Journal of the North Atlantic
108
K. Colls and J. Hunter
2015 Special Volume 9
Introduction
Over the last three decades, a significant amount
of research has been undertaken to advance our understanding
of the archaeological remains and settlement
history of the islands within the archipelago of
the Western Isles of Scotland (for example, Armit
1994, 1996, 2006; Branigan 2005; Harding and Dixon
2000; Parker Pearson et al. 2004; Sharples 2005).
Recent work has begun to move towards a more unified
study of the Isles, looking not only at the lessvisible
monuments but also at incorporating historic
landscape characterization, geomorphology, sealevel
change, and palaeoenvironmental archaeology
in order to study historic landscapes and the interaction
between communities and the places where they
live. The result is an array of published material, as
well as unpublished archives, which provide insights
into patterns of settlement from the Mesolithic onwards
(Bell 2007, Housley and Coles 2004, Rennell
2010). The findings allow researchers to understand
and interpret the complex modern island landscapes,
or “islandscapes”, and the factors which have influenced
their development. However, the Isle of Harris,
which sits in the northern part of the archipelago
(Fig. 1), has been relatively neglected. Perhaps due
to its lack of visible upstanding remains or because
of its harsh and variable terrain—steep mountains,
secluded valleys, and deep machair dunes—little
research has been undertaken to characterize and
interpret the archaeological resource of the Isle nor
has it been possible to define any differences in past
settlement patterns and human activity compared to
the other Western Isles islands.
In archaeological terms, Harris contains evidence
for Mesolithic, Neolithic, Iron Age, Viking,
Post-Medieval, and clearance remains, but monuments
and finds have been discovered largely by
accident rather than by systematic survey. In 2004,
the University of Birmingham began a long-term
research project with the aim of characterizing the
archaeological evidence across a range of the terrain
types throughout Harris (Colls and Hunter 2010).
A wide variety of non-invasive and intrusive archaeological
techniques were applied in an attempt
to understand patterns of settlement, to characterize
the remarkable adaptability of ancient and postmedieval
populations, and to better understand the
driving forces behind these adaptations, whether anthropogenic
or environmental. This dataset has enhanced
the Historic Environment Record for Harris
considerably; it now allows a much more accurate
and balanced archaeological signature to be produced
for the Western Isles. This paper provides a
summary of the main factors which have influenced
and impacted upon the development of the present
Harris landscape and introduces several themes and
case studies (both invasive and non-invasive) from
fieldwork. Individual site reports and comprehensive
data analysis will be reported on elsewhere.
Harris lies towards the northern end of the
Western Isles, an archipelago of approximately 500
islands off the west coast of Scotland covering 2900
Archaeological Signatures of Landscape and Settlement Change on the
Isle of Harris
Kevin Colls1,2,* and John Hunter3
Abstract - Between 2004 and 2011, a program of archaeological investigation by the University of Birmingham on the Isle
of Harris, a distinctive island forming part of the Western Isles of Scotland, has allowed the archaeological remains of this
enigmatic place to be further characterized and understood. Despite intensive archaeological interest in the archipelago for
a number of decades, the Isle of Harris has been overlooked, and only now are we beginning to identify the archaeological
resource and make comparisons to the wealth of published data from islands such as the Uists, Barra, and Lewis. This paper
highlights some generic overall patterns of archaeological signatures on the Isle that have been identified through a range of
archaeological methods including field walking, intrusive excavation, aerial reconnaissance, geophysical and topographical
survey, and documentary research. Several key case studies will be introduced including upland shieling complexes and
mulitperiod settlement sites on the west coast machair systems. The purpose of the paper is not to present a gazetteer of the
results of the work to date, but to highlight some of the key findings with a view to demonstrating that the Isle of Harris is
directly comparable with the archaeologically rich landscapes o f the other islands.
Special Volume 9:108–124
2010 Hebridean Archaeology Forum
Journal of the North Atlantic
1Institute of Archaeology and Antiquity, University of Birmingham, Birmingham B15 2TT, UK. 2Current address - Centre
of Archaeology, Computing, Engineering, and Science, Staffordshire University, R102, Science Centre, Leek Road, Staffordshire
ST4 2DF, UK. 3School of History and Culture, Arts Building, University of Birmingham, Birmingham B15 2TT,
UK. *Corresponding author - Kevin.colls@staffs.ac.uk.
