Content uploaded by Tony Pollard
Author content
All content in this area was uploaded by Tony Pollard on Jul 28, 2014
Content may be subject to copyright.
The excavation of four caves in the
Geodha Smoo near Durness, Sutherland
by Tony Pollard*
with contributions by D Alldrit, J H Barrett, E Photos-Jones,
R Cerón-Cerrasco, C Smith and R Squair
* GUARD, Department of Archaeology, University of Glasgow, Glasgow G12 8QQ
Scottish Archaeological Internet Report 18, 2005
www.sair.org.uk
SAIR 18 web page SAIR 18 contents
Published by the Society of Antiquaries of Scotland, www.socantscot.org
with Historic Scotland, www.historic-scotland.gov.uk
and the Council for British Archaeology, www.britarch.ac.uk
Editor Debra Barrie
Produced by Archetype Information Technology Ltd, www.archetype-it.com
ISBN: 0 903903 87 3
ISSN: 1473-3803
Requests for permission to reproduce material from a SAIR report should be sent to the Director of the
Society of Antiquaries of Scotland, as well as to the author, illustrator, photographer or other copyright
holder.
Copyright in any of the Scottish Archaeological Internet Reports series rests with the SAIR Consortium
and the individual authors.
The maps are reproduced from Ordnance Survey material with the permission of Ordnance Survey on
behalf of The Controller of Her Majesty’s Stationery Office. ©Crown copyright 2001. Any unauthorized
reproduction infringes Crown copyright and may lead to prosecution or civil proceedings. Historic Scotland
Licence No GD 03032G, 2002.
The consent does not extend to copying for general distribution, advertising or promotional purposes, the
creation of new collective works or resale.
Contents
List of illustrations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .v
List of tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .vi
1 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
2 Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
2.1 Site location and description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .2
2.2 Project background. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .3
3 Smoo Cave . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
3.1 Background and methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .6
3.2 The Smoo Cave section . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .6
3.3 Summary and interpretation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .8
4 Glassknapper’s Cave, Antler Cave and Wetweather Cave. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
4.1 Background and methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .9
4.2 Glassknapper’s Cave . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .9
4.3 Antler Cave. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .12
4.4 Wetweather Cave . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .13
5 Material Culture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
5.1 Pottery by Robert Squair . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .15
5.1.1 Methodology. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .15
5.1.2 General description of fabrics. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .15
5.1.3 Vessel catalogue . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .15
5.1.4 Interpretation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .17
5.1.5 Depositional practices and post-depositional processes . . . . . . . . . . . . . . . . . . . . . . . . . 18
5.1.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .19
5.2 Bone and antler artefacts by Tony Pollard (with species identifications by Catherine Smith) . . . 19
5.3 Metal objects. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .20
5.3.1 Iron nails . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .20
5.3.2 Copper-alloy pin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .21
6 Analysis of Slag and Fuel Samples from Smoo Cave by E Photos-Jones . . . . . . . . . . . . . . . . . . . 22
6.1 Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .22
6.2 Results. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .22
7 Animal and Plant Remains. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
7.1 Aminal bones from Smoo Cave by James H Barrett . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .23
7.1.1 Methodolgy. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .23
7.1.2 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .23
7.1.3 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .26
7.2 The fish remains from Glassknapper’s Cave, Antler Cave and Wetweather Cave
by Ruby Cerón-Carrasco . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .27
7.2.1 Methodology. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .27
iii
7.2.2 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .27
7.2.3 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .28
7.3 The mammal and bird bone from Glassknapper’s Cave, Antler Cave and Wetweather Cave
by Catherine Smith . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .28
7.3.1 Methodology. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .28
7.3.2 Species present . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .29
7.3.3 Age of animals at death . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .30
7.3.4 Economy of the site: evidence of butchery and bone working . . . . . . . . . . . . . . . . . . . . 30
7.3.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .32
7.4 Marine shells by Ruby Cerón-Carrasco . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .32
7.4.1 Methodology. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .32
7.4.2 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .32
7.4.3 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .35
7.5 Plant remains by Diane Alldritt . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .35
7.5.1 Methodology. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .36
7.5.2 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .36
7.5.3 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .37
7.5.4 Summary and overall conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .38
8 Conclusions and Discussion. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
8.1 Prehistoric activity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .40
8.2 Viking/Late Norse and later activity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .43
9 Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
10 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
iv
List of illustrations
1 Site location . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .3
2 Site plan of Glassknapper’s Cave and Antler Cave. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .4
3 Smoo Cave: section through the midden and occupation deposits. . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
4 Smoo Cave . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .11
5 Detail of section G–H. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .12
6 Site plan of Wetweather Cave . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .14
7 Metal and pottery artefacts. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .16
8 Bone artefacts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .19
9 Bone artefacts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .20
10 Metal artefacts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .21
v
List of tables
1 Radiocarbon determinations from Smoo Cave and Glassknapper’s Cave (GKC).
Dates have been calibrated using OxCal v3.8 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .10
2 Fish specimens from Smoo Cave by fragment count. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .24
3 Comparison of nine (Q1) haddock cranial and ‘appendicular’ elements from
Smoo Phase 5 and Earl’s Bu . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .25
4 Chi-square comparison of haddock cranial and ‘appendicular’ specimens from
Smoo Phase 5 and Earl’s Bu . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .25
5 Total numbers of mammal bones found in Glassknapper’s Cave (GKC),
Antler Cave (AC) and Wetweather Cave (WWC) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .30
6 Total numbers of bird bones found in Glassknapper’s Cave (GKC) and
Wetweather Cave (WWC) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .31
7 The marine shell representation from Smoo Cave . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .33
8 Marine molluscs from Geodha Smoo: Glassknapper’s Cave 100% sorted samples. . . . . . . . . . . . . . 34
9 Identified plant remains by dated phase . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .36
10 Identified cereals by context . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .37
11 Number of weeds recovered from each phase, divided into ecological groupings . . . . . . . . . . . . . . . 38
12 Comparison of artefactual and ecofactual material from the four caves . . . . . . . . . . . . . . . . . . . . . . 41
vi
1 Summary
In response to the threat posed by marine and river
erosion, a series of deeply stratified midden deposits
was excavated in caves leading off a narrow, rock-cut
inlet known as the Geodha Smoo, near Durness,
Sutherland. These included the famous Smoo Cave
(NGR: NC 4136 6714), at the southern end of the
inlet; two smaller caves cut into the western wall of
the inlet (Glassknapper’s Cave and Antler Cave);
and a fourth cave (Wetweather Cave) in the eastern
wall. The majority of excavated deposits from these
caves appear to relate to Viking/Norse or post-Norse
activity, with fish bones, marine shells and mammal
and bird bones representing the processing and
consumption of marine and terrestrial foods.
Possible evidence for metalsmithing in the form of
iron slag and boat nails could suggest that boats
were repaired in the sheltered inlet. Four radio
-
carbon dates from Smoo Cave and Glassknapper’s
Cave provide evidence for use of these sites between
the eighth and 11th centuries
AD. Convincing
evidence for pre-Norse activity, although unsup
-
ported by radiocarbon dates, was recovered from
Glassknapper’s Cave in the form of probable Iron
Age pottery, while late Neolithic pottery came from
floor deposits in the Wetweather Cave.
1
2 Introduction
This report combines the results of two programmes
of archaeological fieldwork. The first took place in
1992 and involved the cleaning back, sampling and
recording of a midden section inside Smoo Cave prior
to the erection of a protective wall (Illus 1). This
remedial work was necessitated by erosion of the
midden face by the river, known as the Allt Smoo,
which flows down through the cave into the inlet and
the sea. The project was sponsored by Caithness and
Sutherland Enterprise and facilitated by Highland
Council’s Archaeology Service. The second, and more
extensive, programme of work was funded by
Historic Scotland and took place outside Smoo Cave
proper in 1995, in a pair of adjoining caves in the
western wall of the inlet some 80m to the north of
Smoo Cave and in a fourth cave in the eastern wall
(Illus 1).
2.1 Site location and description
Smoo Cave (NGR: NC 4138 6714) is situated at the
head of a narrow inlet (Geodha Smoo) that runs
inland for about 600m from the northern coast of
Durness, Sutherland (see Illus 1). The main cave,
which is of impressive dimensions (approx 35m wide
by 50m deep), has been carved into the Cambrian
Limestone by successive episodes of high sea level
over the past several hundred thousand years. The
main cavern is connected to several smaller fresh
water chambers, eroded by the Allt Smoo as it cut its
way to the sea.
Smoo Cave was carved along the line of a weak
fault in the limestone, both by the river which today
flows through it and by the sea which at times of
very high tide laps the back wall of the cave. The
inlet itself was created as the cave roof progres
-
sively collapsed with the deepening of the cavern
over hundreds of thousands of years (Gleed-Owen
1992).
Still a popular tourist attraction on Sutherland’s
north coast, Smoo Cave has drawn visitors and
comment since at least the 18th century, as illus
-
trated by this early reference, dating from between
1720 and 1745 and cited in Macfarlane’s Geograph
-
ical Collections:
Betwixt the two Sangoes at the shore, there is a
cave stretching pretty far under ground with a
naturall [sic] vault above; Its called Smoa [sic], at
the mouth of it is a harbour for big boats, on the
floor of the cave there is room enough for 500 men
to exercise their arms, there is a burn comes out of
the earth in one side of the said cave and forms a
large and deep pool there where trouts are catched
and then runs out of the pond to the sea; there is
also a spring of excellent water in the other side of
the said cave (Mitchell 1906, 192).
The investigation of Smoo Cave reported here,
though limited in extent, represented the first
systematic archaeological work to be carried out
inside it. The deposits are thought to have been
‘turned over’ in 1904, with bone pins recovered
(Johnston 1981). No evidence for this earlier distur
-
bance was noted during the 1992 investigation. Very
little was previously known about the age of the
deposits, although it has been suggested (Keiller
1972) that recovered bone and stone artefacts
resembled Mesolithic tools recovered from the so-
called Obanian cave sites on the west coast.
