 8 0 4 2 2 2 6 0 0 1 3 0 0 0 4 0 0 0 0 0 0 0 0 0 0 0
crossed
CoYn 25 0 6 9 10 14 11 0 1 1 2 1 2 0 7 0 1 2 0 2 0 0 0 2 0 1
Shroud 9 1 3 2 3 5 4 0 0 0 0 0 0 0 2 0 0 1 0 0 0 0 0 1 0 1
a monastic cemetery at llandough
Marker 3 0 1 1 1 1 2 0 0 0 1 0 0 0 2 0 0 1 0 0 0 0 0 0 0 0
Pillow 6 3 0 1 2 5 0 0 1 0 1 0 0 0 5 0 0 0 0 0 0 0 0 0 0 0
Flexed 4 0 2 0 2 1 2 0 1 0 1 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0
Prone 1 0 1 0 0 1 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 1
Crouched 1 1 0 0 0 1 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0
Pit 1 0 0 1 0 0 0 0 1 1 0 0 1 0 0 0 0 0 0 0 1 0 0 0 0 0
Right-side 1 0 0 1 0 0 0 0 1 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0
11
12 neil holbrook and alan thomas
fig. 7
Selected burials B2–B66.
a monastic cemetery at llandough 13
14th-century roof tile (although cf. p. 32). There was also an instance in Area II
(B265*, M, 45+, and B267*, fetus) where an adult and a fetus, or conceivably a
new-born infant, were buried in the same grave. Double burials occur rarely in
Late Roman cemeteries, but were more prevalent in the early post-Roman
cemeteries at Cannington (5% of the population) and Henley Wood, Somerset
(where there were five examples, including an adult associated with a juvenile).15
Age and Sex Distributions
Of the population available for study, 30% was either certainly or probably
male, 25% certainly or probably female, and 26% children under 18 (Figs. 18–19).
The distribution of male, female and child burials is shown in Figure 12, and more
detailed information is available in the digital archive. It would appear from the
overall plan that the adult graves were not grouped according to age and sex, but
that all areas of the cemetery contain mixed burials. Area III contained a distinct
concentration of child and young child burials, which were scattered throughout
the area, although two adjacent clusters could be recognised. The burials within
the clusters were aligned E.–W. These juvenile burials cut through adult
inhumations, with the solitary exception of B568, which was partially overlain by
the adult B562, and so represent the final use of this part of the cemetery. Within
all parts of the cemetery it is possible to discern localised concentrations or possible
alignments of the same sex, although it is doubtful how much significance should
be attached to these observations. No obvious patterning could be recognised
within the different age categories of the adult burials.
Body Positions
The vast majority of the inhumations at Llandough were extended and supine,
although there were rare examples of differing postures. One burial (B233) had
been placed on its right-hand side, and four burials were flexed (B55, B68, B333*
which produced a radiocarbon date of 610–777 cal a.d., and B652). Of these B68
and 652 were young children, while B55 and B333* were adults. The flexed burial
of a young man was found dug into the ruins of the Roman villa to the south of the
church: it produced a radiocarbon date of 999–1284 cal a.d.16
A single prone burial (B470, C) was found in Area III. Finds from the grave
included two lead balls of shot which must be intrusive. Nearby was B474* (F, 30)
who had been laid on her back with the knees drawn up to her chest. A pit-burial
(B601*, YA) was also found in the same area (Figs. 10, 14C). This burial was unique
in that the upper torso had been laid on a slope of approximately 45o, suggesting
that it had been placed in a pre-existing pit which was at least 0.48 m deep. The
eastern part of this pit, which must have contained the legs of the burial, had been
removed by modern activity. The grave fill contained three sherds of pottery: two
from a late Iron-age/Early-Roman calcite-gritted jar, the other from a possible
15 A. B. Woodward, ‘Part 3: Discussion’, 215–39, in D. E. Farwell and T. I. Molleson, Poundbury Vol. 2: The
Cemeteries (Dorchester, 1993), 227; L. Watts and P. Leach, Henley Wood: Temples and Cemetery Excavations 1962–69
(CBA Res. Rep., 99, York, 1996), 51; P. Rahtz, S. Hirst and S. Wright, Cannington Cemetery (London, 2000), 98.
16 Owen-John, op. cit. in note 4, 144, pl. 14.
14 neil holbrook and alan thomas
a monastic cemetery at llandough 15
fig. 9
Selected burials B306–B898.
fig. 8 (facing)
Selected burials B80–B267.
16 neil holbrook and alan thomas
fig. 10
Burial 601. © Cotswold Archaeology.
Severn Valley ware jar or large tankard. A 1st-century a.d. date is suggested for
this burial (p. 86).
Leg Positions
The vast majority of burials had their legs out straight, as with the post-
Roman cemeteries of Cannington and Henley Wood where this was almost the
only position.17 Exceptions comprised eight burials which had their ankles crossed
(e.g. B39*), one adult burial (B89) which had the right leg flexed and the other
straight, and a further fourteen burials with slightly flexed or bowed legs, probably
a result of post-burial settlement.
Arm Positions
Seventeen arm positions (A–Q) have been defined in the cemetery population
with reference to those identified in the discussion of the Poundbury (Dorset)
17 Woodward, op. cit. in note 15, 222. Rahtz et al., op. cit. in note 15, 76–7.
a monastic cemetery at llandough 17
fig. 11
Pie-charts showing prevalence of arm positions with male, female and child burials.
18 neil holbrook and alan thomas
assemblage.18 These are listed in Table 3 and their occurrence quantified in Table
2.
Table 3
POSITION OF THE ARMS IN THE CEMETERY POPULATION (ILLUSTRATED
EXAMPLES IN BRACKETS)
Arm Position A both arms straight (B6, B10, B39, B64, B898)
Arm position B left arm bent on to pelvis, right arm straight (B114, B206, B251, B265,
B796)
Arm Position C left arm straight, right arm bent onto pelvis (B87, B536)
Arm Position D both arms bent onto pelvis (B212, B537)
Arm Position E both arms crossed on lower chest (B474)
Arm Position F both arms crossed on waist (B66)
Arm Position G left arm straight, right arm crossed on waist (B2, B13)
Arm Position H left arm bent onto pelvis, right arm crossed on waist (B622)
Arm position I left arm crossed on waist, right arm bent onto pelvis (B115)
Arm Position J left arm raised by head, right arm bent or crossed onto pelvis (B37,
B333)
Arm Position K both arms raised by head (B80)
Arm Position L left arm straight, right arm raised by head (B98)
Arm Position M left arm crossed on waist, right arm on chest (B101)
Arm Position N left arm raised by head, right arm on chest (B190)
Arm Position O left arm on chest, right arm crossed on waist (B405)
Arm Position P both arms crossed on pelvis (B601)
Arm Position Q left arm crossed on waist, right arm straight (B35)
The most common position was with both arms straight (Position A: 25% of
the population), followed by Position B (19%), Position P (15%), Position C (14%)
and Position D (6%). None of these arm positions was confined to any particular
part of the cemetery and all were found within Areas I–III. Most of the other arm
positions were too rare to allow for any meaningful analysis of their distribution,
and there were five instances where the arm position was unique, and one with
only two examples. Four burials in these latter two categories were found in one
part of Area I: B80*, B98*, B190* (which contained a burial pebble), and B333*
(which produced a radiocarbon date of 610–777 cal a.d.).
Woodward has studied the arm positions in a selection of Late Roman and
early post-Roman cemeteries, and concluded: ‘. . . that some cemeteries are
characterised by very particular combinations of arm positions, although no
general chronological or spatial patterning of the different specific positions may
be perceived.’19 Table 4 presents the data from a number of sites, and the lack of
obvious pattern is apparent. Tables 5 and 6 present a concordance of burials with
independent dating evidence, and it is evident that certain arm positions were
18 Woodward, op. cit in note 15, 222–6.
19 Woodward, op. cit. in note 15, 224.
20 Sources for Table 4. Poundbury, Cirencester, Cannington, and Henley Wood: Woodward, op. cit. in note 15,
fig. 130; Caerwent: E. Campbell and P. Macdonald, ‘Excavations at Caerwent Vicarage Orchard Garden 1973:
An extra-mural post-Roman cemetery’, Archaeol. Cambrensis, 142 (1993), 74–98; Capel Maelog: W. J. Britnell,
‘Capel Maelog, Llandrindod Wells, Powys: excavations 1984–87’, Medieval Archaeol., 34 (1990), 27–96.
a monastic cemetery at llandough 19
Table 4
PREVALENCE OF SELECTED BURIAL TRAITS IN LATE-ROMAN-MEDIEVAL
CEMETERIES
Poundbury (Dorset) and Cirencester (Gloucs.) are late Roman; Cannington and Henley
Wood (Somerset) early post-Roman; Caerwent (Mons.) c. 6th–7th century. At Capel
Maelog (Powys) the cemetery spans the period from the 7th/11th to early 16th century,
although of a population around 500 individuals, sufficient skeletal material to determine
the burial arrangement of the corpse survived only in 16 graves which date to the period
from the mid-13th century to the early 16th century.20
Site Population. Stone Arm Position Arm Position Arm Position Arm Position
lining A D F/P E
Poundbury 1,114 3% 20% 40% 10% 0%
(Main)
Cirencester 453 6% 15% 23% 11% 8%
Cannington 542 5% 32% 0% 38% 10%
(159)
Henley Wood 72 15(+)?% 30% 33% 26% 13%
Caerwent 76 10% c. 100% 0% 0% 0%
Llandough 824 2% 25% 6% 18% 1%
Capel Maelog (c. 500) 3% 0% 0% 30% 30%
utilised over a very long period. Arm Position A doubtless occurred throughout the
life of the cemetery and was certainly still being used after the late 9th century at
least (B228). Arm Position B was used in burials B110 and B65, which produced
radiocarbon dates of 655–938 and 705–1000 cal a.d. respectively; and in three
graves containing burial pebbles. Arm Position C also occurs in four graves
containing burial pebbles (the incidental sherd of medieval tile in the filling of grave
B10* is probably intrusive) and in B87* which produced a radiocarbon date of
776–1018 cal a.d. The arms crossed-on-chest arm position (Position E) has been
discussed in the report on the excavations at Cannington. It has been suggested
that this might be a possible indicator of Christianity, but the authors stress that
too few examples have been found in unequivocal Christian contexts for there to
be any certainty in the matter.21 Its rarity at Llandough demonstrates that it was
not a common early Christian rite in this part of Wales at least. Arm Position G
was used for a burial with a radiocarbon date of 370–640 cal a.d. and in four
graves containing burial pebbles. Arm Position P occurs in six graves containing
burial pebbles (the 13th-century terminus post quem provided by a single sherd of
medieval tile from a burial underlying B811 has been disregarded) and in one
(B688) that produced a radiocarbon date of 779–1024 cal a.d. Arm Position K is
unusual although a single burial at the Anglo-Saxon cemetery at Alton (Hants.)
had the right arm sharply bent back with the hand on the neck which could have
been due to the pain caused by severe arthritis in the vertebrae.22 Similarly there is
21 Rahtz et al., op. cit. in note 15, 417.
22 V. I. Evison, An Anglo-Saxon Cemetery at Alton, Hampshire (Winchester, 1988), 29.
20 neil holbrook and alan thomas
Table 5
BURIALS WITH INDEPENDENT DATING EVIDENCE.
Abbreviations: strat. rel. denotes a burial where dating derives from its stratigraphic
relationship with another burial; TPQ = terminus post quem; TAQ = terminus ante quem; C14
denotes a burial which is dated by radiocarbon assay. On balance finds of later-medieval
and post-medieval pottery and tile within the grave fills have been considered intrusive.
Incidental finds of Roman pottery within the grave fills are most probably residual and so
are likewise not listed.
Burial Method Trait Arm Orienta- Datable Stratigraphic Suggested
No. Posit. tion Finds relationships Date
2* C14 Part G 45° Roman 370–640
stone pot cal a.d.
lining
35 grave good Part Q 70° hobnails, ?
artefact stone Roman pot
lining
34 grave good D 75° iron knife Above B38 11th century
artefact Roman pot
38 Strat. Rel. D 75° Below B34 Pre-11th
century?
40 grave good B 45° melon Above B521 ?
artefact bead
59 grave good B 80° mid-4th Below B47 ?
artefact century coin
65 C14 B 45° Above B355 705–1000
cal a.d.
74 Strat. Rel. B 75o iron frags. Above B106 TPQ 782 cal
A.D.
87* C14 C 70° hobnails, Cuts B333 776–1018 cal
coin, slag, a.d.
Roman pot
93 grave good organic P 70o 2 mid-4th Above B110. TPQ 655 cal
artefact stain century coins; Below B69 a.d.
Iron nails
94 Strat. Rel. C 75o iron nail, slag Above B106 TPQ 782 cal
a.d.
106 C14 N–S A 340o Roman pot Below B74, 782–1024 cal
burial B94, B115, a.d.
B125
110 C14 organic B 75° hobnails, Below B33 655–938 cal
stain Roman coin, a.d.
slag, Roman
pot
115* Strat. Rel. I 65o Above B106 TPQ 782 cal
a.d.
116 casual B 100° Bii Above B130 late 5th
artefact Amphora century +
125 Strat. Rel. Partially 70o 2 hobnails Above B106 TPQ 782 cal
flexed a.d.
a monastic cemetery at llandough 21
Burial Method Trait Arm Orienta- Datable Stratigraphic Suggested
No. Posit. tion Finds relationships Date
228 C14 A 50° burial Cuts B248, 885–1035
pebble, B272 cal A.D.
Roman pot
248 Strat. Rel. Q 50° Cut by B228 TAQ
885–1035
cal a.d.
272 Strat. Rel. A 50° Roman pot Cut by B228 TAQ
885–1035
cal a.d.
333* C14; Strat. J 50° Roman tile Cut by B87 610–777 cal
Rel. A.D.
355 Strat. Rel. B 40° Below B65 TAQ
705–1000
cal a.d.
380 C14 coffin Q 60° 680–980
cal a.d.
458 Strat. Rel. infant 75° Above B854 Late 5th
century +
517 casual B 80° Bii Above B539 Late 5th
artefact amphora & B771 century +
525 Strat. Rel. P 90° Late Iron-age Above B542 Late 5th
pot century +
542 casual organic 75° Bii Below B525, Late 5th
artefact stain amphora, Cuts B679 century +
burial
pebble,
?Late Iron-
age pot
547 grave good B 80° Roman beads Below B503, ?
artefact 501
566 C14 Part of a D 85o Cuts 565 434–687 cal
row of a.d.
burials in
Area III
578 Strat. Rel. C 90° Above B638 ?Early-
medieval +
624 grave good D 80o Roman glass Above 625 ?
artefact bead, Roman
pot
631 C14; grave 75° iron girdle, 361–662 cal
good burial pebble, a.d. Girdle
artefact post-med. suggests
pottery 5th–8th
century date
22 neil holbrook and alan thomas
Burial Method Trait Arm Orienta- Datable Stratigraphic Suggested
No. Posit. tion Finds relationships Date
640 casual 65° late 6th–7th Above B663, Late 6th–7th
artefact century glass Below B644 century +
cone beaker,
Late Iron-
age/ Early-
Roman pot
644 Strat. Rel. 80° Below B645, Late 6th–7th
Above B640 century +
645 Strat. Rel. infant 80o Above B644 Late 6th–7th
century +
650 Strat. Rel. 90o Above B784 TPQ 604 cal
A.D.
688 C14; P 60o Above B705, 779–1024 cal
stratigraph- B755 a.d.
ically late
722 Strat. Rel. B 100o Above B784 TPQ 604 cal
a.d.
784 C14 A 90o Above B783 604–885 cal
Below B650, a.d.
B722
796 grave good B 50° Polden Hill Above B850, ?
artefact brooch, 868, 901,
Roman pot 930;
Below B851,
551, 817,
661, 532
819 casual C 40° Bii Below B755 Late 5th
artefact amphora; century +
burial
pebble
851 grave good 80° Roman Above B796 ?
artefact bead Below B852
854 Strat. Rel. P 70° Below B458 late 5th
Above B864 century +
864 casual A 80° Bii Below B854 late 5th
artefact amphora, century +
burial
pebble
898 grave good A 95° Fe knife ?8–11th
artefact century
934 grave good G 80° Roman Above ?
artefact bead; B768, Below
burial B710
pebble
a monastic cemetery at llandough 23
Table 6
DETAILS OF BURIALS CONTAINING PEBBLES
Burial No. Sex /Age + Pebble Arm Orienta- Other Finds Relationship
No. of Grave trait Position Position tion
pebbles
18 1 M, 45+ P 60o Roman pot, Above B24
animal bone
142 1 Adult G 95o
190 1 F, 25 N 80o Late Iron-
Partial age Early
Stone Roman pot,
Lining fired clay,
animal bone
228 1 F, 35 A 50o Roman pot C14 885–1035 cal
a.d.
Cuts B248 & B272
251* 1 U, 30 B 50o Above 252
Coffin
306* 1 ?F, 20 Chest 80o Roman pot Above B275 &
B334. Cuts B304
322* 2 Adult Legs 70o Above B353
442* 1 M, adult 80o slag, flint, Below B445,
Above B519
4481 5 F, adult I 80o Above B480
456 2 Y Child 80o Above B644
498 1 U, Adult Ribs 80o slag Above B602
& B472
512 2 U, Adult Feet (1) P 75o Roman pot, Above B682
animal bone
521 1 M, 40 Pelvis 70o Roman pot, Below B470
animal bone & B522
533 1 Child 90o animal bone Above B840
542 1 M L arm 75o ?Late Iron- Below B525,
age pot, Above B679
Bii amphora,
animal bone
547 1 F, 60+ C 80o 4 beads, iron Below B501
nail, slag
5512 2 Y Child 80o Above B796
560 1 Fetus Chest 70o Above B656
562 1 F, 45 C 105o ?Later- Below B545,
medieval pot, Above B568
animal bone
570 2 Y Child 60o limpet shell, Above B642
animal bone & B1000
587 1 M, 30 C 80o animal bone Cuts B446
Shroud
589 3 Y Child Chest 80o animal bone
627 2 ?F, 40 L arm (1) A 80o 2 iron nails
Feet (1)
24 neil holbrook and alan thomas
Burial No. Sex /Age + Pebble Arm Orienta- Other Finds Relationship
No. of Grave trait Position Position tion
pebbles
628 3 Child Knees (1) 75o Cut by B617;
Feet (1) Above B618
631* 1 M, 30 Chest 75o iron girdle, C14 361–662 cal
girdle animal bone a.d.
638 1 ?F, 50+ P 80o Roman pot, Above B662,
Post-med Below B578
glass & B537
643 10 Child L shoulder P 80o Below B753
(1) chest/ & B754, Above
abdomen B892 & B618
(5) knees
(3)
680 1 Y Child 70o Below B639,
Above B749
& B723
687 6 ?F, adult 80o animal bone, Below B645, B837
slag, ?Late
Iron-age pot
691 12 80o Roman pot, Cut by B692
Roman bead,
iron nail
727 3 F, 25 Knees (2) P 80o animal bone Below B568
& B562,
Above B956
753–756 25 F, 30 25o fired clay, Above B643,
M, 60+ 80° animal bone B756, B819,
U, adult 30o B764;
U, adult - Below B688,
B810, B755
& B819
760 4 ?F, 35 Knees (2) P 60o Above B777;
Feet (2) Below B545
768 2 ?F, 40 Q 60o Roman pot, Below B459 &
animal bone B934 Above B747
796* 1 ?M B 50o Roman pot, TPQ of 604 cal A.D.
Roman from B784.
brooch, Above B650,
animal bone, B868, B901 &
iron hobnail & B930;
Below B851,
B551, B532, B661
& B817
819 1 M, 40 C 40o Bii amphora, Below B755
animal bone
846 1 F, 40 Knees A 80o animal bone Below B890,
Cuts B848,
864 1 F, 40 A 80o Bii amphora, Above B698
animal bone
a monastic cemetery at llandough 25
Burial No. Sex /Age + Pebble Arm Orienta- Other Finds Relationship
No. of Grave trait Position Position tion
pebbles
868 1 F, 30 A 110o Above B881,
Below B901
890 1 M, 20 Feet G 80o animal bone Below B613 &
B806, Above
B846 & B847
896 1 Y Child 80o Above B916
934 1 M L pelvis G 80o Roman bead, Above B768,
animal bone Below B710
983 1 M, 35 B 85o Above B913
Notes
1 The burial stones from B448 may be associated with groups of disarticulated bone (B480–1) and therefore not
in situ.
2 A burial stone from B551 may be associated with Bone Group B550.
Not listed in the Table: Bone Group B391 (a few skull fragments) produced a single pebble; Bone Group B548
(an upper arm and a few skull fragments) produced three pebbles around the skull; Bone Group B728 ( just skull
fragments) produced a single pebble; Bone Groups B802, B843, B855 each produced a single pebble, and Bone
Group B1021 produced two pebbles.
no obvious patterning in the distribution of the various arm positions, save for the
greater prevalence of the crossed arm positions F/P in the northern part of the
cemetery (Fig. 13A). This may indicate that these positions were more prevalent in
the later period of cemetery use, and certainly the crossed arm positions were very
common in the medieval burials at Capel Maelog (Tab. 4). Figure 11 presents the
prevalence of the different arm positions with male, female and child burials. Arm
Position B is slightly more common in child burials than those of adults, but
otherwise little patterning is evident.
Skull Positions
The position of the skull in the graves is not considered to be significant as
burials with skulls facing to the left, right, and upwards are common throughout all
areas of the cemetery. A similar conclusion has been reached at Cannington and
elsewhere.23 No clear association with any burial trait can be discerned (Tab. 2).
grave-types
Many graves had been cut through the natural limestone bedrock and in
sixteen cases the stone spoil had been re-used as packing around the edge of the
grave-cut (e.g. B2*, which yielded a radiocarbon date of 370–640 cal a.d., B6*,
B35*, which contained 21 Roman hobnails, and B190*). The distribution of these
graves is shown in Figure 13B where the concentration within Area I is apparent.
