1.

1.1 Description of the Site and its Setting.

The Gower Peninsula.

Goat’s Hole cave, Paviland is located (Fig 1.1) on the south coast of the Gower peninsula
(Ordnance Survey map reference SS 437858). The Gower presents for the most part a
gently rolling surface varying from 15-36m above sea-level, and bounded on its north,
west and south sides by cliffs. The cave itself commands an imposing view over the
Bristol Channel, Exmoor and Lundy Island often being visible. It is adjacent to a rocky
gully named “Foxhole Slade”, leading to the cliff known as “Yellow Top”, although the
Ordnance Survey six inch sheet applies this name to a crag a little to the east. Foxhole
Slade provides the easiest access from Goat’s Hole to the plateau of the Gower. Goat’s
Hole can be easily located by following a kilometre-long path sign-posted from the main
road (B4247) at Pilton Green. It is approximately two kilometres east of Mewslade Bay,
and three kilometres west of Port Eynon.
The entire Gower Peninsula is designated a Site of Special Scientific Interest
(SSSI) (Countryside Council for Wales 1993).Two areas of the Gower Peninsula are also
National Nature Reserves, the sands and dunes around Whiteford, and the South Gower
Coast from Port Eynon Point to Worm’s Head. The Paviland cliffs are excluded from the
National Nature Reserve area, as is Goat’s Hole cave itself, Fox Hole Slade and the
Knave Hill Fort (Countryside Council for Wales 1994).

Paviland.
There are three main cave sites, referred to as Goat’s Hole, Hound’s Hole and Fox Hole
Slade cave. A cave in which some lithic artefacts were found (Morgan 1910) runs
beneath Goat’s Hole; its floor lies at the level of the wave-cut platform. This cave has
been referred to as Black Hole. “Paviland” has been used interchangeably with “Goat’s
Hole”, the largest site, and that of greatest archaeological importance.

Goat’s Hole.
Goat’s Hole cave is on the sea cliff, facing south-west onto the Bristol Channel, at an
altitude of 14.50 OD. The cave can usually be reached though the intertidal zone at low
tide, when low tides fall below 4.0m (Swansea, King’s Dock). With higher tides wading
 2.

is necessary. A second route allows the cave to be reached at any time, via a scramble
along a natural ledge on the cliff from the west side of Foxhole Slade.
The cave entrance is pear-shaped and inclined to the east, being 10m high and 7m
wide, with a passage of approximately 30m length. The passage runs into the cliff in a
south-south-west direction. A chimney leads to daylight visible 20m above, close to the
east wall. The upper entrance to the chimney shaft forms a pothole penetrating
downwards from a ledge, of a horizontal keyhole shape, length 5m, maximum width 3m,
and at a height of 32.00m OD.
There are two large hollows in the bedrock floor (A and B). The outer hollow (B)
is almost circular and 3.5m in diameter, its uppermost layer being that of a modern storm
beach. The inner hollow (A) is larger, egg-shaped and 6m long by 4m wide, the outward
pointed end filled with a little sand, wave-tossed pebbles and fragments of collapsed
cemented breccia. Beyond the hollows the floor rises for a distance of 10m and consists
of solid bedrock except for a 1m long narrow hollow containing recent wave-tossed
pebbles and water-worn sub-angular pebbles.
Paviland is accessible (without rock-climbing) at low tide only; although it is not
itself within the tidal zone. However on 9th March 1989 following a storm and during a
high spring tide (10.39m Swansea, King’s Dock) the outer hollow became filled with sea
water. With a maximum depth of 1m it is possible to calculate the volume as 16,000
litres, which drained away during the following month. This scouring by the sea revealed
the surface of an in situ deposit (Davies 1993). Inspection of the deposit by Dr. Stephen
Aldhouse-Green eventually led to the 1997 field season.