2015
Journal of the North Atlantic
K. Colls and J. Hunter
2015 Special Volume 9
109
km2 (Fettes et al. 1992). Although not an island in its
own right, Harris, with the adjoining Lewis, forms
the largest island of the chain (Fig. 1). The geology,
topography, geomorphology, and climate of Harris
and the other Western Isles have greatly influenced
the types of settlement and agriculture that can be
developed there. Steep hills and valleys dominate
the Harris landscape, particularly in North Harris,
which contains more land over 400 m above sea
level than the rest of the Western Isles put together.
Although the geology of the Western Isles is dominated
by the Lewisian Gneiss Complex, a series of
Pre-Cambrian, metamorphic rocks, which are some
of the oldest in the UK (Collins 1986, Fettes et al.
1992), the geological signature of Harris, more
specifically South Harris, is much more complex
than its neighbors (Goodenough and Merritt 2007).
Bands of younger rock formations including granite,
steatite, and gabbro are present, and work is currently
ongoing to compare settlement patterns and
the location of archaeological monuments with the
complex geology of South Harris.
As with other parts of the Western Isles, geomorphological
characteristics are perhaps even more
influential on settlement patterns. Three key characteristics
of the Harris landscape that have impacted
both the past and present are peat deposition, machair
formation, and sea-level change. Predominant in
the uplands and mountains of Harris, blanket peat
has buried much of the landscape to varying depths
between 0.3 and 1.2 m. By the end of the Bronze
Age, the peat coverage of the Western Isles was
probably as extensive as it is today. Along the west
coast of Harrris, blown sand (commonly referred
to as machair) has had a similar obscuring effect,
but has the distinction of being dynamic rather than
static. Rising sea levels throughout the Holocene
controlled the onshore movements of vast quantities
of sediment from the extensive and shallow
coastal shelf, which in turn formed the machair dune
system (Wickham-Jones and Dawson 2006). Jordan
(2004) has outlined that mid-late Holocene sea-level
changes (based on sites on Harris and Lewis) can be
identified as mostly occurring in two major “events”
between 5500 ± 60 years BP and 4500 ± 100 years
BP and between 3000 ± 80 years BP and 820 ± 50
years BP. With the west coast of Harris being predominantly
fertile machair plain, there is the likelihood
that many of the coastal settlement sites of the
Mesolithic onwards may now be submerged below
the Atlantic seaboard. The absence of machair sand
from deposits underlying the earliest known site on
Harris (Mesolithic middens
at Taobh a Tuath; RCAHMS
Canmore ID 10502) suggests
this site predates machair formation
(Simpson 1965, Simpson
et al. 2006) as do the
earlier Neolithic settlements at
Coileagan an Udal, North Uist
(Evans 1971:52–62).
Cultural impacts on the
landscape
Cultural impacts on the
present landscape have been
equally significant. On Harris,
two in particular stand out,
partly through scale of impact
and partly through strength
of historical record. Both lie
outside what might have been
viewed as “traditional” archaeology
20 years ago, but both are
critical in understanding the
present and historic landscape:
the clearances of the mid-19th
century, and the subsequent
increased development of the
fishing industry. Fishing had
Figure 1. Location map of Harris showing key places discussed i n the text. already been established in the
Journal of the North Atlantic
110
K. Colls and J. Hunter
2015 Special Volume 9
eastern parts of Harris towards the end of the 18th
century as a commercial initiative undertaken by
the owner of the island, Alexander MacLeod. This
industry attracted an increased population from both
Harris and elsewhere in Scotland to the eastern villages
(known as the Bays), where the land was mostly
too rocky to cultivate. Elsewhere the traditional
farming soils on the machair to the west remained
the island’s main livelihood and focus of population.
However, from the later 18th century, landowners
throughout Scotland were beginning to recognize
that sheep were a far better source of revenue than
the relatively small and unpredictable rents provided
by farming tenants. The result was a wholesale
clearance of crofters from traditional fertile areas of
settlement to make way for the sheep. This entailed
destruction of dwellings, many being dismantled to
provide stone for sheep walling, and a fundamental
change of boundaries, field systems, and settlement
patterns. So severe was this change, that the preclearance
landscape was virtually obliterated. The
only structures likely to have survived were those
of value to the sheep farmers, such as the excavated
building at Borve (discussed below). Apart from
those, only slight traces of the earlier landscape now
remain within the modern field systems (Fig. 2). In
human terms, the “clearances” as they were termed
entailed an enforced move of families from the
western machair and from fertile islands such as
Pabbay to seek a subsistence from fishing in the
east, notably on the island of Scalpay. This social upheaval
created unprecedented population centers in
areas hitherto rejected throughout prehistory and the
Middle Ages. It was a reluctant move for most, and
for many, emigration to the New World was seen as a
better alternative. One key reflection of the resulting
population pressure and shortage of land can still be
seen in the numerous “lazy beds” created on every
available slope and patch of land irrespective of soil
quality (Fig. 3).