The packing was usually random and not laid with particular care. Exceptions
comprise three burials (B110, which produced a radiocarbon date of 655–938 cal
23 Rahtz et al., op. cit. in note 15, 79.
26 neil holbrook and alan thomas
a.d., B373, and B809) with unworked vertically-pitched thin limestone slabs placed
upright at their feet, and five burials (B61, 81, 490, 508, and 536*), which contained
head support stones (Fig. 13B). In three cases the skull was sandwiched between
two stones; elsewhere only a single stone was used. All of these latter graves, apart
from B61, were either children or young children. The distribution of graves with
stone linings shows a marked bias towards the southern part of the cemetery (Fig.
13B), a pattern also apparent in the distribution of Roman artefacts within grave
fills (Fig. 14C). Burials 829 and 908 had single large stones placed immediately to
the top of the skull. Two graves (Fig. 13B) had possible markers: B39* (F, 45) with
a broken vertically-pitched unworked slab at its head, possibly part of a headstone,
and B60 (M, 45+) which contained a group of stones arranged to one side of the
skull, possibly packing for a timber marker-post. The degree of intercutting evident
in the cemetery indicates either a low regard for spatial organisation, or a lack of
respect for previous burials. It would therefore appear that it was not usual practice
for most graves to be marked, or if they were, that the markers quickly perished.
The use of rough stone-lining in graves is a tradition that occurs in both
Roman and post-Roman cemeteries. Table 4 presents the evidence from a selection
of Late Roman–medieval cemeteries. The high value from Henley Wood may be
of little significance as many of the graves there were rock cut. A slab-lined grave
dated to the early 4th century was found at the Romano-British farmstead at Biglis,
4 km south-west of Llandough, and such graves were also represented in the small
cemetery at the Atlantic Trading Estate, Barry (Vale of Glamorgan), which could
possibly have continued in use from the 2nd to 10th century.24 The use of head-
support stones is a feature which also occurs spasmodically in the graves of a
variety of dates, including Late-Roman; early post-Roman (such as grave 5A at
Henley Wood), and medieval (such as Capel Maelog where in one case at least the
grave pre-dated the late 12th- or early 13th-century church).25 Their function was
presumably to lessen head-rolling. There were no examples from the present
excavation of ‘pillow’ stones, although one was found with a child burial which
produced a radiocarbon date of 668–1022 cal a.d. to the south of the church.26
There were no examples from Llandough of slab graves (i.e. with a floor of flat
slabs) or lintel cists (where the sides, floor and top of the grave were formed by flat
stone slabs). This may be of significance as they occur widely around the western
seaboard of Britain, and are ascribed to the Early Christian Period.27 At the
Caerwent East Gate cemetery there were four slab burials and one lintel cist; a
further cist had previously been recorded in 1910.28 A lintel-cist grave from Caer,
Bayvil (Pembs.), produced a radiocarbon date of 640–883 cal a.d.29
24 J. Parkhouse, ‘Excavations at Biglis, South Glamorgan’, 1–64 in D. M. Robinson (ed.), op. cit. in note 4, 16;
H. James, ‘Early medieval cemeteries in Wales’, 90–114 in N. Edwards and A. Lane (eds.), The Early Church in
Wales and the West (Oxford, 1992), 96–8.
25 R. Leech, ‘Religion and burials in South Somerset and North Dorset’, 229–66 in W. Rodwell (ed.) Temples,
Churches and Religion: Recent Research in Roman Britain (BAR Brit. Ser., 77, Oxford, 1980); Watts and Leach, op. cit. in
note 15, 57; Britnell, op. cit. in note 20, 54.
26 Owen-John, op. cit. in note 4, 144.
27 C. Thomas, The Early Christian Archaeology of Northern Britain (London, 1971), 48–90.
28 Campbell and Macdonald, op. cit. in note 20, 76 and 84.
29 H. James, ‘Excavations at Caer, Bayvil, 1979’, Archaeol. Cambrensis, 136 (1987), 51–76.
a monastic cemetery at llandough 27
fig. 12
Distribution of males, females and children under 18 within the cemetery.
grave furniture
No certain coffin fittings were found. Some 27 graves, however, produced
nails or very small corroded iron fragments. Of these, 23 produced single nails and
three only two nails. Only in the case of B449, which produced a total of seven
nails on the upper chest, the pelvis, and by the right shoulder, might these represent
coffin nails.
Some graves contained traces of black organic staining around the edges or
base of the grave cut, or on top of the skeleton itself. In nine cases (B21, 238, 251*,
256, 380, 404, 672, 784, and 915) this staining is probably the remains of a decayed
wooden coffin (although none produced nails), while in nineteen other cases the
staining is so fragmentary that it could be the remains of some other organic
material. Indeed, one of these burials (B110, dated 655–938 cal a.d.) is unlikely to
have been in a coffin as it had a vertically-pitched limestone slab at its feet. One
coffin burial (B380) produced a radiocarbon date of 680–980 cal a.d., while
another two (B251* and B542) contained burial pebbles. Burial 542 which
displayed some organic staining also yielded a burial pebble and two sherds of Bii
amphora from the grave fill. At least nine burials exhibited a marked ‘hunching’ of
the shoulders, indicating that these were either shroud burials or that they had
been tightly packed into a coffin (e.g. B114*). Numerous other burials display a less
marked hunching of the shoulders. The distribution of the nine hunched burials
closely matches that of the coffins and it is possible that coffin and shroud burials
were a contemporary burial rite (Fig. 15E). It is very probable that more
28 neil holbrook and alan thomas
fig. 13
A: Burials with Arm Positions F and P. B: stone-lined graves, grave markers and head-support stones.
burials were interred in coffins of which no trace has survived. Consequently the
distribution of recorded coffin burials in the cemetery is unlikely to reflect their
true distribution, although it can be observed that few graves with organic staining
a monastic cemetery at llandough 29
fig. 14
C: grave fills containing Roman pottery, Roman coins, Bii amphora and early-medieval glass. D: Hobnail
burials, Roman brooch grave good, glass beads.
30 neil holbrook and alan thomas
fig. 15
E: graves with evidence of wooden coffins and shrouds. F: graves containing burial pebbles, iron knives and the
iron girdle.
a monastic cemetery at llandough 31
occurred in Area I, and that there were small clusters in Areas II and III. Both
males and females, adults and juveniles are represented.
Wooden coffins occur commonly in some Late Roman cemeteries (such as
Poundbury and Lankhills, Winchester (Hants.), while being much less prevalent
elsewhere (such as Cirencester).30 Such coffins also occur in a number of post-
Roman cemeteries in Wales, such as Plas Gogerddan, Ceredigion (radiocarbon
date from a coffin of 265–640 cal a.d.) and Tandderwen, Denbighshire (radiocar-
bon dates from coffins of 560–655 and 886–1012 cal a.d.).31 A higher proportion
of the graves attributed to the early cemetery at Capel Maelog (7th/11th–late
12th/early 13th century) had coffins than in the later phases.32 Shroud burials,
identified in the same manner as those as Llandough, were also found at the
Atlantic Trading Estate, Barry.33
grave fillings
Most of the graves were backfilled with redeposited yellow clay containing
limestone fragments and charcoal flecks. This often made recognition of the precise
edges of the grave cut very difficult. Where there were instances of several
intercutting graves the fills tended to be much darker, resulting in a dark grey,
brown, or greenish-brown silty clay gravesoil heavily mottled with the yellow clay
and limestone fragments. These fills often contained bone fragments from earlier
disturbed burials and most of the artefactual material is either simply residual,
being incidentally incorporated within the grave backfill, or possibly ‘sweepings’ of
the debris from a graveside meal.34 Other than odd nails and hobnails, casually
deposited finds include an iron oval chain loop from the fill of B106; two Late-
medieval pins from B75; a rim fragment of a glass cone beaker of late 6th- or 7th-
century date from B640; a small sliver of plain glass, possibly of Early-medieval
date (Fig. 14C), from B482, and a fragment of a lead window grille, probably
Roman, from B331. In addition, pieces of slag were found in several grave fills and
two globules of post-medieval glass waste in the fill of B457. All of the Roman
pottery recovered from the grave fills is also considered to fall into the accidental
category, although a slight caveat should be introduced concerning the five graves
which produced sherds of post-Roman Bii amphora (Fig. 14C). These amphorae
(and their contents) were doubtless exotic and high-status, and it has been suggested
that such pottery may have been deliberately introduced into the grave fill as part
of the burial ritual. Certainly the discovery of amphora sherds on the ground
surface close to a ‘special’ grave within the churchyard of St Materiana at Tintagel
(Cornwall) has been interpreted thus.35
30 Woodward, op. cit. in note 15, 228.
31 K. Murphy, ‘Plas Gogerddan, Dyfed: a multi-period burial and ritual site’, Archaeol. J., 149 (1992), 1–38 at
p. 22; K. S. Brassil, W. G. Owen and W. J. Britnell, ‘Prehistoric and early medieval cemeteries at Tandderwen,
near Denbigh, Clwyd’, Archaeol. J., 148 (1991), 46–97 at p. 88.
32 Britnell, op. cit. in note 20, 54.
33 James, op. cit. in note 24, 97.
34 For a discussion see Rahtz et al., op. cit. in note 15, 398.
35 C. D. Morris, J. Nowakowski and C. Thomas, ‘Tintagel, Cornwall: the 1990 excavations’, Antiquity, 64 (1990),
848; C. Thomas, Tintagel, Arthur and Archaeology (London, 1993), 103; Rahtz et al., op. cit. in note 15, 424.
32 neil holbrook and alan thomas
Given the frequent intercutting of graves, and the truncated stratigraphy, it
must be recognised that there is a strong likelihood that some artefacts contained
with the grave fills are intrusive, either from later cutting burials or the overlying
topsoil. For example, eight burials contained sherds of High-medieval pottery and
tile, and four post-medieval pottery, including B631* which contained the securely
dated Early-medieval girdle. Given that the site of the cemetery was occupied by
Great House Farm in the post-medieval period, and that the farmhouse itself may
have a Late-medieval origin, this pottery must be considered intrusive, perhaps
introduced during post-cemetery cultivation or animal burrowing, and it cannot
be used to indicate that burial continued until the 13th or 14th century or beyond.
The timber building in the north-eastern part of the site provides further evidence
for the cemetery having gone out of use before this time (p. 39).
potentially significant artefacts contained within graves
This section discusses Roman and post-Roman artefacts (other than pottery)
contained within graves. In some cases deliberate deposition of the object as a
grave good seems reasonably assured (for instance the Roman brooch with B796*
and the post-Roman knives with B34 and B898*). Elsewhere there are varying
levels of doubt whether artefacts were consciously placed with a body as part of a
ritual or accidentally introduced into the grave during backfilling.
Burial 34 (M, 25): a single-edged blade, comparable to 11th-century examples at
Whithorn (Dumfries), had been placed on the right side of the chest (Fig. 35, 3).
Burial 35 (U, 45+): 21 hobnails within the grave fill, possibly from a shoe thrown
in with the backfill. This could have been deliberate. The grave had a partial stone
lining, and the fill also contained sherds of late 3rd- or 4th-century pottery.
Burial 40 (C): a large complete Roman melon bead dated to the 1st or 2nd century
found on the right-hand side of the neck (Fig. 33, 1).
Burial 59 (?F, adult): the grave fill contained a 4th-century Roman coin. This grave
had been heavily disturbed and there is a possibility that the coin had been
displaced from elsewhere.
Burial 87* (F, 20): 15 iron hobnails were recovered from the feet area. A Roman
coin of a.d. 330+ was contained within the grave fill along with 4th-century
pottery. The bottom of the grave displayed organic staining. A radiocarbon date of
776–1018 cal a.d. proves the Early-medieval date of this grave, as does the date of
610–777 cal a.d. from the stratigraphically earlier burial B333*.
Burial 93 (F, 45+): a Roman coin of a.d. 335–40 was found near the left eye
socket. Whether it was originally placed here, or had fallen from the mouth on
decomposition is uncertain. A second coin of a.d. 337–40 was recovered from the
grave fill, as were two iron hobnails, and an iron nail 25+ mm long. The base of
the grave displayed evidence of organic staining. This burial overlay B110 and so
is no earlier than 655 cal a.d.
a monastic cemetery at llandough 33
Burial 110: the grave fill produced four hobnails (widths of heads 10 mm and 12
mm) and two nails, both with bent shafts. The one complete example was 68 mm
long. Two of the nails were found at the feet together with a coin dating to a.d.
337–40. The foot of the grave was lined with vertically pitched stone slabs. A
radiocarbon date of 655–938 cal a.d. once again confirms the date of this burial.
Burial 449 (C): a length of copper-alloy rod, possibly part of a pin from a Roman
brooch, was found between the feet (Fig. 34, 2). There must be uncertainty whether
this was a deliberate deposition.
Burial 547 (F, 60+): four small glass annular beads were found close together on
the neck, directly underneath the jaw. Their proximity and arrangement suggests
that they were strung around the neck (Fig. 33, 2–5). The beads are difficult to date
but are possibly recycled from Roman bottle glass. The grave fill also produced a
burial pebble.
Burial 624 (U, adult): a small dark grey annular glass bead (Fig. 33, 7). Position
within grave uncertain.
Burial 631* (M, 30): dated 361–662 cal a.d., contained a series of iron bands
around the waist from an Early-medieval girdle (Fig. 15F). A full description and
discussion of this object is presented below (pp. 53–64).
Burial 796* (?M): a complete Polden Hill-type brooch of the 1st century a.d. The
brooch was in good condition and lay by the left hand, which had been positioned
on the right side of the pelvis (Fig. 34, 1). It is possible that this may have been used
to fasten a shroud, although the body did not display the hunching of the shoulders
recorded elsewhere. The Early-medieval date of the grave is indicated by a
radiocarbon date of 604–885 cal a.d. from a burial (B784) which was stratigraph-
ically earlier than B796*. The grave fill also contained a burial pebble. The brooch
must therefore be regarded as a stray find that was collected as a curiosity.
Burial 851 (YC): large dark blue glass annular bead (Fig. 33, 9). Roman. Position
within grave uncertain. This burial is stratigraphically later than B796* which
contained a burial pebble.
Burial 898* (M, 25) contained a single-edged knife, broadly datable to the 8th–11th
centuries, laid by the right leg (Fig. 35, 1).
Burial 934 (M): a small opaque turquoise annular glass bead (Fig. 33, 8). Roman.
Position within grave uncertain. The grave fill also produced a burial pebble.
Further residual hobnails were found in twenty other graves. Fifteen graves
contained only single nails, four graves two nails (B93, B125, B593, B839), and one
grave (B321) three nails.
The iron knives accompanying two burials by can reliably be interpreted as
grave goods (Fig. 15F). A further fragmentary knife of possible Early-medieval date
was found in the fill of a post-pit of the post-cemetery timber building, and the
blade from a cutting tool in an undated pit 120 (Fig. 35, 2 and 4). They might have
been disturbed from other burials. Iron knives are one of the few grave goods to
34 neil holbrook and alan thomas
occur in early post-Roman cemeteries, such as Cannington where they accompan-
ied seventeen burials, and Ulwell (Dorset) where a single knife was the only item of
furniture among 54 burials.36 Iron knives were also one of the comparatively rare
grave goods found with Period 1 (pre-a.d. 730) monastic graves at Whithorn.37
The complete state of the Roman brooch from B796* (Fig. 14D), and its
position by the left hand, also strongly suggest deliberate burial. Of the Roman
coins (Fig. 14C), those found within the grave fills of B110 and B35 might just be
incidental finds contained within the backfill of the graves, while that within B59
could have been disturbed from elsewhere. That nearer the head of B93, no earlier
than the mid-7th century, is the best candidate, but still a far from certain one, for
deliberate grave deposition. The practice of placing coins in graves, either by the
head or thrown into the grave fill, is a well-attested Late Roman practice. Indeed
Philpott has observed that the frequency with which coins are found in Roman
grave fills suggests that their presence is not accidental, but rather that coins were
often deliberately thrown into the grave pit during backfilling.38 Roman coins
occur in Anglo-Saxon graves, sometimes pierced for use as amulets, and it would
appear that the practice continued on a spasmodic basis well into the post-Roman
period in western Britain, using increasingly archaic coins that no longer served as
currency but may have had a symbolic value.39 At Caerwent a coin of a.d. 335–48
was found with a burial which was dated by radiocarbon assay to the 6th or 7th
century, while at Cannington a 2nd-century coin was found with a burial dated
540–790 cal a.d. and a pierced 3rd-century coin with a burial that associated grave
goods suggest is much later than the coin.40 The collection and retention of coins
for use as amulets and talismans is more credible than an alternative suggestion
that low-value Roman coinage was employed for economic transactions in post-
Roman western Britain.41
Most of the hobnails recovered from the graves occur in such small quantities
that they are best considered residual within the grave fill (Fig. 14D). The only
possible evidence for a body buried with iron-shod footwear is the fifteen hobnails
from the feet area of B87* which dates to no earlier than late 8th century. Footwear
of this type is not so far recorded from Early-medieval Wales, and a total of fifteen
hobnails is not sufficient to prove conclusively that they derived from nailed boots
or shoes.42 Burial with iron-shod footwear is a commonly attested Roman funerary
36 Rahtz et al., op. cit. in note 15, 84–95; P. W. Cox, ‘A seventh century inhumation cemetery at Shepherd’s
Farm, Ulwell, near Swanage, Dorset’, Proc. Dorset Nat. Hist. Archaeol. Soc., 110 (1989), 37–48 at p. 46.
37 P. Hill, Whithorn and St Ninian: The Excavation of a Monastic Town 1984–91 (Stroud, 1997), 74.
38 R. Philpott, Burial Practices in Roman Britain (BAR Brit. Ser., 219, Oxford, 1991), 212.
39 Cf. the discussion by W. Davies, Wales in the Early Middle Ages (London, 1982), 54–5 of the significance and
prevalence of coins in Early-medieval Wales.
40 M. Farley, ‘A six-hundred metre long section through Caerwent’, Bull. Board Celtic Stud., 31 (1984), 209–50;
Rahtz et al., op. cit. in note 15, 85–98.
41 K. R. Dark, Civitas to Kingdom (Leicester, 1994), 200–6.
42 J. M. Mills, ‘The Hobnails’, 99 in Farwell and Molleson, op. cit. in note 15, demonstrates that at the Late
Roman cemetery at Poundbury the numbers of hobnails found in individual graves suggest that there were three
basic types of footwear containing c. 10, c. 35 and c. 50 nails per boot. Study of the Roman footwear from
Billingsgate Buildings, London, identified three types of nailed shoes. Type B, the least heavily nailed, still had in
excess of ten nails per shoe: M. Rhodes, ‘Leather Footwear’, 99–128 in D. M. Jones, Excavations at Billingsgate
Buildings, Lower Thames Street, London, 1974 (London, 1980). The fifteen nails from B87* is therefore on the low side
if a similar type of shoe is envisaged.
a monastic cemetery at llandough 35
rite which was widely adopted in the rural areas of south-central England,
especially in the 4th century.43 The explanation of the practice is usually interpreted
to be a belief that the dead were felt to need a good pair of shoes for the journey
into the underworld, and has therefore been seen as an essentially pagan tradition.
There is now some supporting evidence that the tradition continued, albeit rarely,
into the post-Roman period. At Shepton Mallet (Somerset) an inhumation burial
was found with concentrations of hobnails at its feet and with a large quantity of
3rd- or 4th-century pottery in the grave fill. The burial produced a radiocarbon
date of 430–680 cal a.d.44
The beads from B40 and B547 were found by the neck and were observed
and recorded at the time of excavation. The single beads from B624, B851 and
B934 were only found during washing of the skeletons, and consequently their
position within the graves is not recorded (Fig. 14D). It seems likely that the beads
with B40 and B547 were suspended on necklaces that were being worn at the time
of burial, and as such can be considered grave goods. Greater doubt must attach to
whether the beads from the three other graves were deliberated deposited or
accidentally introduced during backfilling. The melon bead is clearly Roman, and
there are examples of single melon beads in Romano-British graves from Ospringe
(Kent); Horncastle (Lincs.) and Poundbury.45 There is no independent dating for
B40. The beads from B547 are difficult to date, although they are probably Roman,
and the Early-medieval date of this grave is demonstrated by the presence of a
burial pebble. Single beads, which were neither typically Roman or Anglo-Saxon,
were found in four graves at Cannington, one of which produced a radiocarbon
date of 660–1150 cal a.d.46
burial pebbles
A total of 122 pebbles of quartz and other non-local stone were found
associated with burials and disarticulated bone groups (Tab. 6).47 Forty individual
graves produced pebbles (e.g. B306*; Fig. 16), while a large scatter was found over
the group of intercutting burials B753–6. These burials overlay B643, a juvenile
which contained ten pebbles, so the pebbles in the later graves might represent a
disturbed covering to B643. B687, 1 m north of B643, yielded six pebbles. In
addition the stones from two further burials (B448 and B551) may be derived from
associated disarticulated bone groups. Seven other bone groups also produced
pebbles.
The pebbles were found in all areas of the cemetery but concentrate mainly in
the northern part (Fig. 15F). This distribution may be influenced in part by the fact
that the significance of the pebbles, particularly of the non-quartz variety, was not
43 Philpott, op. cit. in note 38, 167–75, Tab. A31.
44 P. Leech, Excavation of a Romano-British Roadside Settlement in Somerset, Fosse Lane, Shepton Mallet, 1990 (London,
2001), 45.
45 Philpott, op. cit. in note 38, tab. A30.
46 Rahtz et al., op. cit. in note 15, 96–8.
47 Geological provenancing by Heather Jackson clearly indicates that few of the recorded stones from Llandough
occur locally (pers. comm. M. Redknap).
36 neil holbrook and alan thomas
fig. 16
Burial 306 with quartz burial pebble on
the chest. © Cotswold Archaeology.
recognised in the early stages of the excavation and thus not recorded. Any such
bias will have affected the distribution plot in Area I, but not Area II.