Hound’s Hole.
A further cave, Hound’s Hole, also known as Little Paviland or Goat’s Hole West, can be
found about 50m west of Goat’s Hole, also fully within the reach of the highest storm
tides. According to Davies (1993) this cave is of phreatic origin, and is wave modified.
Hound’s Hole is a narrow rift 6m high by 4m wide developed on a vein of calcite which
penetrates 30m with daylight reaching the end. The entrance faces south-west. About
halfway in there is a remnant of cemented wall breccia and stalagmite floor, and a
stalagmitic hanging bridge can be seen at a height of 3m. It is clear from the existence of
this bridge, which is presumably Holocene in date, that Hound’s Hole was once
 3.

completely filled with deposits which the sea, and the activities of collectors, has
removed (Aldhouse-Green in press). The cave was excavated by Buckland (1823) and by
Odo Vivian (1912). Buckland (1823, 93) recovered an abundant fauna including horse,
deer and bear. Vivian recovered about 7 lb of bone (1912, 16) but neither exploration
yielded any humanly modified material.

Black Hole.
Below Hound’s Hole and slightly to the east at beach level is a sea cave, “Black Hole”,
running in some 30m and accessible at exceptionally low tides. The sea cave is visible
from the Hound’s Hole entrance as a boulder strewn chasm, 3m high and 2m wide. The
whole length is completely filled at high water and a fresh-water spring was located at
the rear. Until recently Paviland farmhouse was supplied with water from this spring.
<Note: Lowest Spring tide 1998 will be at 13.14 on Sunday 29th March (2.8m Swansea,
King’s Dock). Details of the Sea Cave can be checked then.>

Fox Hole Slade Cave.

This cave had not been explored before the 1997 season. It lies towards the top of
Foxhole Slade, on the western side of the gully. A talus cone had formed at the entrance,
and the surrounding limestone cliff is still rapidly exfoliating. The entrance chamber now
seen is therefore of recent formation in geological terms. Foxhole Slade cave proved to
contain a sequence of deposits, possibly several metres deep, which filled the cave almost
to its roof. It was possible during the 1997 field assessment to draw plans of the cave for
around 20m.

1.2 The 1997 Excavation: the Current Research Project Field Assessment.
Aims of the Field Assessment.
Goat’s Hole, Hound’s Hole and Foxhole Slade Cave were investigated during the field
season from June 21st to July 12th 1997, under the direction of Dr. Stephen Aldhouse-
Green, University of Wales College, Newport. Information gained added to the
 4.

descriptive data available for the Paviland caves. There were three prime aims of the
field assessment (Aldhouse-Green in press):
• To produce a definitive survey of the caves, and record surviving areas of deposit.
• To excavate half of Hollow B. Hollow B is the outer of two sediment-filled hollows in
the bedrock floor of the cave.
• To collect naturally detached specimens of cemented breccia and stalagmite which
litter the modern floor of the cave, for analysis and preservation in the National
Museums and Galleries of Wales.

Goat’s Hole.
An in situ deposit was revealed in Hollow B (Plate 1.x) (Fig. 1.x) composed of small,
angular limestone clasts in a sediment matrix. This deposit probably formed part of a
talus cone, now completely eroded, of colluvial origin (Aldhouse-Green in press), the
accumulation of which probably spanned the entire Devensian. There were a series of
layers of very fine definition within this colluvium (Layers 3,6,7,8). No such stratigraphy
was recorded by Sollas (1913) in the archaeological layers. However an overall
stratigraphy can be reconstructed, as reproduced below from Aldhouse-Green (in press).

Table 1.x: Stratigraphy of Goat’s Hole Cave.

Layer Interpretation Finds Climate Age Source
(OIS)
A - Stalagmite Porous stalagmite and marine shells _ warm 1 Aldhouse-Green
(1997)
B - Upper Scree 1 Small angular scree and rounded clasts artefact and fauna rich cold 2-3 Sollas (1913)
C - Ochreous Clay Not clear - reworked ochre artefact and fauna rich cold 2-3 Sollas (1913)
D - Upper Scree 2 Small angular scree and rounded clasts artefact and fauna rich cold 2-3 Sollas (1913)
E - Grey-white Band Weathering of underlying deposit none ?warm ?5e Sollas (1913)
F - Pebbles and Sand Beach deposit none ?warm 5e Sollas (1913) &
Aldhouse-Green
(1997)
G - Lower Scree Small angular scree and rounded clasts fauna rare cold 6 Aldhouse-Green
(1997)
H - Bedrock _ _ _ _ _