These fundamental social changes are barely
evident in the census returns. From 1801–1951, the
returns for Harris show a population fluctuating between
≈3000–5500 souls (the present population is
now barely 2000), but this data does not shed light
on the different transitional demographic trends:
emigrants from clearances, internal migration from
the fertile west to the inhospitable east, and incomers
drawn to the fishing from outside the island. It
also does not discern the effects of kelping, which
was underway before the clearances occurred. Kelp
collection was already attested in the 1792 Account,
which describes how seaweed was carried from the
shore and used to manure cultivated land (Sinclair
1978:59f). Later, kelp was collected in order to be
burned to produce alkali for glass manufacture. This
harvest did much to remove manpower away from
farming and fishing to the long-term detriment of
Figure 2. Sunken track way identified during site survey .
Figure 3.“Lazy Bed” cultivation visible as topographic
and vegetation change.
Journal of the North Atlantic
K. Colls and J. Hunter
2015 Special Volume 9
111
both. Kelp burning was arguably as close as Harris
ever came to the industrial revolution, but the price
of kelp fell dramatically in the earlier 19th century
and never recovered. By the mid-19th century, the machair
on the west coast of Harris, traditionally home
to farming communities from prehistory onwards,
appears to have been populated by sheep rather than
people, and the character of the earlier landscape all
but obliterated. It never regained its former population
status, the commercial fishing eventually failed,
and in the process, the slowly
evolving cultural landscape of
Harris became completely remoulded.
One example of this transition
can be seen in the ruins of
deserted settlements, particularly
in the coastal inlets to the
east. Agricultural remains, in
the shape of boundaries, drainage
and soil preparation/management,
and sites associated
with the exploitation of the sea,
form an important component
of the surviving archaeological
record, and yet these categories
of evidence seem to be the
least studied. The remains of
stone jetties, fish traps, and
house foundations around the
inlets and natural harbors attest
to the development of those
areas since the 18th and 19th
centuries. For example, at the
small community at Miabhaig
situated within the East Bays
(Fig. 1), over 50 fishing-related
sites are still visible today
along the shore. Mostly seen at
low tide, these weed-covered
jetties, slipways, and traps
are inextricably linked to the
ruined houses and structures
located just a few meters from
the shore (Figs. 4, 5). The
small communities along these
shores thrived and struggled
as the fishing industry and
crofting, and the relationship
between the two, developed.
Shielings
In common with other
farmers practicing a mixed
economy in the Scottish highlands,
those on Harris traditionally utilized the upper
pastures in the summer months. In Scotland, this
process of transhumance, which involved moving
the sheep and cattle to higher slopes in order to take
advantage of the new summer grazing, stems from
Norse times, in places continued through into the 20th
century, and is well documented (Bil 1968). Settlements
were effectively divided into two parts: the
main household or farmstead on the lower ground
where cultivation took place, and the rougher
Figure 4. Ruined structures and houses are common sites around the Isle of Harris.
Figure 5. Disused fishing boat slipway only visible at low tide.
Journal of the North Atlantic
112
K. Colls and J. Hunter
2015 Special Volume 9
buildings known as shielings on the upper slopes that
acted as temporary shelters for the shepherds (often
girls from the village), storehouses, and working
areas during the summer months. These buildings
were usually of drystone construction, sometimes
part-corbelled in bee-hive shapes, and often occurred
in clusters giving the impression of small, nucleated
settlements. Some of them may even have taken on
a more permanent status as a result of population
pressures after the clearances (discussed above).
The same buildings were re-used annually, presumably
being cleaned out and restored to the extent
necessary at the beginning of each new season. The
longevity of individual shielings remains conjectural.
Very few have ever been
excavated but, by virtue of their
locations along remoter valley
sides, many have survived redevelopment,
robbing, and the
ravages of curious tourists, and
may conceivably be many centuries
old. Several are perfectly
preserved and still roofed, such
as two examples set high up the
steep northern slopes of Vigadale
on the border with Lewis
(Fig. 6). One of these shielings
measures approximately 2.5 x
2 m with rounded corners and
with the upper courses gradually
indenting to create a convex
roof, now grass covered. It has
a low entrance (≈0.7 m high)
and four storage niches set into
the inside face of the walls. The
majority of shielings, however,
survive only as grass-covered
shells. Those on Harris tend to
have in common a bright green
swathe of grass around the
outside, presumably testifying
to the disposal of generations
of domestic waste and organic
activity.