Most of the burials contained only one to three pebbles, but a few contained
more. They occur in a variety of grave-types although eleven of the burials were
those of infants or children. Most of the pebbles appear to have been thrown in
with the grave fill but there are several instances where the stones were placed
directly over the burial. Where this occurs the stones appear to have been deposited
randomly. Pebbles were found with three burials (B542, 819, and 864), which also
contained sherds of Bii amphorae in their grave fillings. These burials only provide
a late 5th-century terminus post quem for the use of pebbles, although the radiocarbon
date of 361–662 cal a.d. from B631* shows that the rite was established before the
mid 7th century. The radiocarbon date from B228 of 885–1035 cal a.d. indicates
that the tradition continued until at least the late 9th century, unless the pebble in
this particular grave was simply residual. Pebbles were found with one foetus, six
young children, three children, twelve probable or certain males, and sixteen
probable or certain females.
a monastic cemetery at llandough 37
Quartz pebbles have been found in a number of other Early-Christian Welsh
contexts. At Capel Maelog two graves were specially marked at ground level by
edge-set stones enclosing an area covered by a thin scattering of large quartz
pebbles.48 One grave certainly belonged to the pre-church cemetery (i.e. pre-late
12th or early 13th century), and the other one might have done so too. A similar
use of quartz pebbles to cap graves at ground level has also been found at Pennant
Melangell (Powys) in a pre-12th-century context, and at St Justinian’s chapel near
St David’s (Pembs.).49 A second burial at St Justinian’s contained a white pebble
on the crown of the skull, another on the abdomen, one beneath each knee joint,
and one near the feet. At St Patrick’s chapel, also near St David’s, one burial
contained a pebble at the base of the pelvis and another several pebbles in the
grave fill.50 Incidences of such stones are not confined to Wales, however, as they
have also been found in Scotland, Ireland, the Isle of Man, and Cornwall. Dated
examples suggest a 7th-century or later date.51 The largest assemblage to be
studied is that from the monastic graveyards at Whithorn, where four graves which
can probably be dated to the second half of the 7th and opening decades of the 8th
century were capped with white stones. These pebbles also seem to have been used
as burial talismans in later 13th- to mid-15th-century graves.52 Where such stones
occur as coverings they presumably define ‘special graves’,53 but those that occur
with the body should probably be interpreted differently. Perhaps they are some
reflection of Christ’s messages in the Revelation of St John the Divine (Revelation
Ch. 2, Verse 17): ‘To him that overcometh will I give to eat of the hidden manna
and will give to him a white stone, and in that stone a new name written, which no
man knoweth, saving he that receiveth it.’ At Whithorn it was suggested that the
stones might be an ‘admission ticket to the other world’, while Crowe has
considered them tokens for prayers for the dead, a rite which may have derived
from monastic practice in 4th- and 5th-century Egypt and Syria.54 In a discussion
of the white stones which had been dug up in the churchyard of Penmynydd
(Anglesey) it is stated: ‘these had no doubt been brought here on occasion of
interments, when, as was usual in some parts of Wales during the Middle Ages,
each mourner brought and deposited a white stone on or near the grave of the
departed’.55
48 Britnell, op. cit. in note 20, 54.
49 W. J. Britnell, ‘Excavation and recording at Pennant Melangell Church’, Montgomeryshire Collect., 82 (1994),
41–102 at p. 82; E. J. Boake, ‘Report on the Excavations of the Chapel of St Justinian, St David’s’, Archaeol.
Cambrensis, 81 (1926), 381–94 at p. 387.
50 A. B. Badger and F. Green, ‘The chapel traditionally attributed to St Patrick, Whitesand Bay, Pembrokeshire’,
Archaeol. Cambrensis, 80 (1925), 87–120 at pp. 103–8 includes a general discussion of the practice.
51 C. Crowe in J. A. Nowakowski and C. Thomas, Grave News From Tintagel (Truro, 1992), 9.
52 P. Hill, op. cit. in note 37, 472–3. The excavations at Whithorn uncovered graves containing white pebbles,
usually quartz or granite, and cattle teeth. According to Peter Hill, a total of 13,000 ‘white stones’ was recovered
from stratified deposits and studied. While they were widely dispersed in space and time, those in graves
concentrated in the latter part of Period I, perhaps dated to the second half of the 7th century. Two out of eight
graves in Phase 1, three out of thirty-one in Phase 2, five out of 33 in Phase 3, and 22 out of 47 in Phase 4 contained
pebbles. The practice continued in the Period II childrens’ graveyard, and Period II burials (?8th–late 9th
centuries).
53 Thomas, op. cit. in note 27, 58.
54 Op. cit. in note 51.
55 Anon., ‘Mona Mediaeva no. XXI’, Archaeol. Cambrensis, 3 ser., 5 (1859), 21–7.
38 neil holbrook and alan thomas
alignment
In some cemeteries it has been possible to construct detailed models of burial
process and sequence based upon alignment,56 but the Llandough assemblage is
unlikely to be susceptible to such analysis as the main factors determining
alignment are likely to have been the monastery church and its associated
boundaries. In a total of 781 cases sufficient survived of the burial for its alignment
to be determined. Given the nature of the bedrock it is unlikely that the grave
diggers would have been able to achieve any precision of alignment nearer than
5o –10o. The vast majority of burials lay between 50o and 90o (Tab. 7) and the
essentially E.–W. nature of the cemetery is therefore evident. The Victorian church
has an alignment of 83o east of north, and if this reflects the alignment of earlier
structures (such as the presumably timber monastery church) then this probably
accounts for the most frequently recorded orientation (71o –80o).57 Some burials
Table 7
BURIAL ALIGNMENT (FOOT END)
0° = feet due north; 90° feet due east;
360° feet due south
Alignment total %
0–10o 0 0.0
11–20o 4 0.5
21–30o 12 1.5
31–40o 19 2.4
41–50o 70 9.0
51–60o 89 11.4
61–70o 124 15.9
71–80o 314 40.2
81–90o 108 13.8
91–100o 19 2.4
101–110o 11 1.4
111–120o 2 0.3
121–130o 2 0.3
131–140o 0 0.0
141–150o 1 0.1
151–270o 0 0.0
271–280o 2 0.3
281–290o 2 0.3
291–300o 1 0.1
301–330o 0 0.0
331–340o 1 0.1
341–360o 0 0.0
TOTAL 781 100%
56 Cf. P. A. Rahtz ‘Grave orientation’, Archaeol. J., 135 (1978), 1–14.
57 A study of burial orientation within Early-medieval cemeteries in NW. Wales has shown that there is a
tendency towards burial in the ENE. arc of the compass which is unlikely to be accidental: D. Longley, ‘Orientation
within some Early Medieval cemeteries: some data from North-West Wales’, Antiquaries J., 82 (2002), 309–21.
a monastic cemetery at llandough 39
display a more pronounced NE.–SW. alignment and these may have been aligned
on the postulated curvilinear monastic boundary. There is therefore unlikely to be
any chronological significance to the varying alignments, and this is reinforced by
the stratigraphic evidence: in at least fourteen cases graves on more of a NE.–SW.
alignment cut those that lay approximately E.–W., while the converse was true on
at least eleven occasions. Two graves were dug on a near N.–S. alignment (B106
and 382). Burial 106 lay in Area I and produced a radiocarbon date of 782–1024
cal a.d. It had been cut by several later graves.
post-cemetery activity
In the NE. corner of the excavation part of a timber building represented by
twelve irregular post-pits was found. Its full plan did not survive because of post-
medieval truncation of the ground level (Figs. 5 and 17). The surviving features
presumably represent the W. and S. walls of a rectangular building, although the
post-pits are clearly not all of one phase. Although the dimensions of the post-pits
were large (in a number of cases about 1 m square), this was the product of the
removal of large slabs of limestone bedrock. For example post-pit 2589 had
dimensions of 1 m x 0.88 but the actual post pipe was just 0.1 m in diameter. The
remainder of the pit had been packed with stone. No floors or other internal
deposits survived and it is therefore difficult to assign a function to the building.
The building is clearly later than the abandonment of the cemetery as some of the
post-pits cut earlier graves. Eight post-pits produced a total of 116 sherds of 13th-
or 14th-century pottery (Tab. 16). In some cases the pottery was most probably
contained within the in situ packing of the post-pit (distinction was not always
possible) and therefore can be used to provide a terminus post quem for construction.
A shallow curving ditch 29 was aligned approximately NE.–SW. (Fig. 5). It
cut through several graves and was approximately 1.35 m wide, a maximum of
0.36 m deep, and had been recut once. The original ditch fill 30 consisted of a
gravely clay, and that of the recut 31 a very stony greenish-brown clay indicating
that the ditch had been deliberately backfilled. A second ditch 1214 lay further to
the west and was aligned approximately NNW.–SSE. It was traced for a length of
approximately 6 m, and had been removed at both ends by later quarrying. The
ditch was approximately 1.5 m wide and 0.3 m deep and had been recut once. It
contained a primary fill 1215 of redeposited yellow-brown clay, and that of the
recut 1171 was a greyish-green clay with a single surface patch (1172) of a shelly
grey-brown clay. This ditch cut B37*. No finds were recovered from either ditch
and they presumably relate to the period following the abandonment of the
cemetery but prior to the construction of Great House Farm.
At some point much of the site had been the subject of intense quarrying. One
of these quarries bisected the excavation area, and further large quarries were seen
to the west of the excavation area during the construction of two houses fronting
Penland Road. They had all been backfilled with what appeared to be a slurry-
type deposit. The date of their excavation and backfill is uncertain, but these
quarrying operations must have removed a large part of the cemetery. Stone walls
and demolition deposits apparent in the excavation all belong to Great House
Farm and its outbuildings, which had been systematically levelled in 1988.
40 neil holbrook and alan thomas
fig. 17
Plan of the later medieval building.
a monastic cemetery at llandough 41
Examination of the building by RCAHMW in 1974 suggested that the farmhouse
was essentially 17th-century but may have incorporated some medieval fabric.58
the radiocarbon dates (Tab. 8)
Bones from four burials were submitted for radiocarbon dating by Beta
Analytic Inc. in 1994. Two (B65 and 380) yielded insufficient collagen for
conventional analysis and were sent for AMS determination. Another six samples
were dated at Waikato in 1998, and a further two in 1999 (Fig. 6). In addition
Table 8 also includes the three dates already published from burials recovered in
the 1979 excavation and watching brief.59 The dates as published were not
Table 8
RADIOCARBON DATES60
Lab. Ref. Burial C14 date 2 Sigma 1 Sigma Area Selection
No. (BP) calibration calibration Criteria
(95% prob) (68% prob)
cal a.d. cal a.d.
Beta-76463 B2* 1570±70 370–640 420–590 I Stone-lined grave.
NE.–SW. burial
Wk-6938 B631* 1530±80 361–662 426–639 II Girdle burial
Wk-7021 B566 1450±55 434–687 558–653 III Part of a distinct row
of E.–W. burials
Wk-7323 B333* 1340±45 610–777 652–764 I Burial under B87
Wk-7023 B784 1320±70 604–885 651–777 III Burial stratigraphically
earlier than B796
Wk-7324 B110 1270±50 655–938 687–858 I Coin of 337–40 in fill
Beta-76466 B380 1210+60 680–980 770–890 II Wooden coffin
Beta-76464 B65 1170+60 705–1000 790–970 I NE.-SW. burial
Wk-6936 B87* 1140±50 776–1018 783–984 I Hobnail burial; coin of
330+ in fill.
Wk-7022 B688 1110±55 779–1024 887–997 III Stratigraphically late
burial towards N. limit
of cemetery
Wk-6937 B106 1100±50 782–1024 892–998 I Aligned N.–S.
Beta-76465 B228 1070+50 885–1035 960–1015 III Part of a row on a clear
NE.–SW. alignment
CAR 305 GGAT 1155±81 687–1025 777–794, – S. of church in
122 797–990 watching brief area
CAR 306 GGAT 1175±81 668–1022 775–983 – S. of church in
145 watching brief area
CAR 271 GGAT 885±75 999–1284 1033–1143, – S. of church. Dug into
052 1147–1258 villa demolition
58 Royal Commission on the Ancient and Historical Monuments of Wales, Notes on Sites and Finds in Llandough
Village (unpub. notes by R. F. Suggett, 1988).
59 Owen-John, op. cit. in note 4, 176.
60 References for calibration. Beta Analytic: J. C. Vogel, A. Fuls, E. Visser, and B. Becker, Radiocarbon, 35 (1993),
73–86. A. S. Talma and J. C. Vogel, ibid., 317–22; M. Stuiver, A. Long, R. S. Kra and J. M. Devine, ibid., 1–244.
Waikato: OxCal programme. Swansea: M. Stuiver and B. Becker, ibid., 35–65. Radiocarbon Calibration Program Rev.
3.0. University of Washington, Quarternary Isotope Lab.; M. Stuiver and P. J. Reimer, ibid., 215–30.
42 neil holbrook and alan thomas
calibrated, but this information has kindly now been supplied by Dr Q. Dresser of
Swansea University.
SUMMARY REPORT ON THE HUMAN SKELETONS
By LOUISE LOE and KATE ROBSON-BROWN
A total of 801 discrete skeletons was received for examination by the
Palaeopathology Study Group, University of Bristol.61 These represent the largest
assemblage of skeletons to have been recovered from Wales to date. They therefore
present a unique opportunity to explore the health and physical attributes of the
Early-medieval Welsh and allow comparison with populations of similar date and
type.62 This short report summarises the main findings of a standard anthropolo-
gical and palaeopathological examination that was undertaken on the remains
between 1998 and 2002. A more detailed report is to be published separately in
due course.63
Table 9
AGE CATEGORIES EMPLOYED IN THE
ANTHROPOLOGICAL AND
PALEOPATHOLOGICAL ANALYSES
Fetus = 0–9 months in utero
Perinate = 0 (around the time of birth)
Infant = >0–<2 years
Young child = >2–<6 years
Late Childhood = >6–<18 years
Young Adult = 18–25
Adult = >25–35
Middle Adult = >35–45
Mature Adult = >45 and >60
Ages and sexes were estimated by employing standard anthropological
techniques.64 For the purposes of analysis, summary ages were derived by taking
the middle of the ascribed range and, for adults, rounding it up to the nearest
decade (as appropriate). For example, individuals aged as 17–25 years were
summarised as 20 years. The summary ages were then placed into a series of
categories (Tab. 9) which have been utilised in the analysis presented in Figures
18–23. Further details on the methodology employed to estimate age and sex can
be found in the digital archive and the full specialist bone report.
61 The research was undertaken as part of doctoral study funded by a University of Bristol Scholarship and a
British Federation of Women Graduates Award.
62 Unless stated, all comparisons are based on the first author’s own observations whilst undertaking Ph.D.
research: L. K. Loe, Health and Socio-Economic Status in Early Medieval Wales (unpub. Ph.D. Thesis, University
of Bristol, 2003). They include samples examined from Atlantic Trading Estate, Barry (Vale of Glamorgan);
Cannington (Somerset); Saint Andrew’s Fishergate, York; and Whithorn (Dumfries and Galloway).
63 Op. cit. in note 14.
64 M. Brickley and J. I. McKinley, Guidelines to the Standards for Recording Human Remains (IFA Paper No. 7, 2004).
M. Cox and S. Mays, Human Osteology in Archaeology and Forensic Science (London, 2000). L. Scheuer and S. Black,
Developmental Juvenile Osteology (London, 2000).
a monastic cemetery at llandough 43
Table 10
CONDITION OF THE SKELETONS
No. of Good No. of Fair No. of Poor
Skeletons (%) Skeletons (%) Skeletons (%)
280 (35%) 320 (40%) 201 (25%)
Table 11
COMPLETENESS OF THE SKELETONS
<15% 15–25% >25–50% >50–75% >75%
No. (%) 58 (7%) 74 (9%) 323 (40%) 213 (27%) 133 (17%)
The skeletons were in varying states of preservation and completeness (Tabs.
10–11). Most were sufficiently well preserved to allow full analysis, being in a fair
or good condition and relatively complete. ‘Poor’ skeletons comprised bones with
considerable damage to most joint ends, as well as abraded bone surfaces, while
‘fair’ skeletons were characterised by bones that had reasonably well preserved
surfaces and relatively un-abraded joint ends. ‘Good’ skeletons generally had un-
abraded joint ends and bone surfaces that were in an excellent state of preservation.
Fragmentation was considerable across all condition and completeness categories.
No skulls had survived complete and of the 6,372 major long bones (humerus,
radius, ulna, femur, tibia and fibula) that were present, 373 (6%) were intact, 1,077
(17%) were complete but broken, and 4,922 (77%) were incomplete, some
represented by just a few fragments. Therefore, some analyses (for example,
metrical assessment) were limited by this degree of fragmentation.
The demographic profile (Figs. 18–19) reflects a mixed group comprising
males, females and children, with a normal mortality curve for a population of this
date and type.65 Deaths peaked in infancy, early adulthood, and later adulthood.
Infants were, however, under-represented and this may have been caused by
factors relating to burial practice, extent of excavation, preservation, and cultural
attitudes towards children.66 Twenty-eight per cent of the population failed to
reach adulthood. In particular, this relates to sub-adults aged between the ages
0–<2, 4–<6 and 8–<10 years. Compared to other samples, the ratio of sub-
adults to adults was lower suggesting a lower mortality rate among older children
overall.67
Males were more numerous in the sample than females: the ratio was 1.2
males to 1 female. This is within the range observed for other samples, although
Cannington has a higher ratio of females to males.68 Deaths were more common
among females at 18–25 years and 25–35 years indicating that child-bearing may
65 T. Waldron, Counting the Dead: The Epidemiology of Skeletal Populations (Chichester, 1994).
66 D. R. Brothwell, ‘Palaeodemography’, 111–29 in W. Brass (ed.), Biological Aspects of Demography (London, 1971);
J. Henderson, ‘Factors determining the state of preservation of human remains’, 43–54 in A. Boddington, A. N.
Garland and R. Janaway (eds.), Death Decay and Reconstruction (Manchester, 1987). Waldron, op. cit in note 65.
67 Loe, op. cit. in note 62.
68 Rahtz et al., op. cit. in note 15.
44 neil holbrook and alan thomas
fig. 18
Sub-adult mortality profile for the entire cemetery sample.
fig. 19
Adult mortality profile for the entire cemetery sample — columns from left: total; males; females; and unsexed.
have had a significant impact on female mortality. Mortality among males also
peaked at 25–35 years. It is possible that this is a result of problems associated with
age-estimation,69 or, given the comparatively high levels of morbidity that were
observed among males, that mobility, immigration and exposure to pathogens had
some influence on this mortality peak.70 Fewer individuals lived beyond 60 years
than beyond 45 years. This probably reflects an ageing bias rather than a true
trend because few individuals had the appropriate traits available with which to
estimate an age beyond 45+.
69 M. Cox, ‘Ageing adults from the skeleton’, 61–81 in Cox and Mays, op. cit. in note 64; R. L. Gowland and
A. T. Chamberlain, ‘Detecting plague: palaeodemographic characterisation of a catastrophic death assemblage’,
Antiquity, 79 (2005), 146–57.
70 Loe, op. cit. in note 62.
a monastic cemetery at llandough 45
fig. 20
Sub-adult mortality profile for burials within Area I.
fig. 21
Adult mortality profile for burials within Area I — total; males; females; and unsexed.
Some differences in mortality experience were identified when burials
contained within Area I were compared with those in Areas II and III combined
(Figs. 20–3). Area I had a higher proportion of males to females and children
whereas Areas II and III had a more even ratio of males, females and children.
Most striking is the high number of infant burials in Areas II and III. These occur
as a distinct self-contained group, centrally located in the north of the cemetery
and surrounded by adult burials (Fig. 12). It is impossible to determine why such a
group should be buried in this area. No pathology was identified that would
indicate a specific cause or event but acute disease, the changing use of the
cemetery, circumstances under which infants were baptised, social attitudes
towards the burial of infants and the status of parents (for example, their health,
marital, and religious status) are among the possible explanations for this.71
Stature ranged from 1.45 m to 1.87 m for the entire sample, between 1.57 and
1.87 m for males and 1.45 and 1.69 m for females. This was similar to those
calculated for other contemporary populations, although the range tended to be
greater than these smaller, samples (Tab. 12). The difference in mean male stature
and mean female stature also tended to be greater. A possible explanation for this
may be that the male population comprised immigrants, while the female
population comprised individuals from the local area. However, although stature
is influenced by genetic predisposition, environmental factors (for example,
71 Loe, op. cit. in note 62.
46 neil holbrook and alan thomas
fig. 22
Sub-adult mortality profile for burials within Areas II and III.
fig. 23
Adult mortality profile for burials within Areas II and III — total; males; females; and unsexed.
Table 12
SKELETAL STATURES: LLANDOUGH AND COMPARACTIVE SAMPLES. S.D. = STANDARD
DEVIATION
Llandough Whithorn Saint Andrew’s, York Cannington Atlantic Trading Estate,
(n=151) (n=7) (n=27) (n=37) Barry (n=19)
Mean Range s.d Mean Range s.d Mean Range s.d Mean Range s.d Mean Range s.d
Males 169.57 156.71– 5.28 173.20 160.69– 7.26 168.80 157.04– 5.67 169.16 161.24– 5.01 165.19 160.06– 4.04
186.66 178.52 177.86 178.20 170.90
Females 156.80 144.80– 4.60 165.00 162.04– 4.19 154.72 139.94– 6.51 159.66 149.14– 6.98 156.75 150.92– 3.07
169.12 167.92 162.07 176.95 159.32
Entire 163.56 144.80– 8.09 170.86 160.69– 7.35 161.50 139.94– 9.35 164.28 149.14– 7.71 161.91 150.92– 5.56
sample 186.66 178.52 177.86 178.20 170.90
childhood disease and malnutrition) are also an important influence. For example,
severe, longstanding malnutrition encountered during the growing years may
compromise an individual’s ability to achieve their full height potential.72 Sexual
dimorphism in height may, therefore, suggest a community that had access to a
nutritional surplus where, given the taller mean stature of males, this was invested
in their growth.73 Further analysis is required to test these hypotheses.