The Devensian cave sediments were capped by a stalagmitic floor, probably of

Holocene age. Fragments of this floor containing modern seashells are preserved
cemented to the cave wall at a height of 5m or more from the bedrock floor. The external
 5.

face of the talus cone would have been eroded by rising sea levels at the beginning of the
Flandrian, leading to a collapse and slumping of the deposit, and possibly a reduction in
the accessibility of the cave. This may be part of the explanation for the few post-
Mesolithic finds from the site.
The artefact-bearing levels are B, C and D, which also produced a cold climate,
Coygan-type fauna (Currant & Jacobi 1997) typical of OIS 3. The grey-white band E
was interpreted by Sollas as a weathering horizon which may indicate a hiatus in scree
formation corresponding to a milder climatic phase. On the northern side of the cave it is
coextensive with a beach deposit, a layer of sand and rounded pebbles; Layer F. It is
thought (Aldhouse-Green in press) that Layer F could be of Ipswichian (OIS 5e) age as,
during this interglacial the sea level would have been sufficiently high as to wash into the
cave, as noted for Minchin Hole Outer Beach (ref.), and raised beaches at a similar
height to Paviland in South Wales (ref. Bowen). The fact that Sollas reports the layer of
beach pebbles as thickening toward the wall of the cave, where they “descended in a
vertical sheet” (1913, 336) lends support to the interpretation that Layer F may be
Ipswichian. Some cave deposits undergo “shrinkage” away from the cave walls, leaving
spaces down which overlying - younger - sediments may move (Aldhouse-Green in
press).
The emplacement of the beach would have occurred together with removal of
previous sediments, and reshaping of the cave entrance. Denekamp Interstadial (30,000
BP) high sea level stands during OIS 3 may also have been of sufficient height (Milliman
& Emery 1968, 1123) but the resulting deposits, if any, have been lost through modern
erosion and excavation.
Layer G underlies the beach deposit. No artefacts were recovered from this
Lower Scree, although a cold climate fauna, heavily mineralised and splintered, was
recovered. This fauna is not chronologically diagnostic but, if Layer F is of 5e age, the
Lower Scree would belong to OIS 6 (Wolstonian glacial). A date on the Lower Scree
was produced by TL: 169 +23/ -46 ka BP (QTLS-PVG 11). There is a 68% probability
that the layer F sediment was deposited between 290 ka - 124 ka BP. At 2 standard
deviations, giving 95% confidence, the age of the sediment is >102 ka BP, with an
infinite upper limit. Comment by Debenham (in Aldhouse-Green, in press) states that the
sample was most likely last exposed to light before OIS 5. It is likely to be of OIS 6 age.
 6.

Hound’s Hole.
In 1997 a trench was opened up at right angles to the line of the cave, in proximity to the
breccia on the west wall. A sequence of cemented breccias and stalagmite, as well as
uncemented deposits, was revealed. A sparse fauna was recovered from layer 3, which
although not as heavily mineralised, appears to be similar to that from Layer F in Goat’s
Hole. One TL date from the base of a broken stalagmite pillar gives 130 ± 16 ka BP
(QTLS-PVH 11), an Ipswichian age. The underlying deposit, a breccia which is probably
the oldest encountered in Hound’s Hole (layer 6) may therefore be of OIS 6 age,
contemporary with the basal deposit in Goat’s Hole. A single lithic find came from the
cave - a fragment of thick blade possibly made on volcanic tuff. Whilst not a diagnostic
find, the selection of a non-flint raw material and its thick-blade technology are
suggestive of an Early Upper Palaeolithic age. The find came from the base of the
modern storm beach and so was not securely stratified (Aldhouse-Green et al. 1997).

Foxhole Slade Cave.