Field surveys in Harris have
now identified and recorded a
number of clusters of shielings,
mostly located in the mountains
of North Harris, where they
tend to occur on the valley sides
and are sometimes positioned
high up in order to provide
vantage points over the flocks.
Their precise locations may
have been determined by the
easy availability of building stone, either from natural
geological resources and scree, or from previous
shielings or even earlier collapsed archaeological
remains. One typical cluster of structures, also on
Vigadale, covers an area of approximately 70 m x 50
m, and probably contains at least 10–12 overgrown
structures (Fig. 7). There are remains of at least three
or four grass-covered foundations terraced downslope
to the east, remains of possibly three or four further
grass-covered structures terraced on the higher
slopes to the southwest, together with two better-defined
upstanding remains. One of these measures approximately
5 m x 3 m with drystone walls ≈1 m wide
surviving to a height of ≈0.75 m with a slight annex
Figure 6. Beehive shielings located in the North Harris Estate close to the Lewis/
Harris border.
Figure 7. The ruins of a summer settlement at Vigadale, North Harris.
Journal of the North Atlantic
K. Colls and J. Hunter
2015 Special Volume 9
113
logical remains beneath the machair dunes throughout
the other islands, this area was selected for an
intensive program of fieldwork to assess the date
and character of any buried remains and to provide
patterns of settlement for comparison with other
machair sites along the Hebridean Atlantic coastline.
In 2005, after intensive and systematic field recording,
a number of geophysical surveys (resistivity
and magnetometry) were undertaken across the Aird
Nisabost peninsula, all of which identified the presence
of significant anomalies that were interpreted
as archaeological in origin (Fig. 10).
On the high ground, field recording proved successful
in further characterizing substantial enclosure
systems in Areas 1 and 2 (Canmore ID 10537).
These areas were then also investigated by geophysical
survey. In Area 1, an area covering approximately
6500 m2 is enclosed by two parallel ditches with
off to south and which appears to be constructed from
an earlier curved(?) structure using larger stones. The
other, a small stone-sided feature approximately 2 m
x 2 m was constructed of drystone walls ≈0.5 m wide
surviving to height of 0.5 m with a possible entrance
to the southwest. Nearby parts of a large glacial erratic
rock has been adapted by crude drystone walling
into covered chambers or rooms with lines of
drystone walling. One measuring some 4 m x 3 m x
0.75 m, located at the north side, effectively forms a
type of cave under the rock overhang with a narrow
entrance to north. Most structures of this type still lie
unrecorded, undated, and largely unrecognized for
their social and historical significance.
The Machair
Much of the impetus of the fieldwork was directed
toward the machair on
the western coast of South
Harris, notably at Nisabost
and Borve (Fig. 1). The findings
confirm the machair to
be one of the key archaeological
areas on Harris. Although
many of the placenames are
of Norse origin, the presence
of the Neolithic standing stone
known as Clach Mhic Leoid
(MacLeod’s Stone; Canmore
ID 10532) and two chambered
tombs point to a much older
archaeological legacy. With
the sea on three sides, the imposing
peninsula at Nisabost
consists of an upper terrace,
where the MacLeod Stone is
located, and a steeply sloping
southern face leading to a
very large area of undulating
machair dunes (Figs. 8, 9). A
number of archaeological sites
were identified on both the
upper and lower levels of the
peninsula that included visible
earthworks and enclosure systems,
eroding middens, structural
remains, and spot finds
(Canmore IDs 10528, 10529,
10533, 10536, 10537, 269182,
305970). The area is prone
to erosion damage caused by
a combination of livestock,
wind, and tidal actions. Given
the richness of the archaeo-
Figure 8. The west coast of Harris showing the machair coastal dune syste ms.
Figure 9. The Aird Nisabost peninsula.
Journal of the North Atlantic
114
K. Colls and J. Hunter
2015 Special Volume 9
Figure 10. Three locations of geophysical survey at Aird Nisabost (top) and close
up of the geophysical results (bottom).
internal stone and earthen bank (standing to a height
of 0.5 m). Geophysical survey identified a possible
structure in the western corner of the enclosure as
well as internal divisions (Fig. 11). Similar square
and rectangular enclosure systems were identified
along the eastern section of Aird Nisabost (Area 2),
although here the earthen banks are better preserved,
surviving to a height of approximately 1 m (Fig. 12).