As noted above, several smaller caves lead off the
Geodha Smoo. The most obvious of these are two
shallow caves situated side by side in the western
wall of the inlet, some 80m to the north of the main
cave: Glassknapper’s Cave to the south and Antler
Cave to the north. The cave mouths open directly
onto a narrow shingle beach, which here occupies a
shallow concavity in the inlet wall.
The caves that today exist outside the main cavern
may, in the distant past, have been side chambers
leading off the main cavern now represented by
Smoo Cave. With the southward retreat of the main
cavern, these side chambers were left behind as
separate caves opening out from the walls of the
Geodha. It is likely that further cave systems exist in
the vicinity; work on one of the caves discussed in
this report (Wetweather Cave) brought to light a
partially collapsed passage at its rear (see below).
The formation of cave deposits is a dynamic and
complex interaction between anthropogenic and
biogenic agencies. The type of sediment inside a cave
depends on access to external sediment carriers such
as wind, water, people and animals. Endogenous
sediment sources such as roof-fall (rock fragments
and fine particles), re-precipitated limestones (speleo
-
thems, crusts, flowstone, stalagmites, stalactites,
tufas/traverine), residual minerals left over after
solution of carbonates (silicas, metal compounds and
clays) and fluvial deposits. The importance of human
activities in the formation of such deposits has been
underlined (Butzer 1977, 79–85). External sources of
sediment can be represented by aeolian materials,
drainage into fissures, slope wash and colluvium,
and particles deposited by the high water reaching
the cave mouth.
Secondary complexities arise from post-
depositional alteration (diagenesis) of deposits such
as translocation of solid particles, bioturbation and
chemical weathering, in addition to the larger-scale
2
problems of erosion and collapse of the cave
structure (Courty et al 1989). Gleed-Owen identifies
the complexities in his attempt to reconstruct the
marine history of Smoo Cave including the issues of,
for example, small-scale fluctuations of sea level,
which even over a relatively short period of time can
lead to the repeated reworking of the marine sands
present (Gleed-Owen 1992).
2.2 Project background
When Smoo Cave was investigated in 1992 at the
behest of Highland Council, both Glassknapper’s
Cave and Antler Cave were observed to contain
deeply stratified archaeological deposits. In order to
avoid confusion with the better known cavern, long
ago named Smoo Cave (from the Old Norse smuga,
3
Glasgow Edinburgh
Durness & Smoo Cave
SMOO CAVE
near Durness, Sutherland
Leirinbeg
Antler Cave
Glassknapper's Cave
Smoo
Cave
Wetweather Cave
Geodha
Smoo
A'Ghoil - Sgeir
Sgeir Leathan
SANGOBEG
Allt Smoo
Durness
A838
Plan & Section Location
Remnants of
Stone Dyke/Wall
A
B
SECTION A-B (see Fig. 3)
Water's edge at
time of Survey
Lip of Adit
Edge of Midden
Allt Smoo
Outflow
To other Caves
Talus
Boulders
Dark Brown Humic
Soil with Shells
Rock Pillar
Outflow
Talus
Footbridge/Jetty
010m
Illus 1 Site location
meaning ‘rift’, ‘cleft’ or ‘cave’; Fraser 1995), the two
smaller caves were given names of their own during
the excavation. The presence of fragments of modern
glass, which at first appeared to have been knapped
like flint, in the upper deposits of the southern cave
provided the inspiration for the name Glass
-
knapper’s Cave. The discovery, again early on in the
excavation, of antler tines in the deposits in the
northern cave earned it the title Antler Cave. Bad
weather at the time of the excavation made work in
both of these caves dangerous, as the trenches were
deep and the sections easily weakened by rain water
and melted snow draining into the caves. In order to
maximize working time during these periods of wet
weather, attention turned to a fourth cave on the
opposite side of the inlet, where conditions were less
hazardous; hence its name, Wetweather Cave.
The mouths of both Glassknapper’s and Antler
Caves, which open onto a shingle beach, were
partially blocked by midden-rich archaeological
deposits, in places over 2m deep (Illus 2). These
deposits were vulnerable to erosion by breaking
waves at times of high tide, with quantities of
collapsed material, including animal bones, sitting
on the beach at the base of the exposed archaeolog
-
ical deposits. These caves, like the parent cave,
appear to have undergone considerable collapse
through time, attested by fragments of limestone in
the eroding deposits. The presence of collapsed
midden material on the beach below the cave mouths
attested to more recent erosion. As in Smoo Cave,
these deposits were suffering from active erosion,
but here caused by wave action at times of especially
high tide as opposed to the action of the river.
The fourth of the caves (Wetweather Cave) is in the
eastern wall of the inlet, where it curves to the
south-west to meet the entrance of Smoo Cave (Illus
1). The cave is at the top of a steep, grass-covered
slope, which may itself have resulted from cave roof
collapse, where it meets the present cliff face around
15m above sea level.
Wetweather Cave differs from the other caves not
only in its elevated position, but also in its dimen
-
sions and appearance. Where the other caves have
high ceilings and are longer than they are wide,
Wetweather Cave has a relatively low ceiling
(c 2.5m) and is much wider than it is deep. It is
perhaps better described as an overhang or rock-
shelter than a cave. However, closer inspection
proved that there was much more to the Wetweather
Cave than first appeared; at its eastern edge the
back wall gives way to a calcified fan of talus, above
which is a small, circular chamber. Further to the
east, a narrow gap allows access to a long, narrow
chamber filled with collapsed rock to the extent that
it is not possible to stand up once inside. It is possible
that this passage was at some time joined to a
complex series of chambers and tunnels like those
known to exist elsewhere in north-western Suther
-
land (Lawson 1988). The successive collapse of the
cave roof has concealed any further continuation of
the passage.
4
ANTLER CAVE
(AC)
GLASSKNAPPER'S CAVE (GC)
EH
G
F
C
D
Trench 3
Trench 2
External Section
Column Section
03m
Internal Section
Trench 1
Modern Beach
GEODHA SMOO CAVES
by Durness, Sutherland
Illus 2 Site plan of Glassknapper’s Cave and Antler Cave
5
Wetweather Cave also differs from its neighbours
in that archaeological deposits were not obvious from
the outset. The floor of the cave was covered by a
carpet of sheep dung and, in the absence of erosion by
either river or sea, there was no exposed anthropo
-
genic material. The site was initially investigated
because of its proximity to caves known to contain
archaeological deposits and also because its higher
position and appearance as a sheltered overhang
struck the excavation team, correctly as it happened,
as a likely place for prehistoric activity.
From the outset it was obvious that the deposits
in Smoo Cave and its neighbours had high archaeo
-
logical potential. Shell midden deposits are well
known for promoting good preservation and this
factor, along with the apparent complexity of the
deposits, suggested that excavation would provide a
valuable insight into past lifeways in this part of
coastal Scotland. This research potential aside, it
was also obvious that the deposits were suffering
badly from marine and riverine erosion. It was in
recognition of this last factor that a programme of
rescue excavation was developed, with the investi
-
gation and recording of the archaeological deposits
prior to their total destruction being the highest
priority.
As no formal archaeological recording or investiga-
tion of the Smoo Caves had taken place prior to the
work reported here it was essential that a survey of
the caves and their deposits took place before exca-
vation commenced. All of the caves were subject to
instrument survey and the apparent extent of the
archaeological deposits was mapped as part of this
exercise.
The first programme of work, in 1992, involved the
cutting back and recording of the deposits in Smoo
Cave, in advance of the revetting work which would,
it was hoped, protect the deposits from further
erosion. This work was carried out to a brief provided
by the Highland Archaeology Service and involved
minimal intervention rewarded with informative
results. The protection of the Smoo deposits is also a
cause of public interest, as the site is visited by
thousands of people every year and is an important
tourist attraction in north-western Scotland. The
impact of visitors walking over the deposits as they
enter the cave is a further cause of erosion, with
walkers exposing deposits on the footpath surface
and the pressure exerted through their body weight
pushing the deposits outward through the exposed
erosion face and thus speeding up the process of
section collapse. It was hoped that this additional
problem would be at least temporarily circumvented
through the construction of the revetment and the
laying of gravel over the portion of midden surface
used as a footpath.
In the case of Glassknapper’s Cave and Antler
Cave, it was recognized that protective measures
such as those adopted in Smoo Cave were inappro
-
priate. Here the main priority was to investigate and
record the deposits prior to their total and imminent
destruction through marine erosion. The main aims
and objectives were set out in a brief provided by
Historic Scotland. It was hoped that intensive exca
-
vation of these caves would provide a more detailed
insight into the history of human use of the caves
than that afforded by the limited programme of
recording carried out in Smoo Cave, though this
exercise did result in the recovery of an important set
of data. Excavation was geared to the recovery of
information on the economic function of the caves
and the period over which activities took place. The
excavation of deposits in the cave interiors, away
from the exterior exposed face, would also allow an
understanding of depositional process and pattern-
ing not possible in the more limited cutting back of
the Smoo Cave section.
The investigation of the Wetweather Cave was not
part of the Historic Scotland brief and is best under-
stood as an assessment of archaeological potential of
a site discovered during the survey of the other caves
and their environs and carried out at times when
work in these other sites was prevented by danger-
ous weather conditions.
More specific objectives and the methodologies
adopted in order to achieve them are discussed in the
relevant sections of the report.
Terminology for the chronological periods repre
-
sented by the excavated remains follows that set out
previously (Batey & Graham-Campbell 1998, 2):
Viking and/or Norse are used interchangeably to
refer to the period from the late eighth to 11th
centuries, while Late Norse is understood to have
begun around
AD 1050.
3 Smoo Cave
The archaeological investigation in Smoo Cave
involved the cutting back, sampling and recording of
an eroding section around 17.5m long prior to the
construction of a protective wall along it, and was
carried out over four days in March 1992. The
construction of a causeway (Illus 1) to allow visitors
to walk from the main cave to the smaller, adjoining
chamber had inadvertently diverted the course of
the Allt Smoo. At times of heavy rain or snow melt,
the obstruction created by the causeway caused the
river to flow directly alongside archaeological
deposits lying near the cave’s mouth at its eastern
side. The erosive action of the river exposed these
deposits and, over several years, washed away a
considerable quantity of material. In an attempt to
curtail the erosion, a revetting wall was to be
constructed against the face of the archaeological
deposits. In order to ensure that the wall sat flush
against the deposits, they were to be cut back along
several planes, a process which provided the oppor-
tunity to examine their nature and remove samples
for further analysis.