A trench was excavated to a depth of 0.85m, revealing a sequence evidently much
disturbed by animal activity. There are two main layers, a Humic Scree (Layer 2)
overlying a soliflucted Scree (Layer 3). Both layers produced artefacts of Early
Mesolithic date, as well as modern material which had been introduced through animal
burrows. Also present was a flint blade fragment, potentially of Late Upper Palaeolithic
age. Layer 3 produced a Pleistocene fauna including reindeer and collared lemming.
Foxhole Slade Cave can be identified as only the third Early Mesolithic site known in
Gower - along with Burry Holms (Green & Stanton 1984,) and Cat Hole (McBurney
1959; Campbell 1977, 55). Foxhole Slade Cave is only a few miles from Rhossili where
Worm’s Head cave yielded a human ulna dated to 8,800 ± 80 BP (OxA-4024). It is close
to Goat’s Hole where later Mesolithic human remains have been dated to 7,190 ± 80 BP
(OxA-681). Underlying sediments at Foxhole Slade Cave may well preserve undisturbed
evidence of Upper Palaeolithic occupation (Aldhouse-Green in press).

1.3 The Geology of Gower.

Devensian Geological Succession in Gower.
 7.

Bacon Hole contains a complete early Devensian succession (Stringer et al. 1986),
representing a major part of OIS 5e to 5c (a warmer interval with faunal remains, around
105 ka BP). With the exception of Long Hole Cave identification of discrete deposits of
Middle Devensian age in Wales has not been possible. Deposits of that age are probably
represented in widespread head deposits. “Head” is rubble deposit, on slopes along the
coast, resulting from solifluxion processes. Its surface adopts a terrace form, reflecting
the underlying slope (Kidson & Tooley 1977, 278).
In south-west Wales the Late Devensian glaciation was occasioned by ice coming
from Welsh centres of dispersal (over the high ground of Powys and Brecon) and from
the Irish Sea basin. The extent and nature of the ice-sheets, as well as the timing of their
expansion and shrinkage has been a subject of considerable debate (eg. Bowen 1973) and
is still poorly understood.
Devensian ice only covered the eastern part of the Gower peninsula. In east
Gower, therefore, late Devensian glacial deposits lie on periglacial and colluvial deposits
of the earlier part of the Devensian stage (before 26 ka BP), which in turn rest on the
10m Last Interglacial (Ipswichian) raised beach of south Wales.
In west Gower, in the area around Paviland, the 10m last interglacial beach is
overlain by periglacial slope deposits. These sometimes include glacial material which
George (1933) calls the “Older Drift”. Its deposition, however, is Devensian - it is
redeposited by solifluxion from older glacial residues which occur on the coastal
plateau of south-west Gower.

Drift Geology of Gower.

In addition to being a source of raw material, the distribution of erratics in glacial
depositions - here termed “drift” - enables the strength and dimensions of some Welsh
ice-streams to be estimated. For example, in the South Wales coal field, the Vale of
Neath ice deposited typical Brecknockshire drift at heights above 300m near Pen
Moelgrochlef (Campbell & Bowen 1989, 18), and was therefore at least 300m thick.
Campbell and Bowen give other examples and conclude that most of the Welsh upland
was submerged at the maximum of the late Devensian glaciation. The implications of this
polar climate will be discussed later (p. 00). George (1932; 1933) showed that Gower
had been covered by two ice masses of different origin. One had a northerly source and
 8.

deposited largely local, Welsh rock types; the other had a source to the west and brought
erratics from the Irish Sea basin. The Irish Sea deposits were more widespread on the
south and western margins of the peninsula. The fact that the rock types were mixed was
used as evidence for the contemporaneity of Irish Sea and Welsh ice masses during the
“Older Drift” or pre-Devensian (Bowen 1970) glaciation.

Paviland Site Formation.

Cave passages form through a limestone karst where there is an available flow of water
with chemical potential to dissolve the limestone, in a favourable geological structure
(Waltham et al. 1997, 10). Gower displays a typical karst landscape, but there are far
fewer extensive caves than in the North Crop. Reasons for this are threefold, firstly the
small size of the peninsula and relatively low topography provides a limited rainfall
catchment area. Owing to the absence of peat the water will not be as acidic. In addition
the combination of steeply dipping bedding planes and many faults and joints allow
underground pathways for water drainage. Solution of the rock has therefore not
progressed to as great an extent (Ede & Bull 1989, 212).
On the south Gower coast a number of caves of archaeological and
palaeontological interest are developed on the fault lines mentioned above. These include
the “very old solution cave fragments” of Bacon Hole and Minchin Hole (Waltham et al.
1997, 221). Some larger chambers of equal age such as Llethrid Swallet (Ede & Bull
1989, 215) are found inland on the peninsula.
The faults act to channel water, leading to the formation of small open-air
systems along their line (Jennings 1985, 145). The entrance to Goat’s Hole clearly shows
an elliptical slope caused by solution on a vertical fault. To the west, Hound’s Hole and
Mallow Hole, developed on another vertical fault, have probably been formed in a
similar way. Fox Hole Slade Cave is formed on a possibly more recent fracture line
trending roughly west-east, at right angles to the majority of tear faults. The shape of the
cave, and the presence of ripplestone suggests that at one time it was an active stream
passage, perhaps to feed a conduit in Fox Hole Slade.
Davies (1993, 12) suggests wave action as the main factor in the formation of
Goat’s Hole. Alteration of the cliff and cave entrance by modern wave action is evident,
and probably also occurred during earlier times of high sea level - as at Minchin Hole
 9.