72 R. H. Steckel, ‘Stature and the standard of living’, J. Econ. Lit., 33 (1995), 1903–40.
73 C. Arcini, Health and Disease in Early Lund: Osteo-pathologic Studies of 3,305 Individuals buried in the First Cemetery Area
of Lund 990–1536 (Archaeologica Lundensia. Investigationes de Antiqvitatibus Urbis Lundae VIII, Lund, 1999).
a monastic cemetery at llandough 47
A range of pathological conditions was observed including those relating to
joint disease, nutritional disorders, infection and trauma. Joint disease was by far
the most common and this was mainly represented by osteoarthritis (OA), which
affected 29.7% of all adults, mostly males. A high prevalence of OA is not unusual:
it is the most common condition encountered in both modern and ancient
populations.74 Since it becomes more common with the advancement of age, it was
also unsurprising to find that most of those affected were aged over 40 years. The
most frequently affected sites were the cervical spine (13.5% of all cervical spines)
followed by the hand (9.5% of all joints combined). This pattern has been observed
in other populations of similar date.75 Skeletons with multiple joint OA tended to
show patterns that involved the hand, spine and shoulder and hand, spine and
elbow. In modern clinical contexts OA affecting the upper limbs is believed to be
more closely associated with hereditary factors than the weight bearing joints of
the lower limbs.76
The presence of other joint diseases in the population was indicated, such as
septic arthritis and, most interestingly gout. This disease is more common in males
than females, and is associated with obesity, excessive alcohol intake, high blood
pressure, and kidney problems.77 In dry bone, diagnosis rests on the presence of
punched out oval or round erosions that have overhanging edges (‘Martell hooks’)
and, when viewed radiologically, sclerotic margins.78 The most convincing example
in the Llandough population was observed on skeleton B848, a male aged between
35 and 45 years, and involved the right and left elbow joints and right and left first
metatarsophalangeal joints (Fig. 24), sites that are commonly affected by this
disease. Few examples of gout have been reported in the palaeopathological
literature, but five examples from the Romano-British cemeteries at Poundbury
are among these.79
No specific nutritional disorders (for example, rickets and scurvy) were
identified in the population. However, this does not necessarily mean that they
were absent. Rather, it is a reflection of how difficult they are to identify in human
skeletal remains. Preservation, limited knowledge of how these conditions manifest
themselves in the human skeleton, and variability in the expression of disease in the
skeleton, can all influence this. For example, six skeletons (B68, 233, 304, 468, 682
and 930) were noted for periostitis, a non-specific inflammation of the outer surface
of bone, involving sites that have been described in possible examples of scurvy
elsewhere.80 However, all six skeletons were incompletely preserved and in
74 J. Rogers and T. Waldron, A Field Guide to Joint Disease in Archaeology (Chichester, 1995).
75 T. Waldron, ‘Changes in the distribution of osteoarthritis over historical time’, International J. Osteoarchaeol., 5
(1995), 385–9.
76 D. Resnick and G. Niwayama, ‘Degenerative disease of extraspinal locations’, vol. 3, 1263–1371 in D. Resnick
(ed.), Diagnosis of Bone and Joint Disorders (Pennyslvannia, 1995).
77 C. A. Roberts and K. Manchester, The Archaeology of Disease (2nd ed., New York, 1995), 137.
78 Rogers and Waldron, op. cit. in note 74.
79 Farwell and Molleson, op. cit. in note 15.
80 D. J. Ortner and M. F. Erikson, ‘Bone changes in the human skull probably resulting from scurvy in infancy
and childhood’, International J. Osteoarchaeol., 7 (1997), 212–20. D. J. Ortner, E. Kimmerle and M. Diez, ‘Skeletal
evidence of scurvy in archaeological skeletal samples from Peru’, American J. Physical Anthropol., 108 (1999), 321–31.
D. J. Ortner, W. Butler, J. Cafarella and L. Milligan, ‘Evidence of probable scurvy in subadults from archaeological
sites in North America’, American J. Physical Anthropol., 114 (2001), 343–51.
48 neil holbrook and alan thomas
fig. 24
Ulna and first metatarsal of Burial 848 with marginal erosive lesions affecting proximal joints and suggestive of
gout. Photograph by Medical Illustration, Bristol Royal Infirmary.
none were the lesions pathognomonic of scurvy. Diagnosis, therefore, remains
speculative.
a monastic cemetery at llandough 49
fig. 25
Enamel hypoplasia. Top: linear horizontal furrows (Skeleton 242, 33–45 year-old male); Bottom: pits of
depressed enamel (Burial 458, 3–5 years old). Photographs by Medical Illustration, Bristol Royal Infirmary.
50 neil holbrook and alan thomas
fig. 26
Fractures involving the parietal bones of three
crania. The fracture in A is linear with well-defined
clean edges indicating a sharp force injury, possibly
that of a blade. The fractures in B and C are
curvilinear suggesting blunt force trauma, possibly
from a blunt weapon given their location. None
fulfil all of the criteria for the identification of
weapon-related trauma (for example, the polished
and flaked up surfaces that are associated with
blade injuries), although post-mortem breaks in
addition to remodelling, indicative of healing, may
obscure diagnostic features. A = Burial 124,
30–40-year-old male; B = Burial 103, 20–40-year-
old male; C = Burial 345, adult male. Photographs
by Medical Illustration, Bristol Royal Infirmary.
a monastic cemetery at llandough 51
Perhaps a clearer picture of the nutritional status of the population is to be
gleaned from three non-specific conditions: enamel hypoplasia, cribra orbitalia,
and porotic hyperostosis. Enamel hypoplasia is identified on teeth as sharp lines,
furrows, grooves, or pits of depressed enamel (Fig. 25).81 It is the result of disruption
to the formation of enamel during growth and therefore represents health stress
encountered during the growing years. There are many factors besides malnutrition
that can cause this condition, infection and childhood illnesses such as measles
being among these.82 In the Llandough population 6.7% of all dentitions had one
or more hypoplastic defects. Defects were more frequent among skeletons from
Areas II and III combined (19.7%) compared to Area I (6.2%). Overall, levels of
enamel hypoplasia were not particularly high, implying intermittent periods of
poor health rather than chronic physiological disturbance.
Cribra orbitalia and porotic hyperostosis, also believed to reflect childhood
health stress, were identified as small porosities or large inter-connected trabeculae
on the roof of the orbits (cribra orbitalia) and the external surface of the skull vault
(porotic hyperostosis).83 Both conditions were infrequent, affecting 35.7% of all
orbits and 5.7% of all skulls. Overall, they were more common and more severe
among skeletons buried in Areas II and III compared with those buried in Area
I. Frequencies were not significantly different to other samples, if a little lower.
It is thought that cribra orbitalia and porotic hyperostosis are caused by iron
deficiency anaemia.84 Given the relatively low frequencies, added to the fact that
most lesions were not severe, the overall impression is that the inhabitants of the
Llandough area experienced relatively mild forms of iron deficiency anaemia
during their lives.
Infection was observed in 0.1% of the population or 0.6% of all skeletal
elements that could be observed. It was more frequent compared to other samples
and involved more males than females.85 Almost all lesions were represented by
periostitis. Besides infection, periostitis may also arise in other conditions such as
mild trauma and varicose veins. A more detailed analysis of the distribution of
periostitis in the Llandough population suggested that most lesions were related to
these relatively minor conditions. Most were isolated, were of minimal or moderate
expression, and most involved the lower limb bones. Specific infections were not
generally prevalent, although diagnosis was hampered by poor preservation. Those
suggested are tuberculosis, osteoarthropathy, brucellocis and the previously
mentioned septic arthritis. Evidence for syphilis and leprosy was less determinative
although diagnosis, based on the distribution of lesions around the skeleton, was
hindered by the incomplete specimens. These diseases were not necessarily absent.
Most trauma was represented by healed fractures, involving 11% of the
population, or 0.2% of all elements that were observed. These affected more males,
and tended to involve the upper regions, especially the clavicle and the forearm.
81 S. Hillson, Teeth (Cambridge, 1986).
82 Roberts and Manchester, op. cit. in note 77.
83 P. Stuart-Macadam, ‘Nutritional deficiency diseases: a survey of scurvy, rickets, and iron-deficiency anaemia’,
201–22 in M. Y. Ićcan and K. A. R. Kennedy (eds.), Reconstruction of Life from the Skeleton (New York, 1989).
84 Stuart-Macadam, op. cit. in note 83.
85 Loe, op. cit. in note 62.
52 neil holbrook and alan thomas
Analysis of the distribution of fractures and fracture patterns indicated that the
majority were associated with indirect forces, such as accidental falls.
There was a near absence of fractures affecting sites that are typically involved
during acts of aggression (e.g. the facial skeleton and crania). Exceptions are a
group of fractures which, in modern clinical contexts, are associated with
violence.86 These include one possible scapula fracture (B838), five fractured
metacarpals involving four skeletons (B91, 119, 225 and 329) and, less conclusively,
fractures involving the forearms of three females and three males. In addition,
three possible weapon injuries were identified involving three crania (B103, 124
and 345). None of these fulfil all of the criteria for the identification of weapon-
related trauma but given their location (among other things), they are possible
examples. All show bony remodelling indicating that they were not themselves
fatal (Fig. 26).
Overall, fracture patterns were typical of a rural population for which farming
activities are the most likely cause of injury.87 These patterns were not significantly
different between burial areas. However, a wider distribution of elements was
involved in skeletons from Areas II and III compared with Area I, perhaps
indicating that the people buried in the former engaged in more diverse activities.
These are the main findings of the skeletal analysis. But what of the overall
health status of the Llandough population? It is difficult to draw conclusions about
the mortality and morbidity experience of an assemblage that is undoubtedly
biased by many factors such as preservation, sampling and cultural circumstances.
However, the impression is of a population that enjoyed an adequate diet, perhaps
more than adequate in some instances if we consider the example of gout. Males
tended to have experienced more pathological conditions than females, possibly
because they were more mobile and therefore exposed to more pathogens and
stressors. Conditions that relate to childhood experiences (e.g. cribra orbitalia and
enamel hypoplasia) were more common than those that relate to adulthood (e.g.
trauma and infection), and a greater burden of disease was indicated in Areas II
and III than in Area I. In general, the small amount of evidence for long-standing
conditions and violence-related trauma suggest that the inhabitants of Llandough
enjoyed favourable living conditions. Although not dissimilar in their overall health
and physical attributes, lower levels of disease and a tendency towards a higher life
expectancy are suggested compared to other contemporary populations.
THE FINDS
In the following sections the various categories of artefact recovered from the
excavation are catalogued and discussed. The finds are crucial to the understanding
and interpretation of the cemetery. For instance, in a number of cases there is
uncertainty as to whether an artefact was deposited in the grave incidentally during
backfilling or was deliberately deposited as an item of grave goods. Information on
86 J. Crawford Adams, Outline of Fractures (New York, 1983); L. F. Rogers, Radiology of Skeletal Trauma (New York,
1992).
87 M. A. Judd and C. A. Roberts, ‘Fracture trauma in a medieval British farming village’, American J. Physical
Anthropol., 109 (1999), 229–43.
a monastic cemetery at llandough 53
fig. 27
Burial 630 showing the iron bands in situ,
prior to lifting. © Cotswold Archaeology.
the completeness of an artefact and its location within the grave can in some cases
assist in differentiating between these two possibilities. It is equally important that
the Roman artefacts, including the pottery found within the grave fills, are fully
reported as they are of importance when considering questions such as whether the
cemetery originated in the Roman Period, or the survival and use of Roman
artefacts in the post-Roman period.
an early-medieval girdle from burial 631
By mark redknap
During the first stages of the excavation of the grave fill of B631*, a large
number of fragments of corroded iron ‘banding’ were found lying above and in
front of the pelvis of the extended body (Fig. 27). Following discussions with the
National Museums & Galleries of Wales (NMGW), it was decided that the
ironwork should be lifted as a block for excavation in laboratory conditions.
54 neil holbrook and alan thomas
Recovery
Once the skeleton, which lay on its back (head to the west), had been fully
uncovered and recorded, the legs and upper body were lifted by the excavation
team. The whole pelvic area which remained, including the lower arms, wrists,
vertebrae and associated rib fragments, was lifted as a single block encased in
Plaster of Paris bandage and transferred to the Conservation Laboratory of the
Department of Archaeology & Numismatics at the NMGW, Cardiff.88 Field
conservation was undertaken by Diane Dollery, and laboratory conservation by
Kathy Bird and Penny Hill.
In the laboratory, the block was x-radiographed, and a wooden box was
constructed around it. This was then filled with a polyurethane foam to provide a
support during excavation. The block was carefully excavated in an inverted
position. Each stage of exposure was traced on melinex film, providing an
additional record of the details of the precise orientation of bones and iron, and
their relationships.
As the soil was removed, it became clear that there were two (not one)
fragmentary bands of iron lying across the back of the pelvis (Fig. 28). Across the
front of the pelvis lay two further bands, one fragment orientated at right angles to
the other fragments. The disposition of these two iron fragments and of some bones
within the block suggest that the pelvic area of the body had been disturbed by
animal activity and elements displaced. This is also supported by the discovery of
the jaw and teeth of a small rodent, within the corrosion of one fragment.
Description
The original surfaces and fractures of the iron fragments remained in good, if
unstable, condition. With careful cleaning of the fractures, it has been possible to
re-orientate the various pieces, and to re-assemble them in their original positions.
This has revealed two separate, parallel bands of iron, designated ‘upper’ (Fig. 29)
and ‘lower’ (Fig. 30). These iron bands vary from 20 to 30 mm in width. All
fragments are slightly convex to the body, varying in cross-section from flat back
with slightly raised edges, to shallow U-shaped guttering. No decorative detail has
been found on the banding. In both cases, the ends terminate and meet in the
centre of the back, close to the spine. The method of fastening the ends together
was in both cases by means of a ‘hook’ forged out of one end and being passed
upwards through a loop forged on the other, the point being bent over to secure
the junction. The loops on both bands had slightly ‘waisted’ or tapered edges.
The corrosion products on the concave sides (i.e. those that face away from
the body) of the fragments contained impressions of organic matter which were
consolidated. Mineral-replaced wood was present in two discrete areas above the
vertebrae, the wood rays being aligned parallel to the vertebral column in both
cases.89
88 For details of the lift see K. Bird, Excavation and investigation of iron banding associated with a human burial
from Llandough (unpubl. diss., University of Durham, 1995).
89 Bird, op. cit. in note 88.
a monastic cemetery at llandough 55
fig. 28
Top: the fragments after excavation of the block, showing their relative positions in the ground.
Bottom: Detail of the upper band, showing conserved sections of ironwork. © National Museum of Wales.
56 neil holbrook and alan thomas
fig. 29
The upper band. Scale 1:2; details 1:1. Drawing by Tony Daly; © National Museum of Wales.
Impressions of textile were found on the external surface of three adjacent
sections of the lower iron band, within the concave ‘guttering’. These impressions
have been examined by Hero Granger-Taylor and identified as a coarse tabby of
average quality, with open weave and loosely spun threads of long fibres, indicative
of linen. Both systems are Z-spun, with 12 threads per cm longitudinally, and 14
per cm horizontally. The weave is diagonal to the iron banding, suggesting
accidental impressions of textile pressed tightly against the banding. The impres-
sions are on frontal sections of the banding, which would have been uppermost in
the grave at the time of burial. Backfilling of the grave may therefore have pushed
a monastic cemetery at llandough 57
fig. 30
The lower band. Scale 1:2; details 1:1. Drawing by Tony Daly; © National Museum of Wales.
the cloth against the ironwork, but it is not possible to establish whether the
impressions represent a shroud or clothing.
Casts of cyclorrhaphous fly puparia were found on approximately half of the
fragments (identified by John Deeming, Curator of Diptera, NMGW) during
cleaning. The majority of these casts, preserved by the formation of iron-corrosion
products around the abandoned puparia cases, appear on the outside of the bands,
58 neil holbrook and alan thomas
though a small number have also been noted on the inner face of the bands. Two
seed cases were also noted within the corrosion products.90 The convex side of the
banding was featureless (no organic remains).
The object is now in the collections of the Department of Archaeology &
Numismatics, NMGW (NMW acc. no. 95.56H)
Date
The radiocarbon date for the skeleton in B631* (i.e. when metabolism ceased)
gave a result of 1530±80 B.P. (Wk-6938) (cal. a.d. 361–662 at 2 sigma calibration)
or a.d. 340–660 using OxCal c2.18. A 5th- or 6th-century date is compatible with
the other archaeological evidence, such as the position of the burial within
cemetery.
Discussion
The skeletal evidence indicates that B631* was a robust adult male, aged
25–35, and about 1.75 m in height (based on left radius).91 One rib had a healed
fracture. It is clear from the surviving remains that the iron bands enclosed the
waist of the adult, and that no attempt appears to have been made to remove them
on his death. Their purpose is less clear, but two interpretations are indicated by
historical sources: penitential or hernia belts. No changes were observed on the
skeleton that would suggest a medical condition requiring a surgical truss.
Bands made of iron were sometimes worn as a form of penance to expiate sins
committed during life (‘penitential belts’). Most recorded instances post-date the
6th century, a late practitioner being King James IV of Scotland (1473–1513) who
wore an iron belt under his shirt in penance for the death of his father at
Sauchieburn (1488), discovered on his death at the Battle of Flodden (1513).92
One of the direct inheritors of the austere ascetic traditions of the East was
John Cassian (c. a.d. 360–435), a native of Dobrudja (Romania/Bulgaria) on the
Black Sea, who after visiting the famous monastic communities of the eastern
deserts, was active in Gaul by a.d. 415, sharing his experience of eastern asceticism
with a new audience. His writings were known in Anglo-Saxon England, but this
type of spirituality is thought to have grown in popularity in the 12th century. At
the basis of the penitential life was the Christian realisation of sin.93 In the 11th and
12th centuries an eremitical life could provide a career of expiation — the purging
of the soul by hurting the body. The characteristic external austerities of the
penitential life as practised in England around this time are set out in the Ancrene
Wisse (12, 10–12) — bodily pain, with fasts, vigils, scourgings, harsh clothes, rough
lodging, disease, and hard physical work. The ascetic life was thought of as a life of
service, particularly military service (the militia Christi). The coats of mail worn by
some hermits (loricati) seem to have been put on, not to burden the flesh so much
as to defend against the physical assaults of the devil: ‘I know one man who wears
90 Bird, op. cit. in note 88.
91 Pers. comm. Louise Loe.
92 L. Stephen and S. Lee (eds.), The Dictionary of National Biography, vol. X (London, 1921–2), 582–90 at p. 582.
93 G. Shepherd (ed.), Ancrene Wisse Parts Six and Seven (Manchester 1959), xxiv.
a monastic cemetery at llandough 59
at the same time a heavy coat of mail and a hair-shirt fastened painfully about his
waist, thighs, and arms with broad bands of iron.’94 In one case recorded by Peter
Damian, an Italian called Dominic, the so-called loricatus (a.d. 955–1060) ‘wore for
about three years an iron corslet next to the skin, but girt about the body with two
iron hoops and his arms fastened by another pair’.95 An early loricatus in Wales was
a William, knight of Hugh Lacy, who withdrew around a.d. 1100 to Llanthony
Abbey (Mons.) and wore ‘An habergeon with which to defend himself against the
darts of the Enemy’.96 One English visionary, Godric of Finchale, is said to have
worn out three shirts of chain-mail during his period of seclusion.97 Another form
of religious devotion recorded by Walter Map is the case of an English hermit who
had a long iron chain with iron peg attached to his leg, which restricted his
movement by the week.98
The preacher, Caesarius of Heisterbach (c. 1170–c. 1240), whose ecclesiastical
histories and ascetical writings were very popular in Germany, recorded in his
early 13th-century Dialogue on Miracles two instances of ‘sinful women’ wearing such
bands as punishment. In one account, a nun named Clementia, in the convent of
St Gertrude, at Nivelle in Brabant, ‘gave her whole time to washing with tears each
day the tunic of her innocence, which she had polluted . . . with sins of the flesh . . .
When she was dying, there were found round her body nine iron bands’.99 In
another account, a maiden of Liège called Osilia, grieving for her sins, became a
nun near the church of St Severin, Metz, where:
. . . in exceeding remorse she often shed tears of blood; moreover she had bound herself with
bands of iron. But as she stood at prayer with these upon her, they all burst asunder. Now
fearing that those chains had not been acceptable to God, she bound herself again. When she
revealed this to the abbot of S. Lambert, he said to comfort the sorrowing woman: ‘He who
said to Mary Magdalen ‘‘The sins are forgiven thee,’’ has broken thy bonds.’100
Caesarius described this event as having happened ‘not long since’, and presumably
it can be attributed to the late 12th or early 13th centuries.
Closer in time and place to Llandough, a visionary brother at the monastery
of (Much) Wenlock (Shrops.) was instructed by angels:
. . . to confess all his own sins, which the impure spirits had imputed to him, to this priest
[Begga] making amends for them as he judged fit; and, as a token of the angelic instruction he
was to testify to the priest that unknown to men he [the priest Begga] had worn an iron girdle
round his loins for love of God . . .101
This episode is recorded in a letter from Boniface to his correspondent Eadburg,
before 719, and Boniface may have encountered the visionary in England before
he went to Frisia in 716. The visionary is thought to have been a full member of
94 Ancrene Riwle fol. 103v, 11–12, trans. M. B. Salu, The Ancrene Riwle (London, 1955), 169.
95 Patrologia Latina 144, col. 1017; Shepherd (ed.), op. cit. in note 93, 51.
96 Shepherd, op. cit. in note 93, 51.
97 Ibid.
98 Walter Map, De Nugis Curialum, ed. and trans. M. R. James, rev. C. N. L. Brooke and R. A. B. Mynors, Courtier’s
Trifles (Oxford, 1983), note 55, 131.