3.1 Background and methodology
The first aim of the work was to carry out a theodolite
survey of the main cave, recording the location and
visible extent of the archaeological deposits on the
plan produced (Illus 1). In order to understand the
nature and extent of the deposits, the eroding face
was first cleaned and drawn. This initial section
drawing is not reproduced here, as in many respects
it closely resembles the final drawing. However,
several ephemeral features present in the initial
section did not survive a further trowelling. These
included a cut pit (015), which may have been a
post-hole, and the remnants of a possible hearth
(012).
After recording the eroding section, the process of
cutting back the face began. In order to accommodate
the revetting wall, while at the same time removing
as little material as possible, the section face was cut
back on three planes, shown on Illus 1.
The Highland Region Sites and Monuments
Record gave the dimensions of the midden as 3m in
diameter by 0.2m high. Cursory examination of the
site soon established that this was a considerable
underestimate of the scale of the deposits. The
exposed midden face was found to extend south
-
wards into the cave for some 17m, with deposits
possibly several metres in depth stretching back for
at least 8m towards the eastern wall of the cave.
The area around a natural pillar in the north-
eastern part of the cave and a small recess in the
northern curve of the cave wall (Illus 1) were
occupied by a flat-topped mound which rose for well
over a metre above the top of the eroding section.
Marine shells were observed eroding from the humic
soil which covered this mound, and a sondage rapidly
excavated on its top and western slope established
that archaeological deposits, consisting of shells and
concentrations of charcoal, lay about 0.30m below
the surface. It also became apparent, after initial
cleaning back of the section, that the shell midden
was not the only evidence for archaeological activity
inside the cave; deposits below the shell midden
provided convincing evidence for several occupation
horizons.
Time constraints did not permit an assessment of
the deposits’ full extent as this would have involved
extensive trial trenching. The presence of marine
shells eroding onto the cave floor seemed at first to
suggest that the deposits roughly corresponded with
the extent of the mound. However, limited excava-
tion later revealed that earlier archaeological
deposits are stratified beneath the present floor of
the cave, and these may extend beyond the edges of
the mound.
It was immediately apparent that the threatened
section was in a seriously denuded state. Recently
collapsed portions of the exposed face covered the
narrow pebble beach, which at the time separated
the midden from the river. However, during the
course of the work, with several days of only
moderate rain, the water level rose dramatically. At
its highest point the river was observed flowing
along the base of the section, totally submerging the
pebble beach.
3.2 The Smoo Cave section
The following discussion of the section investigated
in Smoo Cave moves from the latest deposits to the
earliest; Illus 3 illustrates the section.
The upper layers in the section (contexts 006a and
006b) consisted of large numbers of shells in a matrix
of dark, organic-rich sandy soil. The uppermost layer
(006a) comprised shells (mostly limpets and winkles)
in a dark brown sandy matrix and was c 0.2–0.3m
deep for most of its length. However, towards the
northern end of the section (Illus 3, point A), this
deposit was considerably deeper, and in the initial
section there was some suggestion of a cut pit which
had been filled with shells. Again, this feature did
not survive more than the initial clean, but after the
section had been cut back the northern side of this
cut was still evident and can be seen in Illus 3, where
the upper deposit (006a) rises to the surface.
6
Underlying the deposit (006a) was another shell
midden layer (006b), which on initial cleaning
appeared to be the same as context 006a but proved to
contain more charcoal and represented an earlier
phase of deposition. A thin, charcoal-rich layer (027),
devoid of shells, lay between the two midden deposits.
The upper shell midden layer (006a) terminated
towards the southern end of the section, giving way
to a deposit of dark greasy soil (020), containing
butchered animal bone, which lay in a shallow cut
(019). This was overlain by a thin spread of a similar
soil with a higher charcoal content (025).
Beneath the lower shell midden layer (006b) was a
deposit of reasonably clean yellow sand (008) which
may have been deposited by wave action during
marine inundation of the cave some time in the past.
Such inundations are still a regular occurrence, with
northerly winds and spring tides carrying waves into
the back of the cave. It is difficult to establish when
this inundation took place or how long it lasted.
Immediately below the marine sand (008), though
not visible throughout the entire section, was a thin
band of grey sand (009), which may have been
stained as a result of human activity.
7
Bulk Sample
Sondage
A
Red / Brown
Staining
Sample
Column
Sondage
Bulk Sample
?Cut
019
D
Bulk Sample
Cut
022
017
010
026
020 025
020
?006b
008
010
017
?006b
026
028
Stone
006a
006b
008
009
010
017
023
024
002/006b
001
006a
009 008
010
017
023
024
027
B
Key
Charcoal
Shell Midden
Stained Sand
Clean Sand
SMOO CAVE
Section through the Midden & Occupation Deposits
(cleaned back to the line of the Revetment Wall - see Fig. 1)
02m
Illus 3 Smoo Cave: section through the midden and occupation deposits
8
A hearth or firepit (021), extremely rich in char
-
coal, lay towards the southern end of the section.
Birch and hazel charcoal from this feature provided
a radiocarbon date which falls within the range cal
AD 780–1020 (GU–4545) at the 2-sigma level of confi
-
dence. The charcoal sat in a cut (022) in the marine
sand (008). It thus appears to represent activity on
the waterborne sandy surface (008) just before the
shell midden (006b) began to accumulate. The thin
bands of washed sand – including context 026,
sandwiched between the lower shell midden layer
(006b) and the charcoal-rich layer (020) – are
perhaps more likely to represent riverine deposition
(by the Allt Smoo) than marine inundation.
Lying beneath the marine sand (008) and the grey
sand (009) was a deposit of water-rolled and fract
-
ured stones with some gravel and pebbles (context
010). Most of these stones probably represent a
natural beach surface, with the stones smoothed and
rounded by the action of the waves. However, this
beach surface had seen intensive human activity, as
evidenced by bands of dark humic sandy soil and
frequent charcoal. Discolouration caused by human
occupation was most obvious in the upper part of the
beach deposit (010), but there were also clear archae
-
ological horizons in its lower levels. Artefactual
material, including butchered bone and possibly
struck stones, was found throughout it. Marine
shells were also present, but in nowhere near the
numbers present in the shell midden (006).
Some of the larger stones in the beach deposit
(010), including Boulder 028, may represent struc-
tural features, but it was not possible to substantiate
this from the section alone. However, it is interesting
to note that the possible post-hole (context 016),
which was visible only in the initial section and is not
illustrated, actually sat next to Boulder 028, which
only became visible once the section was cut back.
Their proximity strengthens the argument that the
boulder represented a structural element, perhaps
part of a wall, and helped to brace the post that stood
in Post-hole 016.
The lower levels of beach deposit (010) merged
with a substantial deposit of laminated sands (017),
again representing successive episodes of marine
inundation. The upper levels of these sands were
heavily stained and contained both charcoal and
butchered animal bone, once more providing evi
-
dence for human activity.
In order to establish the full depth of deposits, two
small sondages were dug along the base of the
section (see Illus 3). These revealed a layer of clean,
washed sand (context 023) 0.65–0.8m deep, lying
below stained sand (context 017) and representing a
period when this part of the cave saw no human
activity. Excavation of these sondages stopped
where a surface was encountered that provided
evidence for an earlier phase of human activity on
the site; time did not permit excavation below this
level. On the surface of a layer of heavily stained
sand (024) lay pieces of quartz, possibly chipped by
human action, butchered animal bones and flecks of
charcoal.
3.3 Summary and interpretation
The cleaning back and recording of the eroding
section revealed considerable evidence for sequen
-
tial phases of human activity.
The charcoal-stained, artefact-littered surface
(024) encountered in the sondages provides evidence
for the earliest known human activity in the cave
(Phase 1), at a time when the cave floor was substan
-
tially lower than today. Unless the sea level was
considerably lower than at present, marine inunda
-
tion would then have been a more common
occurrence than it is today. The thick layer of washed
sand (023) sealing the surface was deposited over an
unknown period of time by these incursions of the
sea. Without dates or diagnostic artefactual material
it is not possible to say when or over how long a
period this sand accumulated. The absence of lami
-
nation within this deposit and the lack of identifiable
soil horizons may suggest a single event such as the
‘tsunami’ type episode thought to be responsible for
sand deposits investigated at Wick (Dawson & Smith
1997).
Following the inundation that deposited the sand,
the cave saw another phase of human activity (Phase
2), evidenced by the charcoal and animal bone in a
heavily stained sandy matrix (017). Subsequently, a
stony beach deposit (010) formed during further
human activity (Phase 3), which left behind
abundant charcoal, animal bone and possible stone
structural elements. After a further phase of inunda
-
tion (context 008), a hearth was constructed (Phase
4). This was sealed by the charcoal-rich shell midden
(006b) which was then overlain by another shell
midden (006a). For the purposes of this report both
phases of shell midden have been grouped together
as Phase 5. The radiocarbon date from the hearth
immediately below the lower shell midden deposit
provides a terminus post quem of cal
AD 780–1020 for
its formation. Activity related to the shell midden
may have continued into the medieval period. The
final phase of activity was marked by the upper
horizon (025), which consisted of dark greasy soil
with a high charcoal content.
The presence of six main archaeological horizons is
certainly a simplified breakdown of the stratig
-
raphy. As became obvious with the later phase of
work in the other caves, the real picture is likely to be
a great deal more complicated, with various layers
separated by thin lenses of both clean and stained
sands, the discolouration of the latter probably the
result of human activity. With only the section
recorded, it is very difficult to offer a more complex
interpretation of these deposits.