and Bacon Hole where raised beach deposits are found. However, the existence of
phreatic or solutional features probably allowed access and widening by the sea.
The age of Goat’s Hole is not known, although it must be pre-Ipswichian, as
plausibly OIS 6 sediments were uncovered in the most recent excavations (Layer F). As
Sollas (1913, 331) remarks, the sides of the cave are smooth but undulating, and have
many small tributary channels and crevices which are likely to be solutional. Degraded
but probably Holocene flowstone can be seen above the presumed highest level of
sediment on the cave.
The “chimney” or phreatic tube on the east side of the cave suggests a period of
more intense, possibly post-glacial water flow through the cave, which must be noted as
it may have caused erosion or removal of the sediments. Previously water-filled passages
such as this tend to be circular in cross-section; since water contact is present on all sides,
solution occurs equally in all directions (Trudgill 1985, 72). The top of the chimney
bears Holocene flowstone formations; there are also many phreatic solution marks in the
rock around the mouth of the pothole and on the walls inside. Some sediments remain in
small alcoves but none are artefact-bearing. In July 1992 the relatively clean walls of the
chimney were examined for any prehistoric human markings but nothing was found
(Davies 1993, 14).

Local Raw Materials.

Carboniferous chert, ochre and possibly manganese ores are the only materials strictly
local to Gower which have been utilised during the Palaeolithic. The quartz mentioned
by Sollas (1913, 353) was probably naturally present in the cave sediments, rather than
being imported for use by man.
Carboniferous chert, which is black in colour, forms bands in the limestone;
having been deposited around 360 Ma at the same time as the coal fields. The cherts vary
in form, from tabular to a more homogenous matrix which breaks with a conchoidal
fracture and responds well to knapping. This raw material will be described in more
detail later, as it was extensively used in the blade point and Aurignacian II phases of the
Early Upper Palaeolithic. Chert outcrops have been described in Three Cliffs Bay; and
beach pebbles of flawed Carboniferous chert can be readily acquired from the south
Gower beaches, for example Port Eynon Bay (Jenkins 1997, 26). Although the material
 10.

is fine-grained, it does not seem to include micro-fossils. This, together with its wide
range of morphology at any one outcrop, renders sourcing impossible (pers. comm. Dr.
Jana Horek, December 1997).
The ochre - iron oxide - from Goat’s Hole is primarily associated with the
inhumation. It is attached to the skeletal remains and perforated shells, and in the
perforation of the “pendant”. Samples of sediment also exist (Young, unpublished 1997,
1). Kidney (botryoidal) ores and “layered” ores have come from the cave, and these are
found locally in the major ore bodies in south Wales, suggesting deliberate collection
from a large specific source or sources. Ochre is not present in the cave itself - red
staining in Goat’s Hole is probably biogenic (i.e. algal) in origin. In Fox Hole Slade red
sediments are probably Triassic sandstone infill and degraded limestone. Sources for
non-haematitic red ochre include High Meadow Mine, and Clearwater Mine near
Hereford, Forest of Dean (Sibly 1919, 32).
Manganese oxides and hydroxides, to be referred to as “wad” (Young,
unpublished 1997, 1) and labelled by Sollas as “psilomelane” are rare in south Wales,
although trace elements have been found in some iron mines (for example Sibly 1919,
17).Wad occurs extensively in the Mendips, Devon and Cornwall. Young suggests that it
has been used as a pigment, possibly to give a black colouration (Young, unpublished
1997, 1), whereas Sollas attributes Palaeolithic human interest in the mineral to its
“strange heaviness” (1913, 353). Poor quality flint and igneous rocks (George 1933, 214)
are also available locally to Gower, in the glacial, or “drift” deposits.