99 Caesarius of Heisterbach, Book XI, Ch. XXVIII, trans. H. Von E. Scott and C. C. Swinton Bland, The Dialogue
on Miracles by Caesarius of Heisterbach (1220–1235), (London, 1929), 262–3; C. Daniell, ‘When penance continued in
the grave’, British Archaeol., 19 (1996), 7; idem, Death and Burial in Medieval England 1066–1550 (London, 1997), 157.
100 Caesarius of Heisterbach, Book XI, Ch. XXIX: op. cit. in note 99, 263–4.
101 P. Sims-Williams, Religion and Literature in Western England, 600–800 (Cambridge, 1990), 243 and 246.
60 neil holbrook and alan thomas
Abbess Mildburg’s community at Wenlock (though he was not called ‘brother’,
frater). The priest Begga was an ascetic, and appears to have worn an iron girdle
round his loins for many years.102 The device around his waist would appear to
belong to the same category of penitential device as that recorded by Caesarius of
Heisterbach in the 13th century.
An alternative interpretation for the Llandough object is that provided by the
occasional discovery of iron banding around the pelvic area of skeletons on the
Continent, notably from Merovingian graves in France and Germany, where they
have been identified as hernia belts or Bruchbänder. There is no evidence on the
Llandough skeleton for a medical injury requiring treatment, but an ailment such
as a hernia would not leave any trace. Hernia is a condition in which there is a
projection of part of an organ or tissue through a weak area in the muscle or tissue
that normally contains it, the projection appearing as a tumour or swelling on the
outside. In males, this frequently occurs in the groin (inguinal or scrotal hernia),
while in women it can occur in a similar region or towards the centre of the upper
thigh (femoral or crural hernia), and in children it can occur at the umbilicus
(umbilical hernia). Hernias can also occur in the central upper abdomen (epigastric
or abdominal hernia), in which the intestine bulges out at a point between the
navel and the breastbone. This can be three times more common in men than
women, and it is most likely to occur in people between the ages of 20 and 50. An
incisional hernia is when the intestine bulges through a weakness caused by a
wound or surgery; in modern cases a mesh may be required to repair the bulge.
Hernias are often caused by extra pressure being exerted on weak places, through
lifting heavy objects or frequent coughing. Treatment usually involves returning
the protrusion within the belly and keeping it there (reducible hernia) with purpose-
made pads kept in place by a strong spring band that passes round the body.
Trusses have to be carefully fitted over the place from which the rupture comes,
and it is clear from the Merovingian examples that the metal pads have been
placed directly over such spots (usually lower abdomen, suggesting epigastric or
inguinal hernias).
The first medical references to hernias go back far in time, and they must have
been an occupational hazard in antiquity. Direct evidence for their treatment can
be found in mummies from the necropolis at Sakkara dated ca. 2500 b.c.103
Hernias were well known in the Classical world. Medical authors such as
Hippocrates (c. 460–375 b.c.) or Aulus Cornelius Celsus (first half 1st century a.d.)
described treatment principles which formed a basis for treatment well into the
Middle Ages. In his De medicina, Celsus mentions fixing the intestine with a ‘strip of
linen, to one end of which has been stitched a ball of rags’.104 In a well-known
epigram, the Roman poet Martial (c. a.d. 40–104) names a number of surgical
specialists, including one specialising in healing hernias: enterocclarum fertur Podalirius
102 Ibid., 248.
103 P. Ghalioungui, Magic and Medical Science in Ancient Egypt (1963).
104 Celsus, On Medicine, Books VII-VIII (trans. W. G. Spencer, Loeb, Cambridge and London, 1994), 20.
a monastic cemetery at llandough 61
Hermes, ‘. . . Hermes is said to be the Podalirius of hernias’.105 The Naturalis Historia
of Pliny the Elder (Gaius Plinius Secundus, a.d. 23–79) refers to hernias: inguinal
hernias are called enterocele (XXVII, 89); umbilical hernias are called umbilici
eminentia (XXVI, 49).106 Hernias are thought to be represented on Gallo-Roman
anatomical ex-votos (wooden and bronze) which show the torso with a protuber-
ance adjacent to the genitals or on the abdomen.107
The Compendium Medicinae written about 1240 by the English physician
Gilbertus Anglicus (born c. 1180), contains advice on the care of those suffering
from conditions such as hydrocele, orchitis and hernia.108 According to the Chirurgia
of Roger of Salerno, written about a.d. 1150, hernia was first treated by medication,
and then with a truss. If these failed, surgery could be performed, but was risky.109
The first direct archaeological evidence for hernia belts consisting of iron
bands around the waist was first noted from Merovingian contexts by antiquaries
in the 19th century, such as A. Danicourt who described one from Marchélepot,
Dép. Somme, in 1884. In 1900, the Belgian doctor Victor Deneffe produced a
small monograph on Merovingian Bruchbänder in a series on Chirurgie der Antike,
citing Frankish examples from Devise and Fluy (Dép. Somme).110 More recently,
further examples have been discovered in France at Lavoye (Dép. Meuse), Grave
126, dated to the mid-7th century,111 and from grave 32, Royaumeix/Menil-la-
Tour (Meurthe-et-Moselle), which was attributed to the second half of the 6th
century;112 in Germany at Schweningen, Schwarzwald-Baar-Kreis (dated to the
second half of the 7th century),113 in Switzerland at Bülach (two examples) and
Matzingen, and a Visigothic example from Spain at Acequilla near Madrid.114
Examples have also been reported from Westgothic and Langobardic graves, and
from post-Merovingian contexts in the north.115 The form of the Continental belts
is fairly distinctive, each being forged to specific dimensions to fit the patient. An
105 R. Jackson, Doctors and Diseases in the Roman Empire (London, 1988), 123; Martial, Epigrams (ed. and trans. D. R.
Shackleton Bailey (vol. 2, Cambridge Mass. and London, 1993), X, 56; G. Penso, La médicine romaine: L’art d’esculape
dans la Rome antique (Paris, 1984), 110.
106 Pliny, Natural History (trans. W. H. S. Jones, Loeb, Cambridge and London, 1956), Bks. 20–3 (pp. 322–3);
Penso, op. cit. in note 105, 330.
107 R. Bernard and P. Vassal, ‘Études médicales des ex-voto des sources de la Seine’, Rev. Archéol. Est et Centre-Est,
9 (1958), 328–37 (Abb. 124); Penso, op. cit. in note 105, 110, figs. 149, 150 and 152; M. Vauthey and P. Vauthey,
‘Les ex-voto anatomiques de la Gaule Romaine (Essai sur les maladies et infirmités de nos ancêtres)’, 111–17 in
A. Pelletier (ed.), La médicine en Gaule: villes d’eaux, sanctuaires des eaux (Paris, 1985) at p. 114; S. Deyts, Les bois sculptés
des sources de la Seine (42nd supplement to Gallia, Paris, 1983), no. 73, pl. XXII; Jackson, op. cit in note 105, 164.
108 S. Rubin, Medieval English Medicine (Newton Abbot, 1974), 203.
109 R. S. Gottfried, Doctors and Medicine in Medieval England 1340–1530 (Princeton, 1986), 211.
110 V. Deneffe, Chirurgie antique: les bandages hernaires a l’époque mérovingiennes (Antwerp, 1900/90).
111 G. Joffroy, Le Cimetière de Lavoye (Meuse): nécropole mérovingienne (Paris, 1974), 115, Taf. 14, 1; L. Buchet, ‘Die
Landnahme der Franken un Gallien aus der Sicht der Anthropologen’, 662–7 (Abb. 532) in Die Franken: Wegbereiter
Europas. Vor 1500 Jahren: König Chlodwig und seine Erben (Mainz, 1996).
112 A. Liéger and R. Marguet, ‘Le cimetière Mérovingien de Royaumeix/Menil-le-Tour (Meurthe-et-Moselle),
Rev. Archéol. L’Est et Centre-Est, 43 (1992), 99–149, esp. fig. 32. Others are reported from Euville (Meuse), Behaut
(Vosges), and Lay-Saint-Christophe (Meurthe-et-Moselle): ibid., 141–2.
113 K. W. Alt and G. Oehmichen, ‘Ein frühmittelalterliches Bruchband von Schwenningen. Schwarzwald-Baar-
Kreis’, Fundbericht Baden-Württemberg, 17 (1992), 407–22 at p. 408.
114 G. C. Koenig, ‘Schmane und Schmied, Medicus und Mönch: ein Überblick zur Archäologie der
merowingerzeitlichen Medizin im südlichen Mitteleuropa’, Helvetia Archaeol., 51/52 (1982), 75–154 at pp. 129–30;
R. Baumgartner, ‘Fußprothese aus einem frühmittelalterlichen Grab aus Bodanuz’, Helvetia Archaeol., 51/52 (1982),
155–62.
115 Koenig, op. cit. in note 114, 130.
62 neil holbrook and alan thomas
iron band encircles about half the body, and one end of the band has been forged
into an expanded end, bent to position the plate over the ailment. Loops in the
band enabled the wearer to fasten the device by means of a leather strap.
In Britain, such objects are extremely rare. Deneffe cited an ‘Anglo-Saxon’
example in the collections of the British Museum attributed to the 13th century,
though inquiries have so far failed to locate such an object.116 The medieval
skeleton of an adult male excavated in the N. transept of Merton Priory (Greater
London) was furnished with an accessory over the pelvis which has been interpreted
as a hernia belt.117 This has a very different form to the published Merovingian
examples, comprising a belt which did not continue around the body, with a wide
downward extension on the front right hand side of the body. As in the case of the
Llandough find, the Merton Priory fitting was made from iron beaten into a thin
sheet, without any evidence for rivet holes or joins. Fibrous material overlying the
belt was identified as a woven textile, and though degraded, animal fibre, possibly
wool, was suggested by the scale pattern observed by scanning electron microscope.
The sporadic archaeological survival of trusses provides a poor reflection of
the fairly continuous history of treatment, and some hernia trusses may have been
made of perishable materials. The hernia was an occupational hazard for many,
resulting from the strain of lifting, hauling and similar exertions. A hernia truss
developed by Geiger and introduced in 1651 overlapped and fastened at the front
of the pelvis, the strap passing right around the body.118 Samuel Pepys’s father had
a rupture of some 20 years standing which occasionally escaped the truss which
supported it, causing him great pain.119 An early 19th-century example from
Wales, formerly in the possession of Mr Owen Griffith (died 1865) of Tryfau
(Caernarfonshire) has the form of a plain band of iron which encircled the waist,
and provides a good parallel to the Llandough find (Fig. 32).120 Lumbar support
belts for back pain provide firm support for the vulnerable lower lumbar area, but
these tend to support the back, rather than frontal area. Very little of the iron
banding from Llandough has been lost through poor preservation, and it has been
possible to estimate the circumference or waist measurement of both bands at
about 32 in./81 cm; the small diameter suggests that they must originally have
fitted above the hips, at epigastric or umblical level. This was the position of the
fragments on the body when first uncovered (Fig. 27).
Original appearance
Evidence for textile has been noted on the interior face of the expanded
terminal of a 6th-/7th-century example from Champigneul-sur-Vence, while the
iron band from grave 45 at the cemetery at Bülach (Kt. Zürich, Switzerland), of
similar date, has been reconstructed with a leather sleeve and belt.121 An example
116 Deneffe, op. cit in note 110; Alt and Oehmichen, op. cit. in note 113, tab. 1, 411.
117 B. White and H. Ganiaris, ‘Excavating bodies: excavating and analysing human skeletons’, 14–21 and fig. 6 in
A. Werner (ed.), London Bodies (London, 1998); H. Ganiaris, ‘Merton Priory hernia belt’, MoLAS report (in prep.).
118 E. Bennion, Antique Medical Instruments (London, 1979), 154.
119 L. M. Beier, SuVerers and Healers: The Experience of Illness in Seventeenth Century England (London, 1987), 147.
120 Museum of Welsh Life: NMW acc. no. 33.237/13.
121 Koenig, op. cit. in note 114, 127.
a monastic cemetery at llandough 63
fig. 31
Conjectural view of bands, seen from the front. Drawing by Tony Daly; © National Museum of Wales.
fig. 32
Early 19th-century iron hernia belt (‘truss-iron’), formerly in the possession of Owen Griffith: MWL/NMW
acc. no. 33.237/13. © National Museum of Wales.
from grave 17 at Acequilla near Madrid, appears to show some similarity to the
Llandough find, with the fragmented strap passing right around the front of the
pelvis from behind.122 All such ancient devices appear to have been forged in one
piece, and some may have been sheathed in a softer material to prevent chafing.
Modern trusses have in principle two functional elements, an elastic spring (Feder)
and a pad (Pelotte). An elastic, canvas or padded metal appliance can be used to
hold an abdominal hernia in place — in antiquity, all forged in one piece. The
122 Koenig, op. cit. in note 114, fig. 71; L. Vasques de Parga, ‘Informe provisional sobre as excavaciones
arqueológicas en Azuquecal (Guadajara), Finca de Acequilla, Termino de la Cabana 1962’, Noticiario Arqueol. Hisp.,
7 (1963), 224 (Taf. 62).
64 neil holbrook and alan thomas
strap was only slightly elastic, being thick and not varying in thickness — unlike
Llandough.
Despite the literary references to penitential devices, there seems to be little
direct archaeological evidence to support an identification of the Llandough object
as such worn ‘about the loins’ as described in the episode involving the visionary of
Wenlock. However, he does remind us of the existence of very religious laymen
and conversi in Early-medieval Britain. For example, Guthlac spent nine years as a
soldier, partly on the Welsh border, and then entered the monastery at Repton;
after two years he began a solitary life in the Fens (c. 704–16). Samson of Dol first
moved from Llantwit Major to Caldey Island, then to a cave near the Severn,
before becoming a bishop and, in the course of travelling, setting up three
monasteries in Brittany and Cornwall.
If the Llandough object had served as a hernia belt, it provides rare evidence
for palaeopathology which does not usually manifest itself on skeletons. There is
one significant difference between many of the Continental hernia belts and the
Llandough example — the ability in the case of the former for the appliance to be
removed by the wearer. This would have been impossible in the case of the
Llandough straps, as a second person would have been needed to hammer the iron
points over behind the wearer when putting them on, and similarly, someone
would have needed to hammer the hooks open in order to release the bands from
the body. While this raises some doubts about their function as a temporary
remedy, the care with which both belts were shaped to reduce the risk of chafing
favours this interpretation. With the emphasis on punishment and penance, would
a penitential belt have been constructed in this way? It is not difficult to imagine
penance being proposed as an explanation for iron hernia belts found in medieval
burials. Perhaps a higher significance became accepted by some wearers — a
remedy for hernia being regarded at the same time as a penance for sin? Methods
of deadening pain were rudimentary, and much discomfort would have been
caused by such chronic conditions, which were probably fairly common. In the
later Middle Ages, sufferers may have tried hard to accept the pain with humility,
on the assumption that they would be spared a worse ordeal in purgatory; it has
been suggested in such cases that the promise of a speedier passage through
purgatory may have brought partial relief.123
the coins
By edward besly
Seven Roman coins of the late 3rd and 4th centuries were discovered, as
follows.
1. Tetricus I, in Gaul a.d. 271–4; radiate, mint I (Trier), reverse SPES PVBLICA. Elmer
764/Cunetio 2610.124 Corroded. 007, ?former cultivation soil; NMGW 95.56H/2.
2. Urbs Roma; nummus, reverse wolf and twins, //TR·P; irregular, d. 16 mm. After a.d.
330. Unworn? B87*; NMGW 95.56H/4.
123 C. Rawcliffe, Medicine and Society in Later Medieval England (Stroud, 1995), 9.
124 Elmer: G. Elmer, ‘Die Münzprägung der gallischer Kaiser in Köln, Trier und Mailand’, Bonner Jahrb., 146
(1941), 1–106. Cunetio: E. Besly and R. Bland, The Cunetio Treasure (London, 1983).
a monastic cemetery at llandough 65
3. House of Constantine I, uncertain ruler; nummus, reverse GLORIA EXERCITVS, one
standard type, chi-rho // [PLG?]. c. a.d. 335–40. Unworn. B93; NMGW 95.56H/5.
4. Theodora, commemorative issue; nummus, reverse PIETAS ROMANA, - // [?] (Trier).
c. a.d. 337–40. Unworn. B93; NMGW 95.56H/6.
5. Theodora, commemorative issue; nummus, as no. 4. Unworn. B110; NMGW 95.56H/7.
6. Constantius II, a.d. 337–61; nummus, reverse VICTORIAE DD AVGG Q NN,
branch // [. . .]. Trier, cf. RIC VIII Tr. 204, a.d. 347–8.125 Corroded but probably not
worn. Unstratified; NMGW 95.56H/8.
7. Uncertain; certainly 4th-century, but details of reverse type are unclear through
corrosion. Perhaps an irregular Fel Temp Reparatio ‘fallen horseman’ type, d. 15 mm,
mid-350s. B59; NMGW 95.56H/3.
the non-ceramic artefacts
By mark redknap
Roman and Early-medieval
Glass
Beads
1. Large melon bead in light blue/turquoise frit. Roman, 1st–2nd century.
Diameter 27 mm (B40; NMW 95.56H/19; Fig. 33, 1).
2. Small annular bead. Pale green metal. Diameter 9 mm (B547; NMW 95.56H/
21; Fig. 33, 2).
3. Small annular bead. Pale green metal. Diameter 8.5 mm. Roman (B547; NMW
95.56H/22; Fig. 33, 3).
4. Small annular bead. Pale green metal. Diameter 8.5 mm. Roman (B547; NMW
95.56H/23; Fig. 33, 4).
5. Small annular bead. Dark blue metal. Diameter 7.2 mm. Roman (B547; NMW
95.56H/24; Fig. 33, 5).
6. Annular bead. Pale green metal, iridescent surface. Diameter 9 mm. Roman
(unstratified; NMW 95.56H/25; Fig. 33, 6).
7. Small dark annular bead. Dark green glass, iridescent surface. Diameter 4.5
mm. Roman (B624; NMW 95.56H/26; Fig. 33, 7).
8. Small opaque turquoise annular bead with flat ends. Diameter 3 mm. Roman
(B934; NMW 95.54H/388; Fig. 33, 8).
9. Large opaque dark blue annular bead. Diameter 21 mm (B851; NMW 95.56H/
387; Fig. 33, 9).
Melon-shaped beads of the type represented here in turquoise glass paste are
found in Britain in 1st- and 2nd-century contexts.126 The small undecorated
annular beads of spheroid form (Guido Group 7) have slightly differing diameters,
suggesting that some may have originated from graduating necklaces, while others
may have been on necklaces alongside beads of a variety of sizes. They are difficult
to date. Roman examples of similar small size (2.5–9 mm in diameter) in green or
125 RIC VIII: J. P. C. Kent, The Roman Imperial Coinage. Volume VIII: The Family of Constantine I ad 337–364 (London,
1981).
126 N. Crummy, The Roman Small Finds from Excavations in Colchester 1971–9 (Colchester, 1992), 30.
66 neil holbrook and alan thomas
fig. 33
The Roman and early medieval glass. Scale 1:1. Drawing by Tony Daly; © National Museum of Wales.
opaque blue glass have been published from Drain Group 4 (c. a.d. 160–230) at
the Legionary Fortress Baths at Caerleon;127 Guido lists further examples in natural
greenish translucent glass from Coygan Camp (Carmarthenshire), Merthyr Mawr
Warren (Bridgend), Whitton Roman villa, near Barry (Vale of Glamorgan),
Caerwent Roman town (Mons.), and blue beads from Caerwent and Usk
(Mons.).128 A bead from Cadbury-Congresbury (Avon) similar to no. 6 was
attributed to the pre-Roman/Roman Iron Age;129 bead no. 9 is similar in size to
an annular bead in blue-green glass from a late 4th-century context at Segontium
Roman fort, and has the same profile as one in blue-green glass (also with a pitted
surface) from Leucarum Roman auxiliary fort (Loughor, Swansea), attributed to the
1st century a.d.130 In light of the quantity of Roman pottery and other material
from the Llandough excavations, a Roman date appears likely for the Llandough
beads. However, some beads may have been manufactured from recycled Roman
bottle glass: an example from Dinas Powys was described as either Roman or
‘Teutonic’.131 Beads nos. 2–5 were found in the fill of B547, which also contained a
burial pebble.
127 R. Brewer, ‘The beads and glass counters’, 146–56 in J. D. Zienkiewicz, The Legionary Fortress Baths at Caerleon:
Vol. II The Finds (Cardiff, 1986) at p. 147.
128 M. Guido, The Glass Beads of the Prehistoric and Roman Periods in Britain and Ireland (London, 1978).
129 P. Rahtz et al., Cadbury Congresbury 1968–73: A Late/Post-Roman Hilltop Settlement in Somerset (BAR Brit. Ser., 223,
Oxford, 1992), B0573.
130 D. Allen, ‘Roman glass’, 219–28 in P. J. Casey and J. L. Davies with J. Evans, Excavations at Segontium
(Caernarfon) Roman Fort 1975–79 (CBA Res. Rep., 90, York, 1993), fig. 13.4 no. 47: idem, ‘Glass’, 387–94 in A. G.
Marvell and H. S. Owen-John, Leucarum: Excavations at the Roman Auxiliary Fort at Loughor, West Glamorgan 1982–84
and 1987–88 (London, 1997), fig. 140, no. 53.
131 L. Alcock, Dinas Powys: An Iron Age, Dark Age and Early Medieval Settlement in Glamorgan (Cardiff, 1963), fig. 41, 6.
a monastic cemetery at llandough 67
Early-medieval Vessel Glass
10. Fragment of the rim of a cone-beaker with rounded rim, pale yellow metal with
opaque white marvered trails. Clear/pale bluish-green glass. Estimated rim
diameter about 80 mm, falling within the known range (60–159 mm). Late 6th or
7th century. (B640; Fig. 33, 10).