Further, more intensive, fieldwork has been completed
on the low-lying machair dune systems covering
the southern area of Aird Nisabost (Area 3). Already
noted for the number of prehistoric and Norse
find spots which have been recorded throughout
the last century (Lawson 2002), and the volume of
possible sites recorded during fieldwalking in 2004
and 2005, this area has great potential to further our
understanding of ancient machair settlements on
Harris and to provide us with a comparable dataset
for other islands. An area totalling 15,200 m2, which
included mounded dunes of varying sizes and the flat
areas in between, was surveyed with a range of geophysical
equipment. Many anomalies were identified,
including linear and discrete
features across the survey area. In
addition, a topographic survey of
the dune system was completed
using differential GPS, and the
subsequent digital terrain model
was used as base heights for the
geophysical survey (Fig. 13).
A program of archaeological
excavations at Aird Nisabost,
Area 3 followed from that survey
and is still in progress. A series
of small pits, located to test the
results of the geophysical surveys,
were excavated in 2006 and
revealed a complex array of stone
structures that represents significant
and important remains of
human occupation, in association
with artifacts dating to the Iron
Age and Norse periods. Between
2009 and 2011, a substantial excavation
of one settlement mound
was completed, resulting in the
identification of a Late Iron Age/
Pictish building containing three
separate elements (Structures
1–3; Figs. 14—17).
Structure 1 is circular in plan
with a diameter of ≈4.8 m and
standing in part to a maximum
height of ≈1.3 m (Fig. 15). The
local geology on Harris does not
lend itself readily to the production
of laminar building materials.
Although many of the walling
stones were appropriately
faced, those on the uppermost
courses were much cruder and
presumably reflect a lower need
for structural stability against
the eaves. Construction consisted
of an inner single face of stones
packed against the sand. It was
Journal of the North Atlantic
K. Colls and J. Hunter
2015 Special Volume 9
115
of part-corbelled character, although it was not
clear whether the internal batter of the stones was
a deliberate corbelling effect or a result of the pressure
of sand from the outside. Inconsistency of the
batter suggested the latter. Sections cut through the
walling in two places illustrated the absence of any
backing material (e.g., turf), thus indicating that the
structure had been cut into an existing machair dune.
The basal courses had been constructed directly on
sand, and no foundations were evident. No obvious
post supports were identified, although two possible
contenders (F209 and F210) were positioned at the
edge of the walling. Their shallow profile suggests
that, if anything, they are more likely to represent
post-pad impressions.
Exposure and part excavation of the building in
its entirety demonstrated that the circular walling
(F106) was not of a single phase; evidence of likely
blockings and rebuilding could be interpreted in
at least two places (F202, F203). These alterations
may have been necessitated by change of use or by
collapse, the latter being evident at the northwest
and possibly being the cause of abandonment. There
were two existing openings in the walling circuit, one
to the northeast (≈0.6 m wide)
and one to the southeast (≈0.5
m wide) marked by a kerb and
staining from a threshold stone
(F211). It seems likely from
a topographic point of view
that the northeast opening may
have led directly outside while
the latter provided access
into Structure 2. The possible
blocking F202 may originally
have led through into Structure
3 to the south, and F203 to a
further (undefined) structure
to the west. Structure 1 was
clearly not an isolated unit,
but part of a larger complex of
units which together held some
form of social or functional
integrity.
Internally, excavation was
undertaken down to what was
interpreted as the floor surface.
This feature consisted of a compacted
mix of clay, charcoal,
and burning and was exposed
in all but the northwest quadrant,
which was not excavated
further. A hearth (F204; Fig.
16) was located slightly offcenter
to the structure. This feature
was formed by three kerb
stones, measured approximately
0.6 m x 0.4 m, and was open
to the east. It contained a heavily
compacted burnt fill resting
on a pad of bright red clay.
There was a further hearth-like
but less-formal feature (F205)
located against the wall at the
east. This element was subcircular,
approximately 0.4 m
Figure 11. Area 1 geophysical results overlain onto topographic data (provided by
Microsoft Google Earth).
Figure 12. Ancient enclosure identified as a visible earthwork in the location of the
area 2 geophysical survey.
Journal of the North Atlantic
116
K. Colls and J. Hunter
2015 Special Volume 9
in diameter and was surrounded by, and infilled with,
small cracked burnt stones.
The south and southwest of the circuit was characterized
by wall “units” consisting of a possible
cupboard or enclosed area (F206) and the remains of
a stone-sided tank with luting of grey clay (F207).
An orthostat located to the west also seems to have
been a component of this group providing a total
enclosed area of some 3 m x 1 m. A further stonesided
unit (F208) was located against the north wall.
This unit measured approximately 1.5 m x 0.6 m.