Gower Field Trip.

Aims.
The aim of the field trip was to find sources of lithic raw material on the Gower, which
yielded stone of similar morphology to that in the Goat's Hole collections. It is not
assumed that the sources were identical to those used in the Early Upper Palaeolithic,
simply to discover whether workable materials are present in the general area of Gower
(Zone 1 in fig. 5.1), which may have been utilised during the Palaeolithic. As
conclusions have import for the movement of humans during the Palaeolithic, the search
was made with care, three hours in each of the three areas detailed below. This study was
 11.

made in addition to that by Emma Jenkins (1997). Identification of the rock types was
done by eye, by Heather Jackson.
Flint was searched for specifically, as it was not known in what shape this
material most commonly exploited could be available. It will be suggested that humans
collected flint from many sources, among which a ‘drift’ source is possible but by no
means exclusive.
Possible sources of flint on the Gower are:
1. Beach pebbles. Flint is widespread on the south Gower beaches as found by Jenkins
(1997, 31). Although deeply patinated and very often flawed, some of the pebbles are
large in size, the largest-found measuring 29.5 x 16 cm. They are also abundant, with an
approximate density of one every two m2 in Overton Mere (OS ). Difficulty in knapping
would come primarily from the rounded shape, but if this is overcome by splitting, it
should be possible to work the pebbles.
The primary origin of beach pebble flint is not known for certain. Beach
formation is Holocene, so the appearance of beach pebble flint is probably not relevant to
Early Upper Palaeolithic studies. Flint now seen on in shingle may have come from four
sources:
1. Reworked glacial drift deposits.
2. Raised beaches, such as those seen between Three Cliffs Bay and Caswell Bay.
3. From solid geology, via erosion of the cliffs and subsequent transport by marine
action.
4. From offshore geological deposits, having been moved by the marine action.
The first three origins were explored during the field trip.

Topsoil/ Devensian till. Sollas (1913, 337) concluded that flint of all kinds was probably
collected from ‘the adjacent glacial drift’. Lowe (pers. comm., 1999) agrees with the
explanation presented by Sollas. Flint present in unconsolidated till would be easy to
extract and presumably would be easily noticed in water channels downcutting through
the deposit. The accumulation of flint types by glacial action would account for the wide
variety of types of flint present, but not for the differing conditions of their cortex. The
origin of flint in glacial till, and the distance it has been transported, is not known; some
at least may have originated in the solid geology of the Gower itself.
 12.

Direct from solid geology. Cretaceous (142 - 65 Ma) chalk, which probably included
flint, was laid down on the Gower peninsula (Lowe pers. comm.) Coverage was
discontinuous, and these rocks, soft in comparison with the underlying
limestone/millstone grit, were eroded in the subsequent era, as well as being covered by
glacial deposits. No Cretaceous rocks remain today, but the resistant flint and chert
within the soft chalk may have survived, gradually washed to the sea or dispersed on
upland surfaces. Flint would then appear in glacial till and beach deposits.

In the raised beach platform. The surface of the Ipswichian (OIS 5e) Patella raised beach
was studied for presence of flint inclusions. The flint would have been accumulated from
solid geology and from gravel on the beach of the last interglacial. The raised beach was
considered as it forms a prominent feature which may have been visible in the Upper
Palaeolithic, in proximity to Goat's Hole.

On Upland Surfaces. Scatters of material on an upland surface was investigated, for

material brought by the most southern expansion of the Irish Sea (pre-Devensian) ice, or
for erratic rock originating from Gower solid geology. It is admitted that field-walked
geological collections today are probably not comparable to the land surface yield in the
late Devensian.