4 Glassknapper’s Cave, Antler Cave and
Wetweather Cave
4.1 Background and methodology
During the examination of midden deposits in Smoo
Cave reported above, archaeological deposits,
including midden material, were also noted in the
two marine caves in the western wall of the inlet and
observed to be vulnerable to serious erosion caused
by high tides and storm-driven waves. Conse
-
quently, a grant was provided by Historic Scotland to
enable the rescue excavation and recording of these
deposits before they were entirely lost to the sea.
This work was carried out over four weeks in
February and March 1995.
By the time of the excavation the deposits had
already suffered heavily from marine erosion, with
between 1 and 2m of deposit taken away between the
Smoo Cave investigation in 1992 and the commence-
ment of excavation in 1995. Given the rescue
motivation of the work and the limitations on budget
and time available, a pragmatic approach was
adopted and the main objectives were set out by
Historic Scotland. It was proposed that a sample
assessment of between 30 and 50% of the deposits
would establish the depth of midden, extent of
midden, nature and date of stratigraphy forming the
midden and the nature of any internal structure
within the midden (R Hingley, pers comm).
Prior to excavation, the exposed section (Illus 2,C–
D) was only partly visible behind a loose slope of
earth and stones which had collapsed from the
section face. Much of this material had accumulated
since the site was first identified in 1992, clearly
indicating the rapid rate of erosion. In order to obtain
an impression of the nature and depth of the deposits
it was necessary to remove this material. However,
in case it contained residual in-situ deposits, two
slots were first excavated through it. This controlled
removal and examination of the sections confirmed
that it entirely comprised loose material that had
fallen from the section face.
The section face was cleaned by trowel and
recorded by measured drawing. It was immediately
obvious that there were differences between the
deposits in the southern cave and those in the
northern cave (Illus 2, C–D). The southern cave
(Glassknapper’s Cave) appeared to contain a far
more complex series of deposits, which included
several strata rich in marine shells and animal
bones. The deposits in the northern cave (Antler
Cave) were less well-defined and varied, with
marine shells at this stage visible only at one level.
The collapsed material in front of Antler Cave
proved to contain fragments of red deer antler, which
were also observed in this part of the cleaned-back
section.
The presence of substantial fragments of lime
-
stone in various parts of the section indicated that
the roofs of both caves had suffered collapse at some
point in the past. The caves may therefore have been
somewhat larger than they are now, which would
have made them more fitting for human use than
they appear today. Nevertheless, the presence of
tractor batteries, boating equipment and even a
length of Scalextric track in the southern cave
clearly indicated its use an equipment store and
dumping ground in recent times.
After surveying the cave interiors (Illus 2), it
became apparent that the most efficient means of
fulfilling the excavation objectives would be to cut a
single trench from the exposed section to the back of
Glassknapper’s Cave. The same would also be
attempted for the Antler Cave, but priority was given
to the southern cave, as it appeared to contain more
complex archaeological deposits. The cave floor was
divided in two roughly along its central axis (Illus 2).
By the end of the excavation, most of the material had
been removed from the southern half, while the
northern portion remained intact, proviHding a full
section through the deposits. A lateral section was cut
through the southern half of the cave (Illus 2,G–H),at
right angles to the axial section, thus providing
insight into the nature of the stratigraphy in a north/
south plane (across the cave interior) as well as in an
east/west plane from cave mouth to cave interior
(Illus 2, E–F). The deposits were excavated in spits
down onto the former beach surface, at which point
the concreted nature of the gravels and safety consid
-
erations prevented further investigation.
An important aim of the project was to recover bulk
samples from the excavated deposits, as coastal
deposits rich in marine shells represent a rare oppor
-
tunity to recover well-preserved faunal, palaeo
-
botanical and organic artefactual evidence. Where
possible, samples were removed from individual
deposits. However, the stratigraphy in Glass
-
knapper’s Cave was of such complexity that sampling
individual contexts was not always possible. In order
to overcome this problem, a column sample was taken
through the deposits, with samples removed in spits.
Wet-sieving of samples was carried out on site.
4.2 Glassknapper’s Cave
Glassknapper’s Cave displayed the most extensive
and complex series of archaeological deposits. The
9
external section (Illus 2) contained a considerable
amount of overburden in its upper portion, with a
gritty deposit (001) overlying a black humic layer
that contained many small fragments of quartz and
other stone (003). The upper deposit (001) contained
sherds of bottle glass, some of which, on first exami
-
nation, gave the appearance of having been modified
(they had not been), hence the name Glassknapper’s
Cave. The humic deposit (003) represented organic
soils washed down from the cliff face above and did
not extend far back into the cave.
An earlier episode of largely natural build up
(context 004) was evident directly beneath the humic
deposit (003). The presence of angular fragments of
limestone in it suggested that this deposit (context
004) was at least partially composed of cave roof
collapse. As in the case of the humic layer, the roof
collapse appeared to be limited to the area of the cave
mouth. However, it also contained rounded stones of
various types, which may have been driven there by
high tides and storm waves. Excavation of the trench
through the cave (Illus 2, E–F) revealed relatively
little difference between these upper deposits.
Although they were largely sterile, the presence of
the bottle glass and a number of brown-glazed
pottery sherds suggests they accumulated during
the 19th and early 20th centuries, although a piece of
White Gritty Ware was also recovered from context
004.
These upper layers sealed a series of deposits rich
in archaeological material, possibly accumulated
over a long period of time. Evidence for this human
activity was clearly visible within the eroding
section (Illus 2, C–D), where deposits of marine
shells and animal bone were visible throughout the
lower half. The presence of midden-rich layers strati
-
fied between washed sand layers suggested the
periodic use of the site interspersed by times when
high water levels, perhaps promoted by spring tides
or winter storms, washed marine sands into the
cave. However, some of these thin, clean sand
deposits may have been laid by those using the cave,
perhaps to minimize dampness or to cover
unpleasant organic deposits. The most substantial
sand deposits lay toward the rear of the cave, where
they were deposited by marine action before the high
concentrations of archaeological material
accumulated.
Excavation and recording of both the main
internal cave section (Illus 2, E–F) and the internal
lateral section (Illus 4; Illus 5, H–G) revealed a
complex sequence of deposits, which bore only
limited resemblance to those observed in the
external, eroding section. The internal deposits
(contexts 008, 012 & 013) on the whole consisted of
numerous thin layers and lenses of organic material,
clays and silts, ash, charcoal, crushed shell, animal
bones and washed deposits. It was impossible to
excavate each of these deposits individually, with
many hundreds of individual contexts being strati-
fied within the deposits. In order to overcome this
problem, a column sample was removed from the
deepest portion of the deposits (see Illus 2 and Illus 4
for location), with samples bagged in approximately
0.02m spits. Environmental analysis of these
samples has revealed a wide variety of plant remains
(see Section 7.5 – Plant remains) with recovered
charcoal providing three radiocarbon dates (Table
1). The latest of these was from Spit 2 (toward the top
of the column and consisting of birch and willow),
with the date range being cal
AD 890–1160
(OxA-8210); the second was from Spit 15 (middle of
column, birch) and gave a range of cal
AD 770–980
(OxA-8211); the third came from Spit 33 (bottom of
column, hazel) and gave a range of cal
AD 820–1000
(OxA-8212); all dates are expressed at the 2-sigma
level of confidence (or 95.4%). These relatively
closely spaced dates clearly indicate quite rapid
accumulation of considerable quantities of material,
with the 0.95m depth of the column sample forming
in perhaps 100–150 years.
Excavation continued down through the tightly
stratified deposits within the cave, onto what
appeared to be the cave’s primary floor, character
-
ized by hard-packed, water-rolled stones. While
these appeared to represent a beach surface, it is not
possible to state for certain that earlier archaeolog
-
ical deposits were not sealed beneath; marine shells
10
Table 1 Radiocarbon determinations from Smoo Cave and
Glassknapper’s Cave (GKC). Dates have been calibrated using OxCal v3.8
Lab code Sample material Lab age ¶13C% Calibrated dates
1-sigma 2-sigma
GU-4545 Birch and hazel from Hearth
021 (Smoo Cave)
1120 ± 50 –27.1
AD 880–1000 (68.2%) AD 780–1020 (95.4%)
OxA-8210 Birch and willow from Spit 2
(near top of column sample,
GKC – see section drawing)
Context 008
1030 ± 40 –25.7
AD 900–920 (2.6%)
AD 970–1040 (65.6%)
AD 890–930 (8.4%)
AD 940–1050 (78.3%)
AD 1090–1160 (8.7%)
OxA-8211 Birch from Spit 15 (middle of
GKC column – see section
drawing)
1160 ± 35 –27.1
AD 780–800 (5.5%)
AD 810–900 (38.5%)
AD 910–960 (24.1%)
AD 770–980 (95.4%)
OxA-8212 Hazel from Spit 33 (bottom
of GKC column)
1120 ± 30 –25.7
AD 890–980 (68.2%) AD 820–840 (1.1%)
AD 860–1000 (94.3%)
and animal bones were found intermixed with the
loose beach gravels at the base of the external
section. Unfortunately, the hard-packed nature of
the basal deposit in Glassknapper’s Cave and the
obvious safety risks involved in digging the trench
any deeper made it impossible to establish the
presence or absence of earlier deposits.
There was no convincing evidence for substantial
structural elements in Glassknapper’s Cave, although
two concentrations of stones appeared to represent
artificial arrangements (not illustrated). The first of
these (context 038) lay toward the rear of the cave and
comprised a tightly packed layer of limestone chunks
and water-rolled stones, the latter probably collected
from the beach. The rear portion sat just beneath the
modern surface but dipped down towards the mouth of
the cave, following the contour of the sand deposit
beneath it. The purpose of the stone concentration was
unclear, but it did contain a beach pebble hammer
-
stone and several sherds of wheel-thrown, medieval
pottery. The absence of any trace of this feature in the
section drawing (Illus 4) suggests it was confined to the
southern half of the cave.
The only other possible structural element
consisted of several large, angular chunks of lime
-
stone (018) stratified well within (approx 1.1m from
the surface) the complex cultural deposits (008)
observed just inside the cave mouth. These appeared
to have been set into the underlying deposits and
may represent an attempt to cordon off the mouth of
the cave. However, as in the case of the stones (038)
toward the rear of the cave, the concentration
exhibited little regularity and may simply have been
the result of roof collapse.