11. Small plain yellowish fragment; thin walled, some fine bubbles. Possibly Early-
medieval, but form uncertain (B482; NMW 95.56H/35; not illustrated).
While little of the contour of the body of the vessel represented by rimsherd
no. 10 survives, on the basis of profile, surface treatment, colour and method of
production, it probably belongs to a cone beaker, the largest category of glass
found at Whithorn.132 The source of the western British group is thought to be the
Continent (as yet unprovenanced, but possibly northern or western France; some
production has also been identified at Macquenoise, Belgium, during the 6th
century), and most vessels of this type have been dated to the late 6th or 7th
century, on the basis of associations with imported pottery (D and E wares).133
Conical beakers have been recognised at ‘high-status’ settlements in South Wales
such as Dinas Powys (where over 250 sherds were recovered during the 1954–8
excavations) and Hen Gastell near Swansea (38 fragments), as well as undefended
settlements such as Longbury Bank (Pembrokeshire), trading places such as Caldey
Island, and monastic sites such as Margam.134 Other sites producing similar glass
include Cadbury-Congresbury (Avon) and wics such as Hamwic.135 Cone beakers
have a fairly wide distribution, finding popularity across north-western Europe,
some occurring in Scandinavia. In the Rhineland many are attributed to Böhner’s
Stufe III, a.d. 525–600, though the form was popular among Germanic OYzieren
from the late 4th century. The Llandough sherd is important for it is complement-
ary to the material reported from Dinas Powys, and provides a further indication
of the sophisticated material culture circulating in the vicinity of the monastery in
the late 6th or 7th centuries. Cone beakers were certainly in use on sites in Wales,
though lumps of partially fused glass from Hen Gastell and Dinas Powys indicate
that broken glass was also collected and recycled.
Metals
Copper-alloy
1. Polden Hill-type brooch, complete. Length 50 mm. 1st century a.d. (B796*; Fig. 34, 1).
132 Group C1 in E. Campbell, ‘The early medieval vessel glass’, 297–314 at p. 297 in Hill, op. cit. in note 37, fig.
10.7, no. 18; fig. 10.10, no. 71).
133 Ibid., 297.
134 Alcock, op. cit. in note 131, fig. 40, nos. 26, 30; for colour plate: M. Redknap, The Christian Celts. Treasures of
Late Celtic Wales (Cardiff, 1992), 22; for reconstruction see E. Campbell, Imported Goods in the Early Medieval
Celtic West: with special reference to Dinas Powys (unpubl. Ph.D. thesis, University of Wales, College of Cardiff,
1991), illus. 64, no. 23; idem, ‘Continental glass’, 19–22 in P. F. Wilkinson, ‘Excavations at Hen Gastell, Briton
Ferry, West Glamorgan, 1991–92’, Medieval Archaeol., 39 (1995), 1–50; E. Campbell and A. Lane, ‘Excavations at
Longbury Bank, Dyfed, and early medieval settlement in South Wales’, Medieval Archaeol., 37 (1993), 15–77;
E. Campbell, ‘New finds of post-Roman imported pottery and glass from South Wales’, Archaeol. Cambrensis, 138
(1989), 59–66, Fig. 1.
135 Rahtz et al., op. cit. in note 129, 140; J. R. Hunter and M. P. Heyworth, The Hamwih Glass (CBA Res. Rep.,
116, York, 1998).
68 neil holbrook and alan thomas
2. Fragment of rod/shaft, of circular cross-section. Both ends appear truncated. Possibly
part of a pin from a brooch. Roman? Length 21 mm (B449; NMW 95.56H/12; Fig. 34, 2).
3. Small fragment of folded sheet (unstratified; NMW 95.56H/93; Fig. 34, 3).
4. Fragments of bronze sheet whose edges have incomplete perforations (B504; NMW
95.56H/3). Unlike a ‘perforated’ ornamental plate from Dinas Powys, decorated with
repousée dots which had worn away, these fragments have holes which appear to have
been drilled or punched.136 Lengths of fragments 9, 13 and 17 mm. The fragmentary
nature of this object makes identification uncertain; the lines of dots and spacing are
similar to that found on Roman colanders or strainers, but the Llandough fragments are of
thinner sheet metal and appear to bend along one edge in the manner of a binding strip.137
Mounts with ‘serrated’ edges are known,138 but no traces of rivets survive on the Llandough
fragment (Fig. 34, 4).
Lead
1. Lead junction for iron grille (window?). Probably Roman (B331; NMW 95.56H/14;
Fig. 34, 5).
Iron (a complete list of iron objects is deposited in the site archive)
1. Small single-edged knife with angled back. Complete length 103 mm; blade
length 62 mm (B898*; NMW 95.56H/15; Fig. 35, 1).
2. Fragment of large knife blade. Slight angled back. York back-form C, with back
almost straight, curving down towards tip; blade curving up to point. Broken at
tang. Overall length 91 mm (fill 2642 of post-pit 2641 of the post-cemetery timber
building; NMW 95.56H/16; Fig. 35, 2).
3. Narrow, single-edged knife blade, Type C3. Straight back, narrow blade, turning
down slightly at point. Overall length about 112 mm. Whithorn Class C with
convex front part. Most period IV, c. a.d. 1000–50 (B34; NMW 95.56H/17; Fig.
35, 3).
4. Single-edged blade (some of tang missing) from cutting tool. Overall length 185
mm. The poor state of preservation makes interpretation of this object difficult.
There are similarities to the seax from Abbey Green, Chester,139 and larger knife
forms from York, if one assumes that the pointed end has snapped off, but the blade
back is not particularly thick (fill 121 of pit 120: grid 10/3; NMW 95.56H/45; Fig.
35, 4).
5. Oval loop. Width 46 mm; thickness 9 mm (B106; NMW 95.56H/60; not
illustrated).
6. Fragment of iron strap-work. Main section 90 mm long by max. 24 mm wide,
thickening from 2 mm at wide end to 7 mm at narrow end (fill of post-pit 1235 of
the post-cemetery timber building; Fig. 35, 5).
7. ‘Object’/nail? (B110; NMW 95.56H/62; not illustrated).
8. Lump (B60; NMW 95.56H/119; not illustrated).
136 Alcock, op. cit. in note 131, fig. 20, no. 21.
137 E.g. Crummy, op. cit. in note 126, no. 2029.
138 E.g. S. Margeson, Norwich Households: The Medieval and Post-medieval Finds from Norwich Survey Excavations
1971–1978 (East Anglian Archaeol., 58, 1993), fig. 23, no. 282 (c. 1400–1600).
139 Q. Mould, ‘Iron’, fig. 12.1, no. 1 in S. Ward (ed.), Excavations at Chester: Saxon Occupation within the Roman Fortress:
Sites Excavated 1971–1981 (Chester, 1994).
a monastic cemetery at llandough 69
fig. 34
The non-ferrous artefacts. Scale 1:1. Drawing by Tony Daly; © National Museum of Wales.
fig. 35
A selection of the ironwork. Scale 1:2. Drawing by Tony Daly; © National Museum of Wales.
70 neil holbrook and alan thomas
9. Section of bar, length 48 mm, width 22 mm, thickness 8 mm (fill 2617 of post-pit
2616 of the post-cemetery timber building; NMW 95.56H/; not illustrated).
10. Object. Length 47 mm (B69, SF10; NMW 95.56H/46; not illustrated).
11. Two fragments of binding strip (B303; NMW 95.56H/125; not illustrated).
12. Fragment of binding strip or sheet, 15 by 19 mm (B635; NMW 95.56H/94; not
illustrated).
13. Fragment of binding strip. Width 14 mm (B956; NMW 95.56H/107; not
illustrated).
14. Small fragment of iron (B1027; not illustrated).
Apart from nails, knives formed the largest group of iron objects. The knife
blades are difficult to date with any degree of accuracy, but their profiles are known
from Dinas Powys and Whithorn. Similar examples to no. 1 occur at York (Type
A1); Whithorn class A (2 straight parts to back, meeting at angle) where it is
dominant during phases II and III, c. a.d. 730–1000; and Dinas Powys.140 Parallels
for no. 2 have been found at Whithorn 66.8 and Dinas Powys.141 Parallels for no. 4
include Whithorn 66.56 (Class D), c. 730–1250 (most common during Whithorn
phase IV, c. 1000–1250/1300). Mention should also be made of the fact that
medieval ironwork was discovered under the rear wing of Great House Farm
c. 1880.142 It is now in the collections of the National Museums & Galleries of
Wales.
Nails
Nails are present in 28 of the excavated graves. Many are fragmentary, and
there are insufficient large nails from individual graves to have functioned as coffin
fastenings (usually one per grave). An exception is B449 which contained seven
nails, two of which have been bent at right-angles at points 28 and 30 mm below
their head, suggesting a timber thickness, but not necessarily as part of a coffin. A
diversity of original functions for the nails can be assumed, in view of the variation
in size, their fragmentary nature, the bent shafts of many, and the presence in
similar deposits of hob-nails.
In view of the other Roman material recovered during the excavations, a
Roman date seems likely for most if not all of the nails. Many of the large nails
were bent, and a source for their use within the context of the 2nd- to early 4th-
century Roman villa situated some 100 m south of the excavation appears likely.143
Most large nails are incomplete (shafts or heads), the complete examples measuring
68 mm (B110), 44 mm (B175), 74 mm (B449), 63 mm (layer 1255) and 70 mm (fill
2255 of feature 2256).
The nails have on occasions been classified by head-form and length of shaft,
the smaller examples being tacks or segs. Of the small hobnails, their lengths vary
140 P. Ottaway, Anglo-Scandinavian Ironwork from Coppergate (The Archaeology of York, 17/6, London, 1992), fig.
229, no. 2807; A. Nicholson, ‘The iron’, 404–33 in Hill, op. cit. in note 37, p. 427; Alcock, op. cit. in note 131, fig.
21, no. 1.
141 Nicholson, op. cit. in note 140; Alcock, op. cit. in note 131, fig. 21, no. 6.
142 R.C.H.A.M.W., op. cit. in note 58.
143 Owen-John, op. cit. in note 4, 142–4.
a monastic cemetery at llandough 71
Table 13
QUANTIFICATION OF IRON HOBNAILS BY LENGTH AND WIDTH OF HEAD
6– 7– 8– 9– 10– 11– 12– 13– 14– 15– 16–
6.9mm 7.0mm 8.9mm 9.9mm 10.9mm 11.9mm 12.9mm 13.9mm 14.9mm 15.9mm 16.9mm
Length 3 4 4 8 6 3 4 5
Head Width 1 1 9 18 22 8 5 1
from 10–16 mm and their head-widths from 6–13 mm (Tab. 13). Similar hobnails
have been published from Roman Caerleon, where some 600 were found in the
upper sediments of the frigidarium drain of the Fortress Baths. Their domed heads
were typically 10–12 mm across.144 There are too few hobnails from graves (with
the possible exception of B87*) to confirm that they represent intact footwear on
the bodies, and there is no evidence that their occurrence is evidence for continuity
of a late Roman burial rite.145
Samples
?Pigment. Layered (B98*; NMW 95.56H/56; not illustrated). No analysis has been
undertaken.
Similar fragment from B60 (NMW 95.56H/119).
Bone
1. Roughly trimmed tapering shaft and point, probably from an awl or large pin.
Length 34 mm; max. width 6.4 mm. High-medieval or earlier (B69; NMW
95.56H/42; Fig. 36, 1).
2. Knife handle (sawn, split type) with one rivet hole; burnished surfaces. Original
width about 17.2 mm. Date uncertain (unstratified; NMW 95.56H/41; Fig. 36, 2).
Textile-related objects
1. Fragment of irregularly-shaped spindle whorl made from an Oxfordshire bowl
possibly of Young type C45.146 The fabric is orange with a grey core and with
traces of red slip. The vessel-type has a long currency in the period from the mid-
3rd to the late 4th century. Width 24 mm. Possibly Early-medieval (unstratified in
the vicinity of B153; NMW 95.56H/192; Fig. 37, 1).
2. Fragment of spindle whorl cut from sherd of a samian bowl, probably of form
18/31 or 31 and Central Gaulish in manufacture. The original vessel will have
been of 2nd-century date. Diameter 49 mm (007, ?former cultivation soil; NMW
95.56H/191; Fig. 37, 2).
One of the pottery spindle whorls found at Dinas Powys was made from a re-
used sherd of Romano-British greyware.147 Other examples at Dinas Powys were
made from a redware resembling A ware, while two flakes were reported of a
144 Zienkiewicz, op. cit. in note 127, 192–3.
145 D. Petts, ‘Cemeteries and boundaries in western Britain’, 24–46 in S. Lucy and A. Reynolds (eds.), Burial in
Early Medieval England and Wales (Leeds, 2002), at p. 30.
146 C. J. Young, Oxfordshire Roman Pottery (BAR Brit. Ser., 43, Oxford, 1977).
147 Alcock, op. cit. in note 131, fig. 33, no. 1.
72 neil holbrook and alan thomas
fig. 36
The bone artefacts. Scale 1:1. Drawing by Tony
Daly; © National Museum of Wales.
fig. 37
1, 2: ceramic spindle whorls; 3: roundel of mudstone. Scale 1:1. Drawing by Tony Daly; © National Museum of
Wales.
spindle whorl made out of a sherd of samian, as well as one of bone.148 The
Llandough cemetery examples may have a similar origin, and reflect the spinning
of wool in the vicinity of the site. The textile impressions found on the ironwork
from B631* are important evidence for textile used in South Wales during the 5th
or 6th centuries.
Stone
1. Possible honestone (fill 2379 of post-pit 2378 of the post-cemetery timber
building; NMW 95.56H/180; Fig. 38, 1).
2. Honestone, fine-grained, micaceous Devonian sandstone pebble. Heavily worn
on flat side; some longitudinal wear on opposite side. Length 137 mm; max. width
45 mm. Early-medieval? (B127; NMW 95.56H/182; Fig. 38, 2).
3. Disc of irregular shape, made of siltstone. Post-Roman counter? Found on upper
vertebrae just below flat stone on which skull rested (B791; NMW 95.56H/173;
Fig. 37, 3).
148 Ibid., 149, fig. 34, no. 14.
a monastic cemetery at llandough 73
fig. 38
1–4, 6: worked stone; 3: ceramic marble; 5: possible fossil. Scale 1:2. Drawing by Tony Daly; © National Museum of
Wales.
The two definite honestones (nos. 1, 2) are both of sandstone, the most
popular rock-type utilised at Dinas Powys.149 As at this site, hand-sized pebbles
were selected for use at Llandough.
Later Medieval
Copper-alloy
1. Small pin with wound wire spherical head (B75; NMW95.56H/10; Fig. 34, 6).
149 Alcock, op. cit. in note 131, 160.
74 neil holbrook and alan thomas
2. Pin or needle (top missing) (B75; NMW 95.56H/11; Fig. 34, 7).
Pins were used to secure shrouds, though this is not proven in these cases.
Similar examples are reported from Norwich and Beeston Castle.150
Post-Medieval Artefacts
Glass
1. Bottle base. 17th-century (unstratified; NMW 95.56H/27).
2. Glass waste. Opaque white metal. Length 14 mm (2 globules) (B457; NMW 95.56H/
83).
3. Twist of clear cullet. Length 26 mm (B436; NMW 95.56H/33).
4. Thin sliver of plain glass. Clear, no bubbles. Length 12 mm (B638; NMW 95.56H).
5. Fragment of bowl in reddish glass. Length 15 mm (1255, stony layer in the northern part
of Area III which was cut by numerous burials; NMW 95.56H/29).
All other glass recovered from the site is of recent date, the majority being window
glass (listed in site archive).
Stone
1. Fragment of honestone or moulded edging for a sill or doorstep, in Pennant sandstone.
Very straight, rounded edge. Possibly 19th-century (B31; NMW 95.56H/181; Fig. 38, 4).
2. Fragment of fine-grained micaceous sandstone, with two raised, blackened bands on one
face. These may be impressions of the sandstone horizon on top of a coal-bed (B511;
NMW 95.56H/183; Fig. 38, 5).
3. Fragment of rectangular worked block of carboniferous limestone, with a brownish-red
outer surfaces (on all four sides, possibly derived from weathering or water action). One
end is very flat (B532; NMW 95.56H/184; Fig. 38, 6).
Copper-alloy
1. Small spectacle-type buckle (misshapen). Width 22 mm. Probably 16th-century
(unstratified; NMW 95.56H/114).151
2. Thimble (flattened). Post-medieval. 17th- or 18th-century (unstratified; NMW 95.56H/
112). Similar examples found on the wreck of the Kronan, Royal Flagship of the Swedish
fleet (lost 1676).152
Miscellaneous non-ceramic artefacts
Only stratified artefacts are listed.
Flat disc-shaped button with wire loop. Diameter 16 mm. 19th-century (B75; NMW
95.56H/49).
Flat disc-shaped button with cast lug. Diameter 16 mm. 19th-century? (B198; NMW
95.56H/122).
Two pellets — shot (B470; NMW 95.56H/79).
Lead shot (B152; NMW 95.56H/85).
Modern nail (B428; NMW 95.56H/77).
150 Margeson, op. cit. in note 138, no. 31 (c. 1450–1500); ibid., fig. 5, nos. 34–7 (17th-century); P. Courtney, ‘The
medieval and post-medieval objects’, in P. Ellis, Beeston Castle, Cheshire: Excavations by L. Keen and P. Hough, 1968–85
(London, 1993), fig. 101, nos. 44, 46 (17th-century).
151 For similar see Margeson, op. cit. in note 138, fig. 16, nos. 164–5 and 168–9 (17th-century).
152 A. Franzén, ‘Kronan: remnants of a warship’s past’, National Geographic, 175 (1989), 438–65 at p. 457.
a monastic cemetery at llandough 75
Modern nail (B227; NMW 95.56H/123).
Brick
Fragment of hand-made brick. Post-medieval. Rubble 2255 overlaid by B871; NMW
95.54H/205.
iron age and roman pottery
By peter webster
Most, if not all, Iron-age and Roman pottery recovered from the site is from residual
contexts. In this situation, it seems most relevant to restrict published comment to a general
survey of the material. A more detailed listing, consisting of the initial assessment by Jane
Timby and a listing by Museum number and context (by PVW) will be placed in the site
archive. The material will be stored separately from post-Roman pieces in the NMGW,
Cardiff.
The range of material is, in fact, sufficiently restricted to allow us to summarise it
under a combination of source, fabric and form headings (Tab. 14). In the table, some
effort has been made to ensure that different sherds from the same vessel are counted only
once, but, in a collection which consists mainly of sherds, it is often not possible to assign
sherds to specific vessels and some multiple counting is inevitable. Nevertheless, the table
does achieve a reasonable overview of the collection and presents it in a form capable of
comparison to that published from the adjoining Roman site in terms both of its sources
and chronology.153
Sources
Table 14 has been arranged by period. Iron-age/Early Roman fabrics may all be
considered local. The Roman fabrics are arranged in terms of the distance of the source
from the site with the most distant first.
Iron-age/Early Roman fabrics
Fabrics associated with the Iron Age/Roman transition are comparatively well
represented on the site (20% of the total). They consist mainly, or wholly, of jars. Fabrics
have been divided into those with calcite, ‘quartz’, stone (probably limestone) and
unidentified grit filler. Calcite grits are well known from South Wales.154 The stone temper
is presumably that identified by Spencer and further discussed by Allen.155 The quartz-like
temper (from B178, B180, B190*, B195, B552 and the fill 806 of pit 805; Museum nos.
257–9 and 261) may occur with stone tempering and is in small roughly rectangular
blocks. A number of the sherds in this ware are in the Glastonbury tradition. A sherd in
this tradition has previously been noted from Llandough but the quantity from the 1994
site is more comparable with that from Biglis.156 The Iron-age/ Early-Roman material as
a whole would suit a date range within the early/mid- to mid-/late 1st century (i.e. either
side of the Roman conquest of the region). However, both on the 1979 site and on the 1994
one, 1st-century romanised pottery is conspicuous by its absence and a pre-Neronian
153 Owen-John, op. cit in note 4, 160–72.
154 Cf. Whitton: M. Jarrett and S. Wrathmell, Whitton: An Iron Age and Roman Farmstead in South Glamorgan (Cardiff,
1981), 116; Usk: W. H. Manning, Report on the Excavations at Usk 1965–1976: The Roman Pottery (Cardiff, 1993), 230
and refs.
155 B. Spencer, ‘Limestone tempered pottery from South Wales in the late Iron Age and early Roman period’,
Bull. Board Celtic Studs., 30 (1982–3), 405–19; J. R. L. Allen, ‘Late Iron Age and earliest Roman calcite-tempered
ware from sites on the Severn Estuary levels: character and distribution’, Stud. Celtica, 32 (1988), 27–42.
156 Owen-John, op. cit. in note 4, 162; Parkhouse, op. cit. in note 24, 33.
76 neil holbrook and alan thomas
Table 14
ROMAN POTTERY BY CLASS AND SOURCE
Flagon Beaker Jar Bowl Mortarium Dish Unclassified Total %
IA / E-Roman
Calcite Grit – – 7 – – – – 7 5
‘Quartz’ – – 8 – – – – 8 6
Stone – – 7 – – – – 7 4
Gritted – – 6 – – – – 6 4
Roman
Samian – – – 3 – – 2 5 4
à l’éponge – – 1 – – – – 1 1
S. Midland – – 3 – – – 1 4 3
Oxford White – – – – 4 – – 4 3
Oxford Red 1 1 – 8 3 1 – 14 10
BB1 – – 33 9 – 3 1 46 33
Severn Valley – – 1 – – – – 1 1
S. Wales Grey – – 10 1 – – – 11 8
Other oxidised 3 – 13 1 1 – 1 19 14
Other reduced – – 6 – – – 1 7 5
Total 4 1 95 22 8 4 6 140 100
% 3 1 68 16 6 3 4 100
conquest date may be suggested for this part of the assemblage. We may also note in
passing that Glastonbury-style pottery does not appear on Roman sites in South Wales
that have no Iron-age predecessors and it probably, therefore, went out of use before the
conquest. One group is worthy of note, that from the fill 806 of pit 805 which was cut by
B217 (Fig. 5). This pit yielded six fragments of a large Late Iron-age/Early-Roman jar
with quartz filler; a sherd of a large Roman grey ware jar; a sherd from a jar with quartz
and stone temper; seven sherds of Glastonbury Ware with quartz and grey stone filler
(some with cross-hatching between lines and also some deeper grooves, perhaps from the
same vessel as no. 7), and a sherd of black-burnished ware cooking pot, probably with
lattice below a line and therefore of late 3rd- or 4th-century date. Excluding the last as
intrusive, it is likely that this pit dates to the 1st century a.d.