All the various elements had been established directly
into sand. Their use is unclear, and excavation
revealed little evidence of a base or flooring within
them. A cube-shaped, heavily used smoothing stone
was recovered from the sand fill of F206; this item
may have fallen from a ledge in the wall. A further
Figure 13. Digital Terrain Model (DTM) of the dune systems
at Aird Nisabost generated using GPS survey (top) overlain
by the geophysical data (middle) to create a combined dataset
of geophysical data that correlates directly to topographic
properties of the dunes (bottom).
Figure 14. Plan of the Iron Age stone buildings excavated at Aird Nisabost.
Journal of the North Atlantic
K. Colls and J. Hunter
2015 Special Volume 9
117
remains or residues from “industrial” processes such
as metalworking, bone working, potting, or tanning
(malodorous residues). The activities undertaken
required some form of heat and perhaps storage—
possibly for skins or wools rather than for grain or
produce—for which the luted watertight tank (F207)
appears to be focal.
The fill of Structure 1 was almost entirely of aeolian
sand, but with some stabilization lines and dark
root horizons. There was some charcoal, patches of
shell midden and animal bone,
as well as pieces of antler in
the upper fills. The contouring
of the sand infill gives
some idea as to the nature of
abandonment. The collapse at
the northwest side was never
repaired, and sand appears to
have blown in from the west. It
accumulated within the eastern
side of the structure and became
sufficiently stable for a
thin soil horizon to have developed
on its surface and across
the former working floor. This
horizon effectively provided
a stratigraphic marker, and a
seal over the internal features.
Repeated sand and stabilization
events subsequently filled
in the structure completely.
Structure 2 was probably
a later addition accessed
through the passageway ≈2 m
in length leading from Structure
1 (Fig. 17). The passage
widened out (a later event?),
and the walling was somewhat
cruder in construction. The
primary surviving element of
Structure 2 consisted of a
curved wall line (F112) constructed
of thick upright slabs
resting against a sand backing
reminiscent of Pictish period
structures at Bornais, South
Uist. However, this element
had been later superimposed
by slabbed upper coursing as
part of a modification. F112
was of varying thickness, approximately
0.7–1.2 m, poorly
stone-faced on either side.
The fill of the structure
was largely aeolian sand with
smoothing stone was located elsewhere in the building,
and these two objects provide the only clues as
to the structure’s function. The excavation of F208
at the north produced a hammer stone, three small
fragments of pottery dated to the Late Iron Age,
and a small fragment of a copper alloy object. The
remarkable paucity of both pottery and animal bone
throughout the excavation of the structure strongly
suggests that the building was unlikely to have been
“domestic” (i.e., living quarters). Nor did it contain
Figure 15. Photograph of the Iron Age stone building (Structure 1) excavated at Aird
Nisabost showing internal furniture and compartments.
Figure 16. In situ hearth within the central buildings at Aird Nisabost.
Journal of the North Atlantic
118
K. Colls and J. Hunter
2015 Special Volume 9
stabilizing layers and root horizons. At the entrance,
there was a thin dark stain lying parallel to the likely
position of the door, possibly representing a dust line
for the door, or even the rotting door itself. There
was also a hinge socket of soft degraded quartz. This
was the only part of the floor surface that seemed to
have survived. The depth from the current ground
surface to the floor was ≈0.5 m. A sherd of coarse
pottery was recovered adjacent to the socket (context
117).
Structure 3 appeared to have been inserted in
through the south outer face of Structure 1, and consisted
of the curved end of a further cell (Fig. 18).
The walls were crudely constructed, but seemed to
be concave in profile, similar to those of corn dryers
from elsewhere. Again, the fill was of aeolian sand,
but with root horizons and darker lenses. Although
excavated to a similar depth to Structure 1, no floor
level was identified.
The few pieces of pottery recovered would appear
to belong to the later part of the Iron Age. The
character of Structure 1 and the complex of cellular
structures to which it appears to be integral are
strongly indicative of the post-broch period, similar
to, for example, the structural complexes at Pool in
Sanday (Canmore 3422), and Buckquoy in Orkney
(Canmore ID 1799; Hunter 2007, Ritchie 1977) and
Bostadh in Lewis (Canmore ID 4130; Neighbour
and Burgess 1996). In Structure 2, the wall line
constructed of thick upright slabs resting against a
sand backing is reminiscent of the post-wheelhouse
structure at Cnip in Lewis (Canmore ID 4009; Armit
1996:165).
The site at Borve lies within
the west-facing dune system
in the south of Harris on the
fringes of the machair (Fig. 1).
The site was first recognized
through field walking as the
probable remains of a stone
structure eroding from the
sides and base of a large dune.