Locations searched were, from east to west:

1. Shingle beach between Caswell Bay and Brandy Cove (OS 587877), Devensian till
cliffs at Caswell Bay (590877) and the Patella raised beach between Caswell Bay and
Brandy Cove (OS 587874).
2. Land surface at East Slade near Horton (OS 488856)
3. Pre-Devensian till cliffs at Rhossili Bay (OS 414882 and OS 416885).
Two hours were spent field-walking at each location.

Results.
Caswell Bay. Flint does exist in the Caswell Bay raised beach, but the flint observed was
in fragments (c. 1 cm2), too small to use for artefacts, and present in a very low density
 13.

(approximately 1 piece per 8m2). The flint was water-worn. Smooth, and orange-stained.
It also would have been difficult to extract from the concreted raised beach; no samples
were taken in the interests of conservation. The Patella raised beach in the vicinity of
Caswell Bay was not a Palaeolithic source of flint, and did not contribute greatly to the
beach pebble flint.
No flint was found in the Devensian till cliffs at Caswell Bay. The shingle beach,
however, did yield flint, in a size up to 8 cm diameter, in a density of approximately one
every two square metres). Other knappable materials were present, in similar sizes and
densities - Carboniferous chert, unspeckled rhyolite and sandstone, the latter two
probably originating in Pembrokeshire, while the Carboniferous chert is present in
nearby cliffs.

East Slade. The location of East Slade was chosen because the land surface retains
elements of the Irish Sea till, which had a west to east movement across the peninsula.
Fields at a height of 70m OD were searched, and small flint nodules were found in
disturbed areas of topsoil. The flint was present in an approximate density of one nodule
per square metre, the nodules being 5 cm or less in diameter. All flint was grey in colour,
with a deep white patination, and is similar to that found as beach pebbles.
The density of topsoil flint was higher at East Slade than observed in other areas,
such as the fields at Paviland manor. The patina was not orange-stained, as is the case in
the Paviland Manor fields, and the nodules were larger. On the basis of these differences
it was decided to attribute the flint to Irish Sea till rather than remnants of the solid
geology. One might expect any ‘solid geology’ flint to be more uniform in appearance
over the peninsula.
Irish Sea till as seen at East Slade may be a source both for grey beach pebbles
and for the grey flint component of the Goat's Hole collections. Against this, the small
size of the pieces found must be borne in mind, as must the possibility of an offshore
source of grey flint, the origin of both beach pebbles and till. No other relevant raw
materials were found inland at East Slade.

Rhossili Bay The location of Rhossili Bay was chosen because the cliffs at the southern
end of the bay comprise a mixture of Welsh and Irish Sea glacial till. Welsh till had a
 14.

north to south movement across the peninsula, carrying erratic rock from north Wales.
Redeposited and slumped Devensian till ‘head’ deposits were seen/ Rolled and chatter-
marked flint nodules were found in this deposit. As at East Slade, the nodules were of
deeply white patinated light grey flint, the largest of 8 cm diameter. Head deposits
further north along the beach did not yield as much flint. The density of nodules is very
approximate owing to the slumping and gradient of the cliff, one every 10m2. The
Rhossili handaxe (Green 1981) was found in this area, which if it has not been
transported itself, provides evidence that larger nodules of the same material existed in
the area. It is thought that Irish Sea till is the most likely source of this flint.

Conclusion.
The grey flint used at Goat's Hole may have come from deposits west of the Gower,
either as till or exposed solid geology. The river-gravel stained, pitted cortex of some
pieces supports the former explanation. During the Palaeolithic scanty vegetation cover
and the pebbles exposed and cleaned in water channels may have enabled people to find
till flint more easily than today. As seen at Caswell Bay, Carboniferous chert is also
available locally in a naturally fragmented state.
However, questions still remain as to the nature of transport of honey-coloured
flint, green-grey and black flint which appears to have come from primary sources
outside Gower. The presence of useable ‘native’ flint and Greensand chert as a remnant
of Cretaceous solid geology remains to be proven, but Upper Palaeolithic use does not
seem likely in terms of what remains only 30 Ka later.
Rhyolites may have been transported eastward from Pembrokeshire with the Irish
Sea Till and can be found on the south Gower coast, however the large pieces used at
Goat's Hole, and the rugged rather than worm surface of some pieces suggests that
collection form primary sources occurred in addition to ‘drift-combing’ collection during
the Palaeolithic.