In the external section, a brown silty layer con
-
taining limestone fragments (012) appeared at first
to comprise only roof collapse; investigation of the
layer farther into the cave, however, found concen
-
trations of winkles, mussels and limpets as well as a
scallop shell. The midden deposits (013) directly
beneath were very loose, and in places simply
consisted of bones and shells lodged in the gaps
between fragments of limestone and other rocks. In
the interior, however, deposits were on the whole
highly compacted and stone-free, apparently repre
-
senting areas of trample, burning and other
11
Modern Cave Floor
External
Section
CD
Beach Deposits
See Colour Plate Inset
Key
Sand
Silty Clay
Clay & Sand
Pebbles & Sand
Silty Overburden
Grey Silt
Charcoal Rich
Black Organic
Grey/Black Organic Midden
Grey Midden
General Midden
Grey/Black Charcoal-rich Midden
Roof Collapse Midden
Marine Shells
Glazed Potsherd
Column sample
from which C14 came
Internal Section
EF
Level of Excavation
Rock
Outcrop
GLASSKNAPPER'S CAVE
ANTLER CAVE
External Section CD
0
1m
01m
Illus 4 Smoo Cave
activities. In order to clarify the nature of the
deposits in the northern portion of the external
section, a slot trench was cut back into the section
(Illus 2, Trench 2), just east of the rock outcrop
between Glassknapper’s Cave and Antler Cave. The
slot trench was cut back to the rock face that
separates the entrances to the caves.
Archaeological material was present throughout
the lower deposits (013) in this slot trench, those
above representing the same process of silting and
collapse observed elsewhere. However, the shells
and bones were not present in distinct and compact
layers, as in the interior, but on the whole were
mixed with the rubble and stone, though in places
higher concentrations of midden material existed
independent of stone accumulation. A sheep’s skull
was recovered from the niche created by the outcrop
and its juncture with the rock face, in which various
other bones and shells had also lodged.
The appearance of water-rolled stones and
limestone fragments (015) in this deposit suggests
that both roof collapse and marine action had
contributed to its formation. Limestone fragments,
indicating roof collapse around the cave’s mouth,
were generally confined to the front part of the cave.
The presence of both water-rolled stones and
washed sands within the midden deposits provides
evidence for the complex nature of the processes of
marine inundation and beach formation. Today the
upper beach in front of the caves is composed of
water-rolled stones, with sand only visible further
down the beach at low tide. The dynamics of wave
action and beach morphology must be studied in
greater detail before the implications of the appear-
ance inside the cave of both types of beach deposit,
usually mutually exclusive, can be fully understood.
The loose midden material (013) identified in
Trench 2 probably represents refuse removed from
the cave interior and dumped into a semi-confined
space otherwise occupied only by tumbled and
wave-deposited stones. As this area was not subject
to trampling, the deposits did not take on the
compacted, greasy consistency of those inside the
cave, each of which at some time in their history
appear to have formed its floor.
4.3 Antler Cave
Archaeologically speaking, Antler Cave did not
prove as productive as Glassknapper’s Cave. The
relative paucity of archaeological deposits may in
part be due to the possibility that, as far as human
activity is concerned, this cave has always been the
damper cousin of its deeper and drier neighbour.
However, this is not to say that archaeological
deposits were totally lacking, and it is important to
note that the consistently wet conditions regularly
caused the sections to collapse and thus made it
impossible to excavate as extensively as in Glass
-
12
GH
Key
Black Midden
Charcoal-rich
Deposits
Grey/Brown
Midden with Shells
Brown Midden
Pebble Beach
Clear Sand
Greasy Brown Deposit
Fish Bone Concentration
Cave Wall
030cm
Illus 5 Detail of section G–H
knapper’s Cave. It cannot therefore be stated for
certain that more complex deposits, similar to those
in Glassknapper’s Cave, do not exist within the
largely unexplored body of the cave deposits. Exca
-
vation of the Antler Cave deposits was limited to a
small slot trench cut back from the main section for a
little over a metre (Illus 2, Trench 3; sections not
illustrated).
The upper deposits were similar to those observed
in the front section of the Glassknapper’s Cave, with
the same sequence of silting and collapse forming the
upper horizons (contexts 022, 023 & 024 equating
with contexts 001, 003 & 004, respectively). A
number of distinct archaeological horizons were
detected further down in the section. These were
sealed beneath a considerable deposit (contexts 024/
026), around 0.40m thick, of limestone fragments
and chips, apparently from cave collapse. The first of
these archaeological horizons lay directly beneath
collapse (contexts 024/026) and consisted of a thin
layer of winkles and animal bone in an orange/brown
matrix (036). This overlay a less clearly defined
deposit (027), some 0.20m thick; it consisted of large
angular stones in a silty brown matrix, which had
shells, animal and fish bones scattered through it.
This in turn sealed a midden deposit (029) of shells
and fragments of charcoal in a silty grey matrix,
which again also contained angular stones. This
deposit did not appear to extend far back into the
cave, but of course it is impossible to say how much of
the deposit outside the cave’s present mouth has
been lost to erosion. This overlay a sterile layer of
orange silty clay with some stones (039), which itself
sealed a deposit of large angular stones, grit and
gravel (040). The lower limit of excavation was
marked by a deposit of very large angular chunks of
limestone with virtually no matrix (041), which
continued beneath the level of the present beach.
Limited excavation of the Antler Cave succeeded
in identifying a series of deposits related to past
human activity. Unlike the majority of archaeolog
-
ical deposits in the Glassknapper’s Cave, these
generally lay within rubble horizons rather than in
highly compacted lenses and layers. The deposits on
the whole were looser and less dense than most of
those in Glassknapper’s Cave. The cave appears to
have been used on a much more casual basis, with
features such as firespots and artefacts largely
absent.
4.4 Wetweather Cave
An investigation of Wetweather Cave was not
included within the original brief to carry out work in
the Smoo inlet caves that later became known as
Glassknapper’s Cave and Antler Cave. In fact, prior
to the project the presence of this cave was unknown.
The cave was identified during the general survey of
the inlet which accompanied the instrument survey
of the other caves. Although situated on much higher
ground and lacking the eroding sections that made
the presence of archaeological deposits obvious, the
cave seemed a likely candidate for past human
activity.
The original intention was to do nothing more with
the cave than include it on the survey. However, as
the project progressed, work began seriously to
suffer from deteriorating weather conditions. Melt
-
ing snow made conditions in both Glassknapper’s
Cave and Antler Cave extremely hazardous as the
deep strata became unstable. At times conditions
were too dangerous for work in the caves to continue.
The third cave did not suffer from water inundation
to the same extent, remaining dry and sheltered
from the worst of the weather. In order to make the
best of the time available it was decided to carry out a
limited evaluation of the cave, which became known
as Wetweather Cave, during periods when work in
the other caves was inadvisable.
Wetweather Cave consists of three elements. The
first of these is the outer chamber, which takes the
form of a deep overhang that opens out to the
north-west. The rear part of the chamber, closest to
the entrance to Smoo Cave, is occupied by a deposit of
talus and limestone concretion, behind which is a
small chamber into which it is possible to gain access
with relative ease. To the left of the entrance to this
small inner chamber is a third, much larger
chamber. However, gaining entry to this chamber is
possible only by crawling through a narrow gap,
which had been partially blocked by cave roof
collapse, with rubble extending into the chamber as
far as the eye could see.
Excavation of the Wetweather Cave was confined
to the outer chamber, where removal of several
centimetres of sheep dung revealed archaeological
deposits (Illus 6). The first feature to be identified
was a concentration of marine shells, dominated by
limpets (context 1/006), which also contained
butchered animal bones and a copper-alloy pin (SF
050). A number of cut features were identified with
further cleaning. These included several stakeholes
and possible post-holes (contexts 1/012 & 1/009),
which had been cut into the chalk-like soil (contexts
1/002 & 1/008) that covered the cave floor. This
highly mineralized deposit, which appears for the
most part to be formed from dissolved limestone,
was at first thought to be archaeologically sterile,
although it did have features cut into it. However,
cleaning back in spits revealed animal bones and, in
several locations, sherds of late Neolithic impressed
ware. A further shell midden deposit (1/015),
consisting largely of limpet shells, was identified in
the north-eastern part of the trench, lying in a
shallow scoop (1/023), again cut into the deposit of
degraded limestone (1/008) that covers the floor of
the cave.
As the time devoted to this cave was dictated by the
inability to work in the other caves, it was not
possible to achieve anything more than an assess
-
ment of the deposits. However, it does appear that
the cave was occupied as a far back as the late
Neolithic, with features of considerable complexity
13
cut into the floor of the cave. The copper-alloy pin
also indicates it was used in a more recent period. At
the end of the fieldwork, plastic was laid down and
the sheep dung deposit carefully reinstated, along
with the excavated spoil, in order to preserve this
potentially important archaeological site.
14
WETWEATHER CAVE
Plan
Key
Marine Shells
Stakeholes
1/015
1/015
1/007
1/002
1/004
1/009
1/012
1/008
1/006
1/002
Limit of
Excavation
Stones
1/014
1/016
1/004
1/008
1/006
1/002
Back Wall of Cave
0
1m
Illus 6 Site plan of Wetweather Cave
5 Material Culture
5.1 Pottery by Robert Squair
5.1.1 Methodology
The excavation of midden and other archaeological
deposits in Glassknapper’s Cave, Antler Cave and
Wetweather Cave recovered a small ceramic assem
-
blage of approximately 350 sherds.
To facilitate effective analysis, much of the assem
-
blage was washed to reveal more clearly the original
vessel surfaces and fracture profiles. Various aspects
of manufacture, morphology and decoration were
recorded for each vessel identified. The physical
condition of the sherds was also recorded to ensure a
fuller understanding of depositional practices and
post-depositional processes. Sherds from the assem
-
blage, with the exception of minute fragments, were
individually bagged to reduce further deterioration
through abrasion in storage.