Samian
The few sherds of samian ware found could all be of 2nd-century date. The most
remarkable aspect of the samian is, however, its sparsity. It is highly unlikely that any
substantial occupation of the area excavated could have occurred in the period c. a.d.
70–200 without deposition of more sherds of samian than this.
Céramique à l’éponge
This late fineware imported from an area north of the lower Loire in western Gaul is
extremely rare in South Wales.157 In view of the small size of the assemblage, its presence
here is the more remarkable and probably serves to underline the intensity of the nearby
late Roman occupation.
157 P. Tyers, Roman Pottery in Britain (London, 1996), 145.
a monastic cemetery at llandough 77
South Midlands shell-gritted ware
Only four sherds were present (layer 1255, B545, B624 and B625; Museum nos. 202 and
289). They are, however, significant as being the latest Roman sherds discovered on the
site. The fabric is found sparsely in South Wales, but always in contexts later than the mid-
4th century.
Oxford white mortaria
Oxford mortaria are typical of later sites in the region. The form represented here, where
identifiable, is Young M22 (Fig. 39, 1 and 9) the archetypal later Roman Oxford white
mortarium produced probably throughout the period c. a.d. 240–400.158
Oxford red colour-coated ware
This represents the only fineware present in any quantity in the assemblage (a surprising
10% of the total). The ware is typical of the later 3rd and 4th centuries.
Black-burnished ware (BB1)
This is the most prolific of the fabrics found (33% of the total). Of this, the majority were
jars. The overwhelming majority of the diagnostic pieces recovered belonged to the late
3rd and 4th centuries.
Severn Valley ware
The ware is not well represented in South Wales outside the conquest period, so the
presence of a single vessel at Llandough is of some interest (B605, Museum no. 295).
South Wales grey ware
This is normally the predominant local fabric on South Welsh sites. One would expect it to
have been better represented among the 1994 material than is the case. The single
illustrated example (Fig. 39, 6) reflects the relatively low percentage of this fabric present.
Other oxidised and reduced wares
This category covers those fabrics not related to a specific source. Most will probably be
local in origin. Of note, however, is Figure 39, 4 which appears to be an import of a fabric
not previously noted in the area.
In general, the sparsity of finewares other than Oxfordshire red colour coated fabric
may be noted. This could reflect the low level of Roman activity prior to the late 3rd
century.
Chronology
The date of the 1994 material is broadly similar to that excavated in 1979. There are,
however, some differences. As with the 1979 collection, that from 1994 contains a
significant amount of pottery from the late Iron Age or early Roman Period (20% of the
total). As already stated, most of this could be either pre- or post-conquest, but it is
noticeable that there is a marked absence of Roman-style pottery of certainly 1st-century
date. This reinforces the suggestion made in the earlier report that a gap may exist between
the Iron-age and Roman occupation of the site.159
158 Young, op. cit in note 146.
159 Webster in Owen-John, op. cit. in note 4, 162.
78 neil holbrook and alan thomas
fig. 39
Roman pottery. Scale 1:4. Drawing by Tony Daly; © National Museum of Wales.
The majority of the remainder of the pottery probably belongs to the later part of the
Roman period. There are a few pieces which could belong to the 2nd or early 3rd century.
However, of the BB1 (33% of the total assemblage), 20 vessels (14% of the whole) are
diagnostically late 3rd-century or later. Little of the remaining BB1 is certainly earlier than
this. All the Oxfordshire products are likely to date from the mid-3rd century onwards. In
this the assemblage closely resembles that from 1979. The more extensive 1979 collection
suggested a cessation of activity in the ‘first third of the 4th century’. The 1994 collection,
though smaller, has a few pieces which are likely to post-date this. Almost certainly later
than c. a.d. 350 are the few fragments of shell-gritted fabric derived from the South
Midlands (from contexts 1250, B624, B625 and B545; Museum nos. 203 and 289; cf. Fig.
39, 2) which, in South Wales, are typical of activity in the second half of the 4th century.
The single sherd of the fabric known as céramique à l’éponge might also be of this late date
(no. 15 below). It is difficult to be certain quite where in the late Roman Period the various
late BB1 jars fit. However, although many have the flaring rims and obtuse angled lattice
(often below a horizontal line) typical of the period from the late 3rd century onwards,
there are no certain examples of vessels with rims oversailing slim bodies,160 or of the more
extremely flared rims which seem to be the latest in the BB1 jar series (Fig. 39, 3). It is,
therefore, safe to assume that most of the pottery reflects the dating already proposed for
the nearby 1979 site (i.e. late Iron-age and Roman, particularly from the late 3rd to the
early 4th century), but that the 1994 site does have some evidence for some activity in the
mid- and late 4th century. There is, however, no clear indication that this latest Roman
activity was of any intensity.
Function
It would be unwise to draw too many conclusions from such a small assemblage as
this. It may be noted, however, that the division by form in Table 14 highlights the
160 J. P. Gillam, ‘Coarse fumed ware in North Britain and beyond’, Glasgow Archaeol. J., 4 (1976), 57–80, no. 14.
a monastic cemetery at llandough 79
extremely high proportion of jars present. These make up 68% of the entire assemblage.
Even if we remove the Iron-age material (all jars) we still find that there are 67 jars out of a
total of 112 Roman vessels, 58% of the Roman total. Parallel to this is the low proportion
of forms, such as flagons and beakers, which could be associated with the table (a mere
4%). This might normally be taken to indicate a subsistence level of existence, but the 1979
assemblage is against this. It might, however, indicate that the vessels being thrown out
into the area examined in 1994 were derived from kitchen activities and serve as a pointer
towards the location of the service range nearby.
Catalogue
A small number of vessels have been chosen for illustration (Fig. 39). These are
intended as an indication of the range of material present. Much of the Roman material
comes from later contexts, so these vessels show the chronological range rather than
serving to date specific contexts. Vessels are listed in order of their Museum numbers.
1. Mortarium in cream fabric with a greyish core and some greying on the rim and with
grey trituration grits. An Oxfordshire product: cf. Young M22 (c. a.d. 240–400).161 007,
possibly from former cultivation soil; NMW no. 195.
2. Rim-fragment from a jar in South Midlands shell-tempered fabric.162 Although not one
of the most developed typologically, a mid- to late 4th-century date may be suggested.
1255, stony layer in the northern part of Area III which was cut by numerous burials;
NMW no. 202.
3. Jar in Black-burnished ware; cf. Gillam nos. 11–12 (late 3rd to early 4th century).163
B15; NMW no. 219.
4. Jar rim in creamy fabric, shading to grey. The filler includes quartz. There are some
similarities to Portchester D Ware,164 and also some fabrics from the Rhineland.165 A Late-
Roman date for the piece seems most likely. B46; NMW no. 232.
5. Jar in Black-burnished ware; cf. Gillam nos. 11–12 (late 3rd to early 4th century). B111;
NMW no. 246.
6. Flanged bowl in grey fabric, brown near the surface and with a dark grey/black surface.
This is South Wales Grey ware, probably altered by burning. There was a similar form
from Usk (fig. 116, type 62.5) dating to the 2nd to 3rd century.166 B122; NMW no. 251.
7. Shoulder-and-neck fragment from a jar or necked bowl in dark grey/black fabric with
an angular ‘quartz’ filler. A vessel in the Glastonbury tradition seems possible,167 and this
is supported by other fragments of Glastonbury-style pottery from the site (from B552 and
pit fill 806), one of which also has the grooves of this piece. A small fragment of rim from
B195 (NMW no. 261) could be from this vessel and suggests a curving upright rim. B180;
NMW no. 258.
8. Dish in orange fabric with a grey core and faint traces of red slip. An Oxfordshire
product: cf. Young C48.5 (c. a.d. 270–400). B272; Museum no. 269.
9. Mortarium in cream fabric with a greyish core and with a cream surface shading to
grey. An Oxfordshire product: cf. Young M22.2 (c. a.d. 240–400). B306*; NMW no. 271.
10. Bowl in orange fabric with traces of orange slip. An Oxfordshire product: cf. Young
C81–2 (4th century). B320; NMW no. 272.
161 Young, op. cit. in note 146. All further references to Young in the catalogue refer to this publication.
162 For a brief discussion of this ware in a South Welsh context see Webster in Manning, op. cit. in note 154,
294–5. For a complete jar with a similar rim to this, see ibid., fig. 136.1.
163 Gillam, op. cit. in note 160. All further references to Gillam in the catalogue refer to this publication.
164 Cf. Tyers, op. cit. in note 157, 194–5.
165 M. Redknap, pers. comm.
166 Webster, op. cit. in note 162.
167 Cf. B. Cunliffe, Iron Age Communities in Britain (3rd ed., London, 1991), fig. A:21.
80 neil holbrook and alan thomas
11. Mortarium rim in orange fabric with a grey core and orange/grey slip. An Oxfordshire
product: cf. Young C97.3 (c. a.d. 240–400). B523; NMW no. 287.
12. Beaded dish in Black-burnished ware. The angle of the wall suggests a vessel such as
Gillam no. 72 (early 3rd century). B979; NMW no. 310.
13. Flanged and ridged bowl in Black-burnished ware; cf. Gillam nos. 46–7 (late 3rd to
early 4th century). Unstratified; NMW no. 313.
14. Bead-rim jar in dark grey fabric with a white gritty filler.168 The overall date is late
Iron-age to Early Roman, but the absence of other Early-Roman fabrics suggests an Iron-
age date in this instance. Unstratified; NMW no. 314.
15. (Not illustrated). A jar or bowl fragment in céramique à l’éponge, a smooth orange-buff
fabric with red sponged slip. The ware, found mainly in Brittany, Normandy and Anjou, is
rare in Britain and exceptionally rare in South Wales.169 4th century. Unstratified; NMW
no. 314.
Roman Brick and Tile
Seven fragments of certain and probable Roman tile were recovered. Stratified
examples of certain attribution comprised a fragment of imbrex from B41/51, and probable
attributions of tegula from B116 and imbrex from B333*.
the early-medieval pottery
By david williams and mark redknap
Nine sherds of post-Roman Bii amphorae were found, five of which were
examined by David Williams.
1–2. Two small plain body sherds in a sandy light buff fabric with scattered pieces of white
limestone (B542; NMW 95.56H/288; not illustrated).
3. A small closely rilled body sherd in a sandy light buff fabric with some limestone
showing. Probably from the top section of the vessel (unstratified, Area III; not illustrated).
4. A very similar looking sherd to no. 3 (B864; NMW 95.56H/305; Fig. 40, 3).
5. A slightly larger rilled body sherd than the others in this group, with the rilling more
widely spaced. This piece probably came from the central body area of the vessel. Light
reddish-buff sandy fabric with many small pieces of white limestone (B116; NMW 95.56H/
247; Fig. 40, 1).
6–7. Two body sherds of amphora, one with wide horizontal rilling and one with narrow
horizontal rilling (B819; NMW 95.56H; not illustrated).
8. Small, close rilled body sherd in sandy, light buff fabric with some limestone showing
through (B517; NMW 95.56H/285; Fig. 40, 2).
9. Small, close-rilled body sherd in sandy, light buff fabric (1255, stony layer in the
northern part of Area III which was cut by numerous burials; NMW 95.56H/202; Fig. 40,
4).
All of the above sherds belong to the Eastern Mediterranean Bii amphora
form.170 This type was made mainly along the southern coast of Asia Minor,
especially around the Golfe D’Alexandrette (Cilicia, SE. Turkey has been
168 For a discussion of the general type see Webster, op. cit. in note 162, 230.
169 Cf. Tyers, op. cit. in note 157, 144–5, especially fig. 161. British finds are listed by M. Fulford, ‘Pottery and
Britain’s long distance trade in the later Roman period’, 35–84 in D. P. S. Peacock (ed.), Pottery and Early Commerce
(London, 1977) at p. 77.
170 D. P. S. Peacock and D. F. Williams, Amphorae and the Roman Economy (London, 1986), Class 44.
a monastic cemetery at llandough 81
fig. 40
B-ware amphorae sherds. Scale 1:2. Drawing by Tony Daly; © National Museum of Wales.
suggested)171 and is generally considered to have carried olive oil, although this
awaits firm proof and wine has also been suggested.172 These vessels seem to have
reached Britain only in the immediately post-Roman period,173 although they were
in use some time before; early versions of the form are present at Kourion in
Cyprus, for example, which was destroyed by an earthquake during the mid-360s
a.d.174 In fact, their occurrence in Britain may possibly have been restricted to the
period from the late 5th century to the mid-6th century, if importation was allied
to Phocaean Red Slipped ware.175 In South Wales, B-ware amphorae occur at
Dinas Powys176 and Longbury Bank, while in North Wales B ware occurs at Dinas
Emrys. Bii has not yet been found in Roman contexts, and cannot therefore be
explained as an element of the Romano-British scatter on the site.
The combing, wall thicknesses and fabrics of sherds 6 and 7 suggest that one
belongs to the same vessel represented by sherds 4, 8 and 9. This means that burials
B517, B864 and context 1255 contain cross-linked pottery. The second sherd has a
similar fabric to that of nos. 1 and 2 (B542).
later medieval pottery
By mark redknap
As is the case with most of the Roman and earlier pottery from the excavations,
most of the later medieval pottery would appear to be residual in nature. The
material is stored within the medieval collections at the National Museum &
171 J. Y. Empereur and M. Picon, ‘Les regions de production d’amphores imperiales en Méditerranée orientale’,
223–48 in Amphores romaines et histoire économique: dix ans de recherche (Coll. de L’École Française de Rome, 114, 1989).
172 Campbell 1991, op. cit. in note 134, 60.
173 A. C. Thomas, A Provisional List of Imported Pottery in Post-Roman Western Britain and Ireland (Redruth, 1981).
174 D. F. Williams, Roman Pottery from Kourion, Cyprus: An Interim Report (forthcoming).
175 M. G. Fulford, ‘Byzantium and Britain: a Mediterranean perspective on post-Roman Mediterranean imports
in western Britain and Ireland’, Medieval Archaeol., 33 (1989), 1–5.
176 In identical fabric:. Alcock, op. cit. in note 131, fig. 27, no. 3.
82 neil holbrook and alan thomas
Gallery in Cardiff. As with the Roman and earlier material, sherds from the same
vessel have only been counted once, but the same qualifications to the results apply
(Tabs. 15–16). The overall pattern is useful, in terms of allowing comparison with
contemporary assemblages from the vicinity, including Cosmeston, some 3 miles
(4.8 km) from Llandough.177
Table 15
THE SOURCES AND PROPORTIONS OF THE LATER-MEDIEVAL SHERDS
Vessel type
Fabric Glazed Jug Cooking Pot/Jar Inturned bowl Total/%
Vale ware (VF1) 3 74 1 78/52.7%
Vale ware (VF2) – 22 – 22/14.9%
Bristol/Redcliffe ware 1 – – 1/0.7%
Llandough fabric F – 42 – 42/28.4%
Limestone-tempered ware – 4 – 4/2.7%
Uncertain 1 – – 1/0.7%
Total 5 142 1 148/100%
The Fabrics
Vale Ware
Light brown to brown sandy iron-rich fabric, containing moderate to abundant
sub-angular to sub-rounded quartz, and sparse red iron ore, sub-rounded black
inclusions. VF1 is lighter in colour, harder and finer in texture; VF2 tends to be
chocolate brown in colour, softer, and more friable.
Sherd total: (VF1) 78, (VF2) 22.
Llandough fabric F
Light to dark brown fabric, moderately hard, with reduced grey core, and
sometimes narrow brown margins. Inclusions of abundant ill-sorted coarse to fine
(<3 mm) sub-rounded and angular fossiliferous limestone, sparse fine black
inclusions (0.5 mm), sparse opaque grey to white sub-angular quartz (1.5–0.3 mm),
and iron-rich clay pellets (fine, 0.3 mm).
Sherd total: 42.
Bristol/Redcliffe wares
Hard fabric, reduced to light grey core with oxidised light yellowish-brown to
pale brown inner surface and margins. Inclusions of subangular and rounded
quartz (up to 1 mm), occasionally very coarse chert/flint (up to 8 mm). Patchy
green external glaze, at times thick. Wheel-thrown.
Minimum number of vessels: 1.
Sherd total: 1.
177 P. Andrews, ‘Excavations at Cosmeston medieval village, near Penarth, South Glamorgan, 1993’, Archaeol.
Wales, 36 (1996), 11–35.
a monastic cemetery at llandough 83
Table 16
LATER-MEDIEVAL POTTERY BY CONTEXT. THOSE CONTEXTS MARKED *
RELATE TO THE POST-CEMETERY TIMBER BUILDING
Context Vale F1 Vale F2 Llandough Limestone- Bristol/ Uncertain
fabric F tempered ware RedcliVe
Fill of pit 004 2
Empty Grave 64 1
Fill of pit 190 1
*Fill of 1235 59 30
*Fill of 1256 11
Rubble 1422 = 1
1255
*Fill of 2378 1
*Fill of 2388 1
*Fill of 2589 1
*Fill of 2611 4
*Fill of 2641 5 1
*Fill of 2647 1 2
Fill 2652 2 3 2
B10, 11, 19 1
B43 1
B52 3 1
B135 1 (glazed)
B601 2 (glazed)
B764 1 (glazed jug?)
B1022 5 5
Totals 78 22 42 4 1 1
% 52.7 14.9 28.4 2.7 0.7 0.7
Limestone-tempered ware, source uncertain
Moderately hard fabric with dark grey core and brown to pinkish-buff surfaces.
Coarse, heavily tempered fabric with abundant ill-sorted coarse–fine (<3 mm)
limestone inclusions, reddish iron-rich inclusions, and white quartz inclusions.
Sherds represent jars with fairly thick walls. All the sherds are undiagnostic and
very small, so dating is uncertain.
Sherd total: 4.
A small number of sherds have been chosen for illustration, to indicate the
rim forms represented on the site.
1. Cooking pot rimsherd. Vale ware (Llandough fabric ai = VF1).178 Fill of post-
pit 1235 of the post-cemetery timber building; NMW 95.56H/201 (Fig. 41, 1).
178 For similar form see B. E. Vyner, ‘Vale Fabric—a medieval pottery industry in Glamorgan’, Medieval and Later
Pottery in Wales, 5 (1982), 31–9, fig. 1, 21.
84 neil holbrook and alan thomas
fig. 41
Medieval cooking-pot rim sherds. Scale 1:4. Drawing by Tony Daly; © National Museum of Wales.
2. Cooking pot rim fragment. Llandough fabric F.179 Fill of post-pit 1235 of the
post-cemetery timber building; NMW 95.56H/201 (Fig. 41, 2).
3. Cooking pot rim fragment. Vale ware (Llandough fabric ai = VF1). Fill 2642 of
post-pit 2641 of the post-cemetery timber building; NMW 95.56H/211 (Fig. 41,
3).
4. Cooking pot rim fragment. Llandough fabric F.180 Fill 2379 of post-pit 2378 of
the post-cemetery timber building; NMW 95.56H/206 (Fig. 41, 4).
Tile
1. Fragment of plain ridge tile, unglazed. Vale ware, similar to Llandaff wasters. 13th or
14th-century. Fill 190 of pit 189; NMW 95.56H/198.
2. Fragment of roof tile. B10/B11/B19; NMW 95.56H/218.
3. Fragment of glazed ridge tile. 13th or 14th-century. B817; NMW 95.56H/304.
4. Two fragments of ridge tile (similar to Llandaff wasters). Vale ware. GGAT evaluation
test pit 5; NMW 95.56H/315.
Discussion
Seven burials produced sherds of 13th-/14th-century pottery (Tab. 16) and
two graves produced fragments of 13th-/14th-century ridge tile (B10/B11/B19
and B817). Post-medieval pottery was recovered from the fills of four graves (B26/
27, B50, B296, B631*). On balance all of this material is considered to be intrusive
within the grave fills (see above, p. 32). The most productive context for later
medieval pottery was post-pit 1235, with cooking-pot rim-forms in Llandough
fabric F and Vale ware. All the sherds are fairly small, and from cooking pots/jars.
As at Llandough, Vale ware dominates assemblages from Barry Village and
Wrinstone, and occurs in reasonable quantities at Llantwit Major, Coity Castle
and Kenfig Castle. It occurs in smaller quantities at Margam, Swansea and
179 For a similar form from Cosmeston see Andrews, op. cit. in note 177, fig. 5, 12.
180 B. E. Vyner, ‘Medieval and post-medieval pottery from Llandough, Glamorgan’, Medieval and Later Pottery in
Wales, 4 (1981), 12–25 at p. 19.
a monastic cemetery at llandough 85
Loughor, occasionally in Cardiff and a few sherds from the later phases of Rumney
Castle. The concentration in South Glamorgan has suggested a source in the Vale,
and the recent discovery of coarseware, glazed jug and ridge-tile wasters (both
Vyner fabrics VF1 and VF2) at Llandaff indicates one source for at least some of
this sandy ware during the 13th–14th centuries.181 Most of the sherds found at
Llandough are cooking pots.
Llandough fabric F was also poorly represented amongst the later medieval
pottery assemblage recovered during the 1980s excavations, which was dominated
by Vale ware.182 The pottery from the Llandough cemetery reflects this supply to
the area, and presumably reflects activity in and around the church during the high
medieval period. Sherds of 13th- or 14th-century pottery were recovered by E. J.