Reference to the natural section
of the dune itself indicated
at least three former ground
surfaces above the exposed
stones, emphasizing the constant
process of topographic
change. There is little known
archaeological or historical
background to the area, although
the name itself may
derive from ON borg (fort)
and lies within a concentration
of other Norse names. The
landscape was cleared in the
middle part of the 19th century
to make way for sheep; most of
the stone structures were dismantled
to provide field walls.
Excavation took place in 2005
and 2006 and involved the partial
excavation of a multi-phase
stone structure measuring at its
largest stage of development
some 10 m x 5.5 m (Fig. 19).
Although it is difficult to draw
definitive conclusions from the
scale of excavations undertaken,
certain clear observations
can be made.
Figure 17. Photograph of Structure 2.
Figure 18. Photograph of Structure 3.
Journal of the North Atlantic
K. Colls and J. Hunter
2015 Special Volume 9
119
The stone structure excavated
was not primary to the site.
It had been preceded by an
earlier post-hole building of
unknown date (Phase 1; Fig.
20) and the construction of
a clay floor (Phase 2). These
were only partially excavated,
but the presence of a stone
saddle quern of prehistoric
type (Fig. 21) together with
a possible broken megalith
reused within the walling of
the later building and/or its local
context suggested that that
the site may have been of long
duration.
The stone building itself
appears to have been of three
stages: a primary kiln unit
(ostensibly for drying grain;
Phase 3), followed by the
addition of a larger cell to
the southeast (Phase 4; Fig.
22) and subsequent rebuilds
and extensions. Although the
structural elements were on a
similar alignment, there were
clear differences in the quality
and method of construction.
The kiln structure was substantially
double-faced and
mortared, while the added
cell was mostly single-faced,
partly mortared, and revetted
into waste from the kiln.
The kiln itself consisted of a
bowl-shaped feature with a
maximum diameter of ≈1.5 m
surviving to a maximum of
twelve courses held together
with a greenish grey clay mortar.
At the base, was a rectangular
flue opening supported
by a flat lintel block within the
main wall.
At some point following
the layout of these cells, the
entire structure was revetted
on all sides, initially by an
earthen bank and then with a
low, poorly consolidated kerb
wall (Phase 5). A second adjoining
structure was built to
Figure 19. Excavations of a stone building at Borve, West Harris.
Figure 20. Postholes sealed by the existing clay floor of the structure at Borve. This
could indicate either an earlier structure on this site or significant redevelopment of
the building.
Journal of the North Atlantic
120
K. Colls and J. Hunter
2015 Special Volume 9
rather than as a second building of the same constructional
quality as that to the north. Over time, a
series of deposits were dumped against the southeast
exterior wall of the main northern structure. The
evidence for burning within these deposits suggests
that they may have been associated
with some form of
“industrial” activity, possibly
metal-working.
Although the excavations
have clarified certain features
of the site, they have also
raised numerous unanswered
questions. The phasing of the
site needs far more clarification,
as does our understanding
of the actual function of
the complex. The stone kiln
in the northern cell is of an
appropriate type to have been
used for grain drying, but other
interpretations, such as distilling,
are also feasible. Moreover,
if the southern cell/area
was used for metal-working,
then was this activity concurrent
with the grain drying?
the south of the first building, seemingly connected
at its eastern extent (Phase 6). The nature of this
adjoining structure is unclear without excavations
within its interior. However, the construction of the
wall suggests that it may have acted as an enclosure
Figure 21. Photograph of the partial remains of a quern stone utilized in the external
stone wall of the Borve structure.
Figure 22. Phased plan of the Borve structure.
Journal of the North Atlantic
K. Colls and J. Hunter
2015 Special Volume 9
121
surprisingly, the immediate area contains a wealth
of concentrated archaeological remains consisting
not only of Iron Age fortifications overlooking the
harbor approaches—Loch an Duin (RCAHMS 1928,
no. 121) and Dun Stuaidt (RCAHMS 1928, no. 122),
a possible monastery, and the only (late) medieval
church standing in the Western Isles—but also a
plethora of boundary features of different character
dating from prehistoric to clearance times. There is
no doubt that Rodel was strategic in the medieval
shipping lanes of the western seaboard and as a result
was also likely to have been a place of status, now
reflected in the impressive stone church of St. Clement
located on a headland overlooking the harbor and
constructed by Alexander MacLeod of Dunvegan and
Harris (also known as Alasdair Crotach) around 1520
(RCAHMS 1928, no.111). Given the quality of the
land and setting, there can be a reasonable assumption
that settlement continuity spanned from prehistory
throughout the Middle Ages and beyond. At the
time of the clearances, which at Rodel occurred in
the earlier part of the 19th century, “150 hearths” were
recorded by the Highlands and Islands Commission
If so, then the site appears to have been dedicated
to a variety of relatively large-scale industries and
thus raises various questions regarding the nature of
the products, the intended market, and the extent of
local settlement infrastructure. A series of notable,
if minor, earthworks down the slope to the west of
the site suggest extensive archaeological remains,
although the nature of these features is yet to be
clarified.