5.1.2 General description of fabrics
The assemblage comprises approximately 350
sherds and fragments, weighing over 500g. The
largest sherd (Illus 7), from P6, is 107mm across its
maximum dimension. All vessels recognizable in the
assemblage, with the exception of P12, which has
disintegrated, are represented by a meagre number
of sherds, only some of which are conjoinable. The
eclectic ceramic styles represented in the assem
-
blage attest to the extensive chronological range and
diverse cultural origins of the surviving vessels.
A total of 11 fabrics were distinguished, largely by
superficial appearance, frequency, size, degree of
sorting and degree of rounding of the constituent
mineral and rock inclusions, using a hand lens with
10x magnification and a bright overhead light
source. No interpretative significance is necessarily
attached to the occurrence of different sherds,
evidently from separate vessels, in the same fabric.
Unfortunately, due to the small size of the predomi
-
nantly rock inclusions and the post-depositional
concretions on many sherds, it was impossible to
identify conclusively the different types of rock in the
recognizable fabrics. The overwhelming presence of
mineral and rock inclusions, supplemented on occa
-
sion by graminaceous inclusions in Fabrics 2 and 5,
suggests an original prejudice of raw material
selection during production.
It is unclear whether the vessels represented in the
assemblage were manufactured locally or imported
into the region from elsewhere. It is tempting to
envisage a local production source for P2 and P5,
manifest in Fabrics 2 and 5, respectively, each
containing graminaceous inclusion voids, but it is
preferable to interpret P9, manifest in Fabric 8, as
imported. The fabrics of P6, the possible Norse style,
and P12, the late Neolithic impressed ware, differ
substantially in terms of superficial appearance, if
not formal description, from the remaining ceramics.
5.1.3 Vessel catalogue
The following vessel catalogue provides a formal
description of each vessel represented in the
assemblage.
P1
Manufacture: The vessel is manifest in Fabric 1.
Diagonal striations on the interior surface, typical of
lifting, indicate that the vessel was probably wheel-
thrown. The interior and exterior surfaces were both
smoothed and slipped.
Morphology: The vessel, probably a moderately
sized jar, is represented by a substantial body sherd
(015/1) and a possible shoulder sherd (016/1). The
vessel evidently incorporated a subtle shoulder into
a neutral profile.
Function: Heavy sooting or macroscopic food resi-
dues are discernible on the vessel exterior.
P2
Manufacture: The vessel is manifest in Fabric 2.
The manufacturing method, surface treatments and
firing profile remain indeterminate.
Morphology: The vessel is represented by a solitary
diminutive sherd. The size and shape of the vessel
remain indeterminate.
Function: No use-related traces are identifiable on
the surviving sherd.
P3
Manufacture: The vessel is manifest in Fabric 3.
The manufacturing method is indeterminate. The
exterior surface was smoothed and probably slipped.
Morphology: The vessel is represented by three
base sherds (014/1, 017/1, 022/1), none of which are
conjoinable, despite the presence of fresh fracture
profiles on two of these sherds (014/1, 017/1).
Function: A glossy soot is discernible on the
exterior basal surface of the vessel. Abrasion on the
interior basal surface of the vessel is readily inter
-
preted as a consequence of attrition incurred during
use.
P4
Manufacture: The vessel is manifest in Fabric 4.
The manufacturing method remains indeterminate.
15
The interior and exterior surfaces were both
smoothed and probably slipped.
Morphology: The vessel, a relatively fine ware, is
represented by two body sherds (023/1, 025/1). The
size and shape of the vessel remain indeterminate.
Function: Possible sooting characterizes the
exterior surface of the surviving sherds.
P5
Manufacture: The vessel is manifest in Fabric 5.
The vessel was probably manufactured by coiling.
The interior and exterior surfaces were both smoo
-
thed and slipped.
Morphology: The vessel, a heavy-necked bipartite
bowl, is represented by a large carinated sherd (013/
16
GKC 006
GKC 008
GKC 010
GKC 011
GKC 012
GKC 013
GKC 014
050
P6
P12
05cm
Illus 7 Metal and pottery artefacts
1), a body sherd (018/1), recently broken into three
fragments, and a diminutive fragment (021/1). The
vessel incorporated a neck into a bipartite profile.
Function: Striations on the vessel exterior, concen
-
trated above the carination, are preferably interpreted
as the vestiges of a use-related attrition pattern.
P6
Manufacture: The vessel is manifest in Fabric 6.
Coil corrugations, tangible on both the interior and
exterior surfaces, indicate that the vessel was manu
-
factured by coiling. The interior and exterior sur
-
faces were both wiped and slipped. A large finger
mark, incurred at the clayware stage of manufac
-
ture, occurs on the vessel exterior immediately below
the rim. The firing profile is uniformly dark.
Morphology: The vessel is represented by a
substantial rim sherd (1001/1) and a body sherd
(1000/1). The vessel, slightly inturned at the rim,
had a simple, flattened rim moulding, and a barrel-
shaped profile. The base, none of which survives, was
probably flat, although evidently narrower than the
width of the vessel body.
Function: Heavy sooting or macroscopic food
resides are discernible on the vessel exterior.
P7
Manufacture: The vessel is manifest in Fabric 7.
The manufacturing method remains indeterminate,
although the rim moulding was probably formed by
lateral joining. The interior and exterior surfaces
were smoothed.
Morphology: The vessel, a relatively fine ware, is
represented by a rim sherd (1002/1) and a diminu-
tive fragment (1007/1), broken subsequently into
two fragments. The rim form combines an external
expansion with an internal bevel. The vessel proba-
bly had a neutral profile.
Function: Abrasion, concentrated on the vessel
interior and not extending onto the internal bevel, is
readily interpreted as use-related attrition.
P8
Manufacture: The vessel, a relatively fine ware, is
manifest in Fabric 4. The vessel was probably
wheel-thrown. The interior and exterior surfaces
were smoothed and possibly slipped.
Morphology: The vessel is represented by two body
sherds (1003/1, 1008/1). The size and shape of the
vessel remain indeterminate.
Function: Possible sooting characterizes the vessel
exterior.
P9
Manufacture: The vessel is manifest in Fabric 8.
The manufacturing method remains indeterminate.
The interior and exterior surfaces were smoothed
and glazed, respectively.
Morphology: The vessel is represented by two
non-conjoinable fragments of a strap-handle (1004/
1, 1004/2). Presumably, the strap-handle was luted
on to the vessel exterior. The size and shape of the
vessel remain unknown.
Function: No use-related traces are identifiable on
the surviving sherds representing this vessel.
P10
Manufacture: The vessel is manifest in Fabric 9.
The vessel was wheel-thrown. The interior and
exterior surfaces were smoothed and slipped.
Morphology: The vessel, a fine closed bowl, is
represented by three shoulder or neck sherds (003/1,
004/1, 1005/1), none of which are conjoinable. The
vessel had a closed, globular profile.
Function: Possible sooting occurs on the vessel
exterior.
P11
Manufacture: The vessel, manifest in Fabric 10,
was probably wheel-thrown. The interior and
exterior surfaces, wiped and smoothed respectively,
were both slipped. The slip affords the vessel exterior
a lustrous appearance.
Morphology: The vessel, a relatively fine ware, is
represented by a solitary body sherd (1006/1). The
size and shape of the vessel remain indeterminate.
Function: No use-related evidence is discernible on
the surviving sherd representing the vessel.
P12
Manufacture: The vessel is manifest in Fabric 11.
An exposed internal building coil and lateral
fracture along sloping coil joins indicates that the
vessel was manufactured by coiling. The interior and
exterior surfaces were smoothed and burnished,
respectively. The firing profile, incorporating a dark
core, is typical of open firings.
Morphology: The vessel is represented by a rim
sherd, a neck sherd, several body sherds, fragments
from a detached cordon and innumerable diminutive
fragments (1010/1, 1011/1, 1012/1, 1013/1, 1014/1,
1014/2, 1015/, 1015/2, 1015/3). The rim is rolled, with
a convex rim surface. The cordon, presumably
attached in a horizontal alignment, was evidently
luted on to the body of the vessel. The vessel exterior
is decorated with whipped cord maggot impressions.
The individual maggot motifs, each aligned verti
-
cally, are arranged together into horizontal bands
around the vessel exterior. Interestingly, each suc
-
cessive row of maggot motifs is offset in relation to
the row above, affording the resultant decorative
structure an overall uniformity and consistency of
design. The vessel, evidently necked, probably had
an otherwise neutral profile.
Function: Unsurprisingly, given the prevalence
of limestone concretions, no use-related traces are
identifiable on the surviving sherds representing the
vessel.
5.1.4 Interpretation
Ceramic styles represented in the assemblage
The assemblage contains a diverse array of ceramic
styles. Unfortunately, almost nothing is known
17
about local ceramic styles, from the Iron Age onward,
in northern Scotland. The settlement archaeology,
extending from prehistory into the relatively recent
past, fails to inform upon contemporary ceramic
production and use (see Reid et al 1967). Norse and
medieval pottery in northern Scotland derive
primarily from excavations in Caithness and the
Northern Isles (see McCarthy & Brooks 1988, 208–
10). Unfortunately, the nature of the relevant assem
-
blages usually precludes the development of
coherent ceramic sequences (eg MacAskill 1982, 405;
Batey & Freeman 1986, 338). The meagre assem
-
blage from Geodha Smoo is no exception.
At any rate, despite this unsatisfactory if unavoid
-
able circumstance, P12 is interpreted as a late
Neolithic impressed ware; P6 is tentatively identi
-
fied as a Norse style and came from Glassknapper’s
Cave; P4, P7, P8, P9, P10 and P11 are immediately
recognizable as medieval wares; P1, P3 and P5 are
tentatively suggested as post-medieval wares. No
stylistic comparisons are offered for P2, a vessel with
a distinctive fabric, but represented only by a small,
entirely uninformative fragment. Given this dearth
of evidence, the following commentary is largely
provisional.