Beare in 1962–3 on the slope immediately to the north-east of Great House Farm
(to the north of the church).183
post-medieval pottery
By mark redknap
1. Horizontal rim sherd, glazed slipware. Orange fabric, brown external glaze. Yellow slip
pellets between concentric lines on rim top. Chamber pot? 18th-century. 007, possibly
from former cultivation soil; NMW 95.56H/195.
2. One body sherd, glazed redware. 18th-/19th-century. B26/27; NMW 95.56H/223.
3. One body sherd of fine redware with internal green glaze. 16th- or 17th-century. B50;
NMW 95.56H/234.
4. One sherd of white china with printed blue external design, 19th-century; one sherd
glazed redware wall sherd, 19th-century; one redware wall sherd with white slip on interior
surface, glazed. 18th-century. B296; NMW 95.56H/270.
5. One rim sherd from plate, transfer-printed. 19th-century. B631*; NMW 95.56H/296.
6. Marble in light brown fired clay. Diameter 17 mm. Post-medieval. Unstratified; NMW
95.56H/179; Fig. 38, 3.
7. Base of a stoneware jar, c. 1900; eight sherds of glazed and unglazed post-medieval
redware, 18th-/19th-century. GGAT evaluation test pit 5; NMW 95.56H/315.
Unstratified wares have not been not listed.
The presence of some post-medieval artefacts within grave fills reflects the disturbed
nature of some of these burials and the impact of the scrambling of these and later deposits.
Most of these artefacts probably derive from activity associated with the area around Great
House Farm, which used to stand about 30 m north of St Dochdwy’s church. The
farmhouse may have incorporated medieval fabric within parts of the 17th-century house
associated with the Vaughan family. The 19th-century work on the farmhouse is associated
with the Bute Estate.184
181 T. Young, ‘Llandaff Cathedral School (ST 156 779)’, Archaeol. Wales, 41 (2001), 140–2.
182 Vyner, op. cit. in note 180.
183 GGAT, op. cit. in note 1, 4; Beare, op. cit. in note 3.
184 GGAT, op. cit. in note 1, 4; R.C.A.H.M.W, op. cit. in note 58.
86 neil holbrook and alan thomas
GENERAL DISCUSSION OF THE CEMETERY
By NEIL HOLBROOK
an iron-age burial?
Excluding a background scatter of seven flints, to which little significance is
attached, the only evidence for possible prehistoric activity at the site is the recovery
of 28 sherds of late Iron-age/Early-Roman pottery, 16 of which came from the fill
806 of pit 805 (Fig. 5). On balance Webster, above, prefers to date this material to
the pre-Conquest period. The scattered distribution of this pottery is of little
evidential value as in most cases it is residual in later graves. There is only one
possible example of a late Iron-age grave, the pit burial B601* (Figs. 10, 14C). Pit-
burials have been found on a variety of Iron-age sites central-southern Britain,
although the deposition of isolated skeletons in pits was clearly an atypical rite.185
Hill has shown in Wessex that these burials were part of a ritual practice involving
the deposition of a variety of artefacts and materials.186 If this burial is of late Iron-
age date, it may be associated with other burials, features or monuments which
have since been destroyed (it lay on the very edge of an area that has been
truncated by later activities). Iron-age settlement was discovered beneath the
Roman villa 150 m to the S.187
continuity of burial from the late-roman to the early-medieval
period?
Before all the radiocarbon dates became available the authors believed that
there were some burials at Llandough (in particular B87* and 110) to which a
Late-Roman date could be ascribed with reasonable confidence.188 Radiocarbon
assay has, however, confirmed Rahtz’s strictures that Roman artefacts only provide
a terminus post quem,189 and there are now no burials for which a Roman date is
unequivocally claimed. That is not to say that the cemetery does not have a Roman
origin; only that it cannot be demonstrated given the current precision of
radiocarbon dating. Whether the Roman artefacts within the grave fills were just
introduced incidentally into the backfill, or were the product of more structured
deposition (sweepings from funeral feasts, etc.) is open to question (see above).
There is a noticeable cluster in the distribution of Roman pottery and other
artefacts within the graves in the very southernmost part of the excavation area
(Fig. 14C). The presence of such material clearly reflects Roman activity in the
vicinity, although it is noticeable that some of the coins and pottery recovered from
the grave fills are later in date than that found in the excavation of the villa in 1979.
Indeed Webster dated the abandonment of the house to the first third of the 4th
century. It is conceivable that there was another Roman building beneath the
churchyard that was occupied for longer than the villa house, and if a Roman
185 R. Whimster, Burial Practices in Iron Age Britain (BAR British Series, 90, Oxford, 1981), 191.
186 J. D. Hill, Ritual and Rubbish in the Iron Age of Wessex (BAR Brit. Ser., 242, Oxford, 1995).
187 Owen-John, op. cit. in note 4.
188 Cf., for instance, Current Archaeol., 146 (1996), 73–7.
189 P. A. Rahtz, ‘Late Roman cemeteries and beyond’, 53–64 in R. Reece (ed.), Burial in the Roman World (CBA
Res. Rep., 22, London, 1977), at p. 55.
a monastic cemetery at llandough 87
origin for the cemetery is sought, this might have acted as a focus for burial. It must
be stated that this would be a very unusual state of affairs, as while scattered burials
have been found on a wide range of Romano-British rural sites, rural cemeteries
are extremely rare as discussed by Jeremy Knight in the supplementary essay that
follows this report.
There, Knight outlines various models which might explain the link between
the Roman villa and the Early-medieval monastery. The monastery at Llandough
is first recorded historically in the mid-7th century. Although the written sources
cannot establish the matter beyond doubt, they suggest that advent of enclosed
monastic communities in South Wales occurred from the late 5th century, and to
have been well established by the early 6th century.190 As yet radiocarbon dating is
insufficiently precise to ascertain whether there was continuity of burial from the
Late-Roman Period, or whether the cemetery was established de novo around this
date. If burial was continuous, it begs the question of how Llandough should be
interpreted in the period before the monastery. Was there a settlement here that
was post-villa and pre-monastery, or did the former villa lands pass into the control
of a local leader, perhaps the one who lived at the defended hilltop enclosure of
Dinas Powys, only 2 km distant? A link between the two sites is certainly suggested
later by the occurrence at both of sherds of Bii amphora from Asia Minor, for
which the main period of importation into Britain appears to fall in the 75 years
between a.d. 475–550,191 and of late 6th- or 7th-century glass cone beakers. Bii
amphorae are found on comparatively few sites, and in South Wales by far the
most significant is Dinas Powys which has produced not only amphorae but also
imported fine wares in quantity. Exactly how this site should be interpreted is a
matter for debate, although the evidence for high-quality metalworking and the
range of exotic imports clearly marks its high status. The mechanisms by which
sites were selected for monastic foundations are unclear, but one possibility at
Llandough is that lands were granted to it by a local overlord (perhaps even the
one at Dinas Powys). The monastery could therefore have been implanted on to a
site with a long established and continuing tradition of burial (be it Christian or
not) in a similar manner to the chapels which were added to pre-existing rural
graveyards from the late 6th or 7th century onwards (Knight’s Models 3–5).192
The alternative, of course, is that the cemetery dates entirely to the years following
the foundation of the monastery (Knight’s Models 6–8).
Cemetery sites of the immediately post-Roman Period have been discussed by
Rahtz, and some show reasonable evidence of continuity of burial from late into
sub-Roman times (Cannington is the largest with an estimated original population
of 1,500–5,500 burials).193 Like Llandough, the majority of burials from such sites
were oriented E.–W. and had few grave goods. Locally the small enclosed cemetery
at the Atlantic Trading Estate, Barry seems to span the period from the 2nd–10th
190 Thomas, op. cit. in note 27, 21–2; Davies, op. cit. in note 39, 147–8.
191 Fulford, op. cit. in note 175.
192 Generally Thomas, op. cit. in note 27, 67–90, and excavated Welsh examples at Capel Maelog (Britnell, op.
cit. in note 20, 82–5) and Pennant Melangell (Britnell, op. cit. in note 49, 91–2).
193 Rahtz, op. cit. in note 189, 56; Rahtz et al., op. cit in note 15, 420 and 422–3.
88 neil holbrook and alan thomas
centuries.194 The burial arrangements at Llandough display greater affinity to
these types of site than to the early Christian rural enclosed cemeteries which occur
along much of the Atlantic coast of Britain.195 In Wales such cemeteries have
traditionally been recognised by the practice of burial in long-cists, although it is
now clear that more diverse burial-types also occur, none of which have been
found at Llandough.196
overall chronology of the cemetery
The earliest radiocarbon date from the excavated burials only shows that
burial had commenced by the period 370–640 cal a.d. It is possible to refine this
dating, however, due to the presence within five grave fills in Areas I and III of
sherds of imported Bii amphorae (from B116, B517, B542, B819 and B864). While
this pottery only provides a terminus post quem for the individual graves, this material
does nevertheless provide archaeological evidence of late 5th- or 6th-century
activity of some nature at Llandough. The fill of one grave (B640) also produced
fragments of a late 6th- or 7th-century glass cone beaker. Turning to the date when
the cemetery went out of use, the ranges of the latest radiocarbon dates are
885–1035, 779–1024 and 782–1024 cal a.d. respectively. Burial 34 in Area I was
accompanied by an iron knife for which a 11th-century date might be appropriate.
This evidence strongly suggests that burial continued at Llandough until the
demise of the monastery which historical sources suggest occurred in the late 10th
or early 11th century. The rarity of sherds of medieval pottery from the grave fills,
in contrast to the post-pits of a timber building which overlay part of the cemetery,
supports the radiocarbon dates in suggesting that regular burial had ceased within
the excavated area by the 13th century when such pottery became widespread.
morphological and chronological development of the cemetery
No definite boundaries to the graveyard were found during the excavation to
assist in the reconstruction of the cemetery layout and organisation, or in the
formulation of a model of its development. During the preliminary phases of post-
excavation analysis, and before all the radiocarbon dates were available, the
cemetery was divided into three areas. It was considered that a noticeable group of
graves on a distinct NE.–SW. orientation could have been aligned on a putative
curvilinear boundary to the cemetery which has left no archaeological trace (an
attempt to mark the approximate position of the boundary is shown on Fig. 6,
although precision is impossible). The burials contained within the area between
this line and the present churchyard wall were designated as Area I. The burials to
the west of Area I were termed Area II where more clearly defined rows were
apparent, although later burials on a marked NE.–SW. alignment cut across this
arrangement. The full extent of these burials lies outside of the excavation area to
the west and ground levelling has probably removed many burials in the northern
part of the site. A shallow ditch 1214 on an almost N.–S. alignment to the east of
194 James, op. cit. in note 24, 96–8.
195 Thomas, op. cit. in note 27, 50–68.
196 James, op. cit. in note 24; Murphy, op. cit. in note 31, 30.
a monastic cemetery at llandough 89
these burials is probably a later feature as it cut through B37*, removing its legs.
Burials to the north of the putative boundary were termed Area III where burial
extended right up to the edge of the steep scarp. It is difficult to make much of the
morphological arrangements within Area III, save to note the frequent intercutting
and the fact that no evidence of any other boundary was found in the excavated
areas. This was the most intensively utilised part of the cemetery, with burials
generally on an E.–W. alignment. One fairly clear row (B730/731, 567, 566, 592,
614, 599 and 893) could be recognised, but little other obvious patterning. Area III
contained numerous infant or juvenile burials which represent the final use of this
part of the cemetery. By way of comparison it may be noted that 56 children were
found buried by the wall of a chapel in the Period II (8th- to 9th-century) monastery
at Whithorn. These clearly formed a distinct group which had been deliberately
separated from the adult population, and pathological examination demonstrated
that they had a variety of afflictions.197Another possibility might be that the
children were the victims of a specific outbreak of plague or endemic.198
Implicit in the zoning outlined above was a simple model which considered
that burial expanded outwards in a broadly radial fashion from a single focus (the
monastery church which presumably lies beneath the present churchyard). In such
a model some of the burials closest to the churchyard wall would be early in the
sequence, although later burials were evidently dug into the previously utilised
area, so hindering attempts to recognise spatial patterning. It was thought that the
putative boundary between Areas I and III might have dated to relatively early in
the history of the monastery, it subsequently being removed as the cemetery
expanded to the north, perhaps with later graves cut across its former line (the
expansion of monastic boundaries can be paralleled in Period 1 at Whithorn).199
In this hypothesis the burials in Areas II and III should constitute later expansion
to the early core.
Once all the radiocarbon dates were available, it became clear that it was
difficult to uphold such a simple model for the development of the cemetery. Three
bodies were very likely buried in the period of the 5th–7th centuries, and certainly
before a.d. 700. Burial 2 was aligned NE.–SW. and lay within Area I, while the
other two lay within Area III. Burial 566 was part of the distinct row of eight or
more E.–W. aligned burials (B730/731. etc.); it cut an earlier burial. Burial 631*
contained the iron girdle. Burial was therefore ongoing within both the S. and
N. parts of the cemetery by a.d. 700, although it is by no means assured that the
latter represents expansion from an earlier southern core, and it is conceivable that
burial occurred simultaneously across the whole of the excavated area from the
outset of cemetery use. Equally there may have been more than one focus of burial.
In the northern part of the cemetery the probable Iron-age pit-burial may have
been associated with a landscape feature which has since been destroyed. Evidence
for the re-use of prehistoric monuments as a focus for Early-medieval burial is
197 Hill, op. cit. in note 37, 557–9.
198 Cf. the burial of children within the churchyard at Tintagel, Cornwall, which might be associated with a
historically recorded plague in the winter of 1546–7: Nowakowski and Thomas, op. cit. in note 51, 27–9.
199 Hill, op. cit. in note 37, 28–37.
90 neil holbrook and alan thomas
increasingly being found in Wales.200 Some 4 m south-west of the pit burial was
B643 (a child) which may have had an original surface covering of white quartz
pebbles, perhaps a marker of special status. This burial is clearly not a very early
feature of the cemetery as it overlay two burials which both in turn overlay an
earlier one. Any status that the grave may once have possessed was not subsequently
respected as a number of later inhumations cut through it. No patterning in the
burials which immediately surrounded B643 could be detected.
Bodies which were buried no earlier than the a.d. 600 occur in all three parts
of the cemetery. In Area III B784 overlay an earlier burial and was in turn covered
by four sequential phases of intercutting burials, while in Area I they included
burials on both E.–W. (B110) and NE.–SW. (B333*) alignments. The date from
B380 in Area II suggests that burial in regular rows continued here after the late
7th century. There were three bodies buried no earlier than a.d. 700 in Area I and
one in Area III. The N.–S. burial B106, buried no earlier than 782 cal a.d.,
underlay a number of E.–W. aligned burials in Area I, while B688 in Area III (no
earlier than 779 cal a.d.) was the latest in a localised sequence of five phases of
burial. To these radiocarbon dated burials can be added B898* in Area III, which
was accompanied by an iron knife grave good broadly dated to the period from the
8th–10th centuries. The latest radiocarbon date from the site came from B228
(885–1035 cal a.d.), one of the NE.–SW. aligned burials which it was thought
might reflect the putative boundary to the cemetery.
To summarise, burials were being made before a.d. 700 in Areas I and III,
and before a.d. 980 at latest in Area II. Burials which must date to after a.d. 700
occur in both Areas I and III. Bodies were being interred on a NE.–SW. alignment
in Area I before a.d. 640 (B2*) and after a.d. 885 (B228). If these burials do reflect
the alignment of a boundary to at least part of the cemetery then this was therefore
in place before the mid-7th century. Burial in Areas II and III clearly continued
after this, and contrary to the belief that the boundary went out of use as the
cemetery expanded, the late date from B228 suggests that it remained a visible
feature in the latest period of cemetery use. Curvilinear enclosed cemeteries have
been demonstrated to be a feature of pre-Norman Christian archaeology in
western Britain, and monasteries in particular were generally enclosed,201 although
a recent survey suggests that the advent of enclosed Christian cemeteries should be
dated to the period from the 8th–10th centuries, and not before.202 Whatever
structure the putative boundary took it has left no trace: certainly no ditch existed
as at some sites, although less tangible structures are also known.203
Only broad deductions can be made concerning the extent of the cemetery in
its most developed form. Burial is unlikely to have continued to the west of Penland
Road as this probably marks the historic route to Cardiff, while to the north and
200 James, op. cit. in note 24; E. O’Brien, Post-Roman Britain to Anglo-Saxon England: Burial Practices Reviewed (BAR
Brit. Ser., 289, Oxford, 1999), 35–8.
201 D. A. Brook, ‘The Early Christian Church east and west of Offa’s Dyke’, 77–89 in Edwards and Lane (eds.),
op. cit. in note 24; A. Preston-Jones, ‘Decoding Cornish churchyards’, 104–24 ibid.
202 Petts, op. cit. in note 145.
203 Cf. the insubstantial Period 1 (pre-a.d. 730) boundaries at Whithorn, and the references cited therein for
monastic boundaries composed of hedges and wattles: Hill, op. cit. in note 37, 35.
a monastic cemetery at llandough 91
east the break of slope would have formed natural boundaries. To the south the
monastic enclosure is likely to have extended as far as the crest of the slope above
the small stream, and perhaps beyond. The 30 or more burials disturbed during
building work in 1979 to the south of the stream seem to date to the late 8th–9th
century to judge from the two radiocarbon dates (Tab. 8), and are thus
contemporary with burial to the north of the church (Fig. 2). One possibility is that
they represent the lay population buried outside the monastic bounds, although
equally they might represent a second cemetery close to the southern limit of the
enclosure.
demography of the cemetery
Full discussion of the demography and pathology of the burial population
must await the publication of Loe’s analyses,204 although the following broad
conclusions can be drawn. Of the population available for study, 30% were either
certainly or probably male, 25% certainly or probably female, and 26% were
children under 18. Adult males outnumbered females by a proportion of about
20%, although whether this is a particular characteristic of monastic sites of this
date must await further study. Females and juveniles were also present in the
burials in the southern area examined in 1979.205 On some monastic sites the
presence of female burials has been considered to be indicative of a lay
population,206 although it is doubtful if this need necessarily be so as historical
evidence suggests that marriage amongst the Welsh clergy was commonplace from
the outset.207 Equally the cemetery may contain some lay burials as well, as Davies
has concluded that by the late 11th century burial within the cemetery of a religious
community was normal, at least for the aristocracy, on payment of fees.208 To what
degree this practice occurred earlier is unclear, although O’Brien has remarked on
the early 8th-century historical evidence from Ireland where people were being
exhorted to leave family or tribal cemeteries and bury their dead in Christian
cemeteries.209 The excavated cemetery is therefore likely to have contained an
amalgam of clergy, their wives and children, and elements of the lay aristocracy.
Indeed given the estimated size of the cemetery, the latter group may have
comprised a significant proportion of the burial population. It may be pertinent in
this respect that Loe has observed that there is a greater variation in male statures
at Llandough than at some other cemeteries, and she suggests that this could be
due to the males being drawn from a wide geographical area. The skeletal
examination also suggests that there were lower levels of disease and greater life
expectancy than at some other Early-medieval cemeteries, perhaps a reflection of
the status and relatively high standard of living of those buried here.
204 Loe, op. cit. in note 14.
205 Owen-John, op. cit. in note 4, 147.
206 Cf. Monkwearmouth, R. J. Cramp, ‘Monastic sites’, 201–52 in D. M. Wilson (ed.), The Archaeology of Anglo-
Saxon England (Cambridge, 1976), 231, and Whithorn, where the presence of a small number of females suggested
to the excavator either a secular burial ground or the presence of monks and nuns of a double monastery: Hill, op.
cit. in note 37, 74.
207 Davies, op. cit. in note 39, 157.
208 Ibid., 187.
209 O’Brien, op. cit. in note 200, 53–4.
92 neil holbrook and alan thomas
acknowledgements
The excavation was generously funded by Ideal Homes Wales Ltd (now part of the
Persimmon Group) and we would like to thank Ray John, Bryn Davies and James Davies
for their assistance. Paul Vining of planning consultants White Young Green Planning
offered greatly appreciated support and encouragement throughout the project. The post-
excavation work and publication were funded by Cadw: Welsh Historic Monuments and
we are grateful to Richard Avent for arranging and supporting this. Many people offered
helpful advice and assistance, and in particular we thank Jeremy Knight, Louise Loe,
Charlotte Roberts, Juliet Rogers and Charles Thomas for their insights. We are particularly
grateful to Mark Redknap for facilitating the various analyses within the National
Museums & Galleries of Wales, and to all the contributors to this report. Processing of the
skeletal remains was undertaken by inmates of Ley Hill Open Prison at Bristol University
through the good offices of Mark Horton, and by volunteers and staff of the Department of
Archaeology. Grants to support the processing were kindly received from the National
Museums & Galleries of Wales and the Cambrian Archaeological Association, and we
thank Richard Brewer and Peter Llewellyn for supporting these applications. Richard
Avent, Wendy Davies, Mark Redknap, Richard Reece and the Editor kindly commented
on an earlier draft of this report. Mark Redknap is very grateful to Tony Daly for the
illustrations, and Evan Chapman for comments on the Roman finds. He would also like to
thank Kathy Bird and Penny Hill for conservation work on the hernia belt; John Deeming
and Brian Leevey for comments of fly puparia; Hero Granger-Taylor for comments on the
textile impressions; Stephen Lloyd, Scottish National Portrait Gallery, for comments on
the portraiture of James IV; and Ralph Jackson and Sian Rees for helpful comments on
the draft report.
The excavation was directed in the field by Alan Thomas and managed by Neil
Holbrook. Thanks are due to the considerable excavation team that worked on the site
over six months. The illustrations were produced by Peter Moore, and the gazetteers of
burials and finds contained within the digital archive by Emma Harrison and Ed McSloy.
The finds and archive have been deposited with the National Museum & Gallery, Cardiff.