The prehistoric pottery P12, a necked vessel with a
neutral profile, a rolled rim and presumably a hori-
zontal cordon, is an impressed ware datable to the late
Neolithic and comes from Wetweather Cave (see Illus
7). The rolled rim, not unknown on such pottery, is
less usual than the thickened, internally bevelled
rims more typical of impressed wares from elsewhere
in Scotland (see McInnes 1964; Longworth 1967;
Cowie, forthcoming). The individual whipped cord
‘maggot’ motifs combine into a coherent decorative
structure. The decoration on impressed wares from
elsewhere in Scotland, particularly the south-west, is
usually more haphazard (see McInnes 1964, 50).
There is a paucity of comparable vessels in northern
Scotland. However, the decoration on an impressed
ware vessel from Allt Chrisal on Barra in the Western
Isles, comprising successive maggot motifs diagonally
aligned, has a comparable coherent structure (see
Gibson 1995, 110, illus 4.36, no 170:111). The inade
-
quacy of the concept of impressed ware, conveying a
misleading impression of categorical homogeneity
based on a decorative technique ubiquitous during
the late Neolithic and early Bronze Age, requires
mention (cf McInnes 1964, 49).
P1, P3 and P5, from Glassknapper’s Cave, are
tentatively ascribed an Iron Age date. P5, a heavy-
necked bipartite bowl probably manufactured by
coiling, deserves special mention. These sherds bear
some similarity in form and fabric to various Iron
Age assemblages from the Northern Isles, including
the multi-period site at Howe on Orkney (B Ballin-
Smith, pers comm).
The Norse pottery The Norse ceramics from, for
example, Freswick Castle and Freswick Links in
Caithness bear some affinity, both technological and
stylistic, with some of the vessels from Geodha Smoo
(see Batey et al 1984, 105–7, 115–16; Pollard 1996b,
20–1). Indeed, the appearance and form, if not the
fabric, of P6 (see Illus 7) explicitly recalls that of
Norse pottery from elsewhere in northern Scotland,
for example at Jarlshof, Shetland (see Hamilton
1956;cfMcCarthy & Brooks 1988, 208). Interest
-
ingly, the fabric of P2, probably grass-tempered,
recalls vaguely that of Norse or medieval pottery
from Freswick Castle and Freswick Links (Batey et
al 1984); Kirkwall, Orkney (MacAskill 1982, 405,
412); and Jarlshof (Hamilton 1956; McCarthy &
Brooks 1988, 208). Admittedly, grass-tempering is
an unreliable cultural or chronological indicator in
northern Scotland (eg MacAskill 1982, 405, 412).
The medieval pottery P4 and P8, each represented
only by uninformative body sherds, are broadly rec
-
ognizable as medieval. P7, with an externally
expanded, internally bevelled rim, is comparable, in
stylistic rather than technological terms, with
medieval vessels from Kirkwall, Orkney (see Mac
-
Askill 1982; McCarthy & Brooks 1988
, 208, nos 523–
4, figure 114:210). Similarly, P9, represented by two
non-conjoinable fragments of a strap-handle, is pre
-
sumably a Scottish White Gritty Ware jug and, as
such, is broadly paralleled at Kirkwall (MacAskill
1982, 407).
5.1.5 Depositional practices and
post-depositional processes
The nature of the assemblage from Geodha Smoo,
comprising only a few vessels sparsely represented
by diminutive and abraded sherds, suggests that the
pottery was casually discarded and subsequently
disturbed, prior to eventual incorporation into
archaeological deposits. The diversity of styles and
breadth of chronology exhibited by the assemblage
lend support to this interpretation.
The degree of sherd dispersal across contexts
suggest some degree of disturbance of the various
pottery-bearing deposits in Glassknapper’s Cave, a
factor which also appears to have played a part at An
Corran, Skye (A Saville, pers comm). Sherds repre
-
senting P1, P3, P4 and P5 derive exclusively from the
lower midden deposit (context 019) in Glass
-
knapper’s Cave. P2 and P11, each orphan sherds,
derive from a stony deposit within the midden (021)
and the lower layer of collapse (006), respectively,
again in Glassknapper’s Cave. P9 derives exclu
-
sively from tumble within Glassknapper’s Cave.
P12, represented by several sherds and many frag
-
ments, derives exclusively from the degraded
limestone (1/008) in Wetweather Cave.
Interestingly, the two sherds representing P6
derive from both the upper and lower midden
deposits (008 and 019, respectively) in Glass
-
knapper’s Cave. The two sherds representing P8
derive from the upper layer of loose collapse and the
upper midden deposit (005 and 008, respectively) in
18
Glassknapper’s Cave. The three sherds representing
P10 derive separately from a black humic layer, a silt
matrix with limestone fragments and a stone deposit
(contexts 003, 004 and 038, respectively, in Glass-
knapper’s Cave). Medieval sherd P7 came from
collapse (context 004).
5.1.6 Conclusion
The variety of ceramic styles represented in the
assemblage, encompassing vessels of late Neolithic
to medieval date, indicate that the caves in the
Geodha Smoo were a focus for sporadic activity, if not
continuous occupation, over several successive
periods in the past.
5.2 Bone and antler artefacts
by Tony Pollard (with species
identifications by Catherine Smith)
One of the most striking results of the Smoo excava
-
tions, and the investigation of Glassknapper’s Cave
in particular, has been the recovery of worked bone
and antler. The recovery of organic artefacts in
immaculate condition is largely due to a fortuitous
combination of damp and alkaline conditions, the
latter promoted by both the limestone geology and
the presence of concentrated marine shells.
These finds included a carved peg of red deer
antler (SF 003, Illus 8), with cut marks clearly visible
on its surface. This artefact was made on an antler
tang, with a carved cylindrical head topping a curved
and pointed shaft. Its function is uncertain but, in
keeping with other elements of the material culture
recovered from the site, may represent a piece of
ship’s furniture, perhaps an alternative form of
timber fastening to the iron nails discussed below
and sometimes referred to as tree nails. Alterna
-
tively, the piece may be a shroud-pin, having the
same general form as wooden examples recovered
from Hedeby Harbour in Denmark (Crumlin-Peder
-
sen 1997, 134). These are devices used for tightening
rigging, stays and shrouds, in order to fasten or
quickly release them to the side of the ship’s hull.
However, the Danish examples are considerably
larger than the piece from Smoo, and it seems
unlikely that the shaft would be anywhere near long
enough to perform this function adequately.
Other pieces of worked antler and bone took the
form of spatulate or pointed blades (SF 004, SF 005,
Illus 9; SF 009, Illus 8). The first of these (SF 004) is a
heavily worked piece of bone which has been cut and
shaped along its length to create several facets and a
multi-sided profile (six-sided at the narrow end and
seven-sided at the broad end). The function of this
piece is uncertain, although a bevel-ended piece of
antler (SF 005) bears some similarity to the bevel-
ended pieces common to western Scottish coastal
sites and may have been used for rubbing hides
(Pollard 1994).
Another spatulate piece of bone was perforated (SF
002, Illus 9) toward one end, which has been curved
through a series of small knife cuts. This may
represent either the handle of a knife or a netting
needle, but as the piece is snapped it is not possible to
say which is the most likely. The object bears some
similarity to a piece, though in antler rather than
bone, recovered from the Birsay Bay excavations
(Morris 1989, 196), although the long sides taper
rather than being parallel as they are on the shorter
piece from Glassknapper’s Cave.
Also recovered, from context 013, was a small,
finely worked bone pin (SF 003, Illus 8) with a round,
perforated head; this may be made of a pig bone.
The function of the majority of the antler and bone
artefacts is uncertain with direct parallels being
scarce. It has been noted elsewhere (Batey 1996) that
small organic assemblages do not necessarily
19
003 009 033
05cm
Illus 8 Bone artefacts
display culturally specific traits. In general, how
-
ever, the assemblage is not out of place with other
collections of bone and antler artefacts from coastal
sites in northern Scotland, many of which contain
Norse elements.
It is apparent that many early (19th-century)
descriptions of bones recovered from coastal midden
deposits as split or cut to remove the marrow may in
reality have referred to artefacts, or the waste
produced by their manufacture, rather than food
waste.
No antler or bone artefacts were identified during
the excavation of Antler Cave. A number of intact
antler tines were recovered from the lower deposits,
but their relationship to the finished antler artefacts
recovered from the neighbouring cave is unclear,
although it is tempting to suggest that the cave was
used for the storage of this raw material.
5.3 Metal objects
5.3.1 Iron nails
Iron nails were recovered from Smoo Cave and
Glassknapper’s Cave.
Smoo Cave Four nails (SC SF 009, 012, 014, 022)
were recovered from the shell midden deposit (006b)
in Smoo Cave, one (SC SF 030) came from context
020 and a further three (SC SF 002, SF 008, SF 021)
came from the tumble at the base of the section (see
Illus 10). One of these pieces, SC SF 009, had a
flattened, square head, but was snapped just below
the head. The majority had round or semi-rectan
-
gular heads and appeared to be standard handmade
nails. However, one piece (SC SF 002) had opposing
sub-circular heads on either end of the square-
sectioned shank. Closer inspection revealed that the
larger of the two heads was a separate plate, known
as a clench plate. With this in mind, closer inspection
of nail SC SF 009 suggests that this is the clench
plate end of a clench nail, as the beaten-down nail tip
is visible as a raised area on the surface of the plate.
Clench nails have a long tradition. They are a rela-
tively common feature on Norse and later coastal
sites in Scotland and are usually associated with
ships and boats.
The clench nail is really a precursor of the rivet,
and was used to hold a boat’s timbers together. The
rivet is a single-piece fastener with a head and
shank; when in position the tip of the shank is
flattened out to create what is in effect a second head,
thus holding timbers in place. The clench nail,
however, was driven through the timbers and then a
pre-holed clench plate placed over the end of the nail
and the protruding head beaten flat with a hammer,
thus holding the clench plate in place. In this way
two timbers were effectively bound together,
clenched firmly between the nail head and the plate.
The presence of a nail with the clench plate
attached suggests that the piece was removed from a
rotten or burned timber, as the clench plate would
only be added when the nail was actually used, and
removal intact would require destruction of the
timber. This may indicate that boat repair was
taking place on the site, as old timbers were removed
from vessels and then replaced, or alternatively that
old ship’s timbers provided fuel for fires.
Glassknapper’s Cave Seven further pieces were
recovered from Glassknapper’s Cave and in general
were larger and sturdier than the examples from
Smoo Cave. These came from a variety of contexts:
GKC SF 006, SF 010 and SF 012 came from context
008; GKC SF 008 came from context 021; GKC SF
011 and SF 013 came from context 012; and GKC SF
014 came from context 013. All apart from GKC SF
014 had clench plates, or fragment(s) thereof,
attached. On a couple of examples, preservation was
20
002 005 004
003
05cm
Illus 9 Bone artefacts
good enough to show clearly that the clench plate had
at one time been separate and the nail head had been
beaten out across its outer face. The nail heads in
general appear to have been circular or sub-circular,
while the clench plates were square in plan. How-
ever, in the majority of cases, it was difficult to
distinguish between the nail head and the clench
plate.
The association between clench nails and boat
construction is a strong one, but not one limited to a
specific period. A nail with clench plate attached was
recovered along with less well-preserved examples
from the excavation of Norse and other deposits in
Birsay Bay, Orkney, where the excavator notes that
it is difficult to tie this artefact type down to any one
period (Morris 1996, 92). Similar nails were also
found during excavations at Freswick Links, Caith
-
ness, where what are described as rivets are
presumably clench nails (Morris et al 1995). Both
Birsay and Freswick Links have strong Norse com
-
ponents and some of the nails recovered may relate
to the Norse periods of the sites’ use.
5.3.2 Copper-alloy pin
A single copper-alloy pin (Illus 7) was recovered from
a shell midden deposit in the Wetweather Cave.
The pin has a slightly bent shaft, 26mm long,
which was probably straight when new. The shaft is
tipped by a spherical head, some 2mm in diameter.
The head is decorated by a single incised line that
spirals around it.
Numerous examples of copper-alloy pins have
been recovered from Norse and later contexts in
Scotland, varying in style from the plain to the
highly decorated. The pin from Wetweather Cave
bears close similarity to two pins recovered during
Curle’s excavations at Freswick Links in Caithness.
The description of these pins (4.8.85–6 in Batey
1987, 117) states that they have heads of twisted
metal. Although numerous pins with twisted metal
heads were recovered from Freswick (eg 4.8.3 and
4.8.4 in Batey 1987, 466), the writer believes that the
photograph of pins 4.8.85–6 (Batey 1987, 467) shows
them to have solid cast heads with incised decoration
rather than heads of twisted metal. Unfortunately,
these objects, once held in a private collection in
Thurso, appear to have been lost (Batey 1987, 117)
and so are not available for examination.
Although the pin may be Norse, it is equally
possible that the object, which was probably used to
fasten garments or headdress, dates from a more
recent period, with twisted metal-headed pins
ranging in date anywhere between the 14th and 18th
centuries (Batey 1987, 144).
21
002
009 022 030
012 014 021008
05cm
Illus 10 Metal artefacts
22
6 Analysis of Slag and Fuel Samples from
Smoo Cave by E Photos-Jones
Three slag and fuel samples from Smoo Cave (all
from shell midden, contexts 006a/b) were subjected
to scientific investigation in order to establish their
nature and composition. The full report is included
in the archive report, and a summary of the results is
given here.
6.1 Methodology
Scientific examination was carried out with a
scanning electron microscope to which was attached
an energy dispersive analyser (SEM–EDAX). One
sample (SM2) was cut, mounted on metallographic
resin, ground with silicon carbide papers (Struers
260, 600, 800 and 1200) and polished with six-micron
diamond paste. A freshly fractured surface was cut
from the other two and mounted on a stub. All three
were carbon-coated in preparation for analysis (at
20kV and WD 39).
6.2 Results
SM2 This fragment of slag is brownish black, porous
and light. It is very inhomogeneous, comprising
areas that are mineralogically distinct and others
that are amorphous. Fuel (charcoal) inclusions are
also evident, as well as areas of extensive weathering
due to burial in a damp environment. Overall, the
sample is very rich in iron, which occurs in two
phases, either as wustite (iron oxide, FeO) or as iron
silicate (fayalite, 2FeO.SiO2). The abundance of iron
in these two particular phases suggests iron-smith
-
ing slag. However, work on traditional bloomery-
making in the Highlands has revealed a relatively
high percentage of iron (60–70% FeO) in what is
certainly tap slag from a smelting cycle
(Photos-Jones et al forthcoming).
SM3 This fragment of wood charcoal proved to be
rich in calcium. Sulphur and phosphorus are also
present in significant amounts and these can
either be associated with the fuel or may be derived
from impurities obtained in the course of
deposition.
SM1 This fragment of ore or flux has a honeycomb
structure like that encountered elsewhere in the
course of the writer’s analytical work on bloomery
slags (Photos-Jones et al 1998). They appear as
inclusions in smelting slag and some of them appear
not to lose their characteristic structure even after
being heated to temperatures prevalent in a
bloomery furnace. The precise nature of this
material is yet to be established.
7 Animal and Plant Remains
7.1 Animal bones from Smoo Cave
by James H Barrett
This section presents an analysis of fish, mammal,
bird and amphibian bone recovered during
small-scale sampling of cultural deposits inside the
mouth of Smoo Cave in 1992. Most of the bone
derived from the upper shell midden deposits
(contexts 006a/b, Phase 5). Small quantities were
also recovered from other contexts of other phases,
including sand or cobble beach surfaces incorpo
-
rating a lower density of anthropogenic material
(Pollard 1992).
7.1.1 Methodology
Twelve sediment samples of unspecified volume
were collected by hand during the excavation and
subsequently sieved using 1mm and 4mm mesh.
Bone considered here was sorted from the >4mm
fraction. A small number of hand-collected speci-
mens (six fish, 15 mammal) retrieved from the shell
midden have been combined with the sieved
material.
Some specimens will have been lost during sieving
(see Jones 1982; Payne 1992). This issue is of greatest
significance for fish, the only assemblage of sufficient
size to justify analyses of taxonomic and butchery
patterns. A comparison of results regarding >4mm
and <4mm sample fractions from a broadly contempo
-
rary fish assemblage at Earl’s Bu, Orkney suggests
that the bias introduced by using 4mm mesh may be
modest (Barrett 1995). The cod family fishes which
dominate northern Scottish assemblages reach c
0.15m within their first year of life and can grow to
lengthsinexcessof1.5m(Wheeler 1969; Wheeler
1978; Whitehead et al 1986a). Their bones are corre
-
spondingly large. Potential impacts of recovery
procedures on species and element distributions will
be considered further below.
Slightly different analytical methods have been
employed for each class of bone. The fish material
was analysed following procedures discussed previ
-
ously (Barrett 1995). All specimens were counted,
weighed and examined for taphonomic alterations
such as burning, butchery and carnivore-gnawing.
Forty skeletal elements were routinely identified to
family, genus or species. Qualitative data regarding
degree of bone fragmentation, texture and size were
recorded for nine diagnostic elements (Quantifica
-
tion Category 1 or ‘Q1’ elements as defined in Barrett
1995).
All mammal and amphibian specimens were iden
-
tified to the finest possible taxonomic category (with
the exception of mammal ribs which were only
grouped by size). This strategy was appropriate
given the tiny size of the assemblage. All specimens
were counted, weighed and examined for tapho
-
nomic alterations. Bone texture and portion (the
latter based on the Environmental Archaeology
Unit’s diagnostic zone system; K Dobney, pers
comm) were recorded for specimens identified to a
meaningful taxonomic category. Measurements
(after von den Driesch 1976) were taken on the few
bones that were sufficiently intact.
Bird specimens were analysed following a strategy
similar to that used for mammals and amphibians.
In this case, however, no attempt was made to
identify ribs or vertebrae. In all three classes, the
abundance of each taxon has been tabulated by
fragment count. Minimum number of individuals
(MNI) estimates were not tabulated as the dataset
was too small to be treated as more than a species
list. Details regarding the mammal elements
recovered are noted in Appendix D (archive report)
for the benefit of possible future comparative work.
7.1.2 Results
Fish In total, 2115 fish specimens weighing 323.6g
were recovered. The distribution among phases was
very uneven, with 1694 of these specimens coming
from Phase 5, the shell midden deposits (contexts
006a/b). A smaller concentration of 260 fish bones
occurred in Phase 3 (context 010). All other periods
yielded an insignificant number of specimens. Pres
-
ervation was good, but not exceptional, in both
periods.
No specimens from Phase 3 exhibited evidence of
fire-alteration or butchery. Nevertheless, given the
presence of anthropogenic deposits (Pollard 1992)
and the virtual absence of evidence for carnivore
damage, it seems likely that these bones derive from
human activity rather than piscivore (eg otter, see
Nicholson 1991) faeces.
Nine specimens from Phase 5 were fire-altered and
the same number exhibited distinct V-shaped cut
marks. Twenty-one specimens from this period
exhibited crushing, which could be consistent with
mastication (Wheeler & Jones 1989). Given the
virtual absence of partial digestion and the tiny
number of gnawed specimens, however, it seems
likely that most or all of this material also represents
cultural rubbish.
Phases 3 and 5 were both dominated by the cod
family, Gadidae. Phase 5 also yielded small concen
-
trations of flatfish and gurnard bones. All other taxa
were represented by fewer than 10 specimens each.
23
In Phase 3, saithe dominate the small cod family
assemblage (Table 2). Most of these specimens
derived from relatively small fish between 0.15m
and 0.5m in total length. Saithe can reach lengths of
c 1.2m (Whitehead et al 1986a) and large specimens
are common finds at Late Norse (medieval) sites in
northern Scotland such as Freswick Links (Jones et
al 1996) and Robert’s Haven in Caithness (Barrett
1995). In this region, small saithe are typically a
product of shore-based or shallow-water fishing
using a line or net (Fenton 1978; Baldwin 1982).
While the sample size is too small to justify definitive
statements, it is therefore possible that this assem
-
blage represents inshore fishing. In the absence of
evidence for fire-alteration or butchering, it is
difficult to suggest whether these fish were origi
-
nally intended for consumption in the cave or for
transport to another