Perhaps equally importantly, the site also still
remains largely undated, at least to within a narrow
time-frame. The crofts around Borve were abandoned
in the 19th century as part of the clearances,
and no oral history refers to an industrial complex
on the site following this event (G. MacLeod, local
historian, Isle of Harris, pers comm.); nor is there
evidence for settlement on 1st-edition Ordnance Survey
maps. The style and construction of the structure
suggest a medieval or early post-medieval date,
based on comparative architectural analysis. In support
of this interpretation, the pottery from within
the building could date from anytime between the
14th to 18th centuries (Mary MacLeod, then Western
Isles Council Archaeologist,
Carloway, Isle of Lewis, pers
comm.), although these finds
were mainly from post-abandonment
deposits. However,
the recovery of uncontexted
Iron Age pottery from a stratified
context outside the main
structure indicates prehistoric
activity in the area. Likewise,
the broken megalith (?) within
the constructional material of
the wall separating the two main
cells could plausibly relate to
the re-use of a Neolithic/Bronze
Age standing stone and may hint
at a far longer site history.
Rodel
One further site deserves
mention here: the area around
Rodel at the southern tip of the
island (Fig. 1). This location has
the benefits of a natural harbor
(since improved) and is situated
favorably for sea routes to other
parts of the Western Isles and
across the Minch to Skye. It offers
the additional benefits of a
sheltered valley containing rich,
reddish soils more fertile than
elsewhere on the island. Not
Figure 23. Plan of a possible ancient settlement located at Rodel, South Harris identified
by fieldwalking, geophysical survey and excavation.
Journal of the North Atlantic
122
K. Colls and J. Hunter
2015 Special Volume 9
fering from considerable rubble collapse on both
faces (Fig. 24). The faces were uncoursed, and their
unevenness reflected the poor quality of local stone
for building purposes. The whole wall was sunk
into natural clay and facilitated a slightly lower
floor on the interior—a characteristic of blackhouses
(Walker and McGregor 1996). The fill between
the faces was of compacted soil and small stones
and contained a significant quantity of animal bone
and artifactual material, including pottery, glass,
and a small cannon ball.
Trench B measured 3 m x 0.75 m and revealed
what appeared to be the sporadic remains of two rough
stone walls, but further investigation and a trench
extension failed to confirm this. The stone elements
appeared to be geological, and the absence of any do-
(cited in Lawson 2002:71), indicating a significant
level of population.
During field walking in 2004, an area of crofting
land at the bottom of the valley immediately north
of the church was interpreted on the basis of topographical
expressions as containing possible house
platforms (Fig. 23). It was anticipated that these
might relate to medieval or post-medieval settlement
in the vicinity. The field containing these remains lay
adjacent to a byre converted from an 18th-century
building (now listed: HB Number 12908) and clearly
originally of some importance. Some 30 broadly
rectangular individual units were identified in total.
A small number contained clear evidence for stonework,
but the majority appeared to be turf-based. In
an attempt to clarify the extent and nature of this
site, a geophysical survey was
undertaken (resistivity) but
was unable to delineate the
turf-based features further. It
did, however, identify a likely
trackway running through the
“structures” (Fig. 23).
On the basis of topographic
and limited geophysical
evidence, four trenches were
established in order to test the
nature of the remains. Trench
A was used to bisect the walling
of a probable blackhouse
depicted on the 1st edition of
OS 6-inch map (1881), and
the other three (Trenches B, C,
and D) were positioned to test
turf-based features (Fig. 23).
The probable blackhouse was
estimated to be of dimensions
12 m x 7 m, which is towards
the smaller end of the scale
for blackhouse sizes (typically
between 9.14 m and 18.3 m
in length; Fenton 1978). In
Trench A, an initial area of
4 m x 0.75 m was opened to
bisect the wall and include
part of the internal floor; it
was later extended to 4 m x
1.14 m to take into account a
blocking feature in the walling
which prevented the true
nature of walling construction
to be observed. Once exposed,
the wall itself was shown
to be double-faced, approximately
1.5 m wide, and suf-
Figure 24. Photograph of the excavation through the structural remains of a blackhouse
at Rodal.
Journal of the North Atlantic
K. Colls and J. Hunter
2015 Special Volume 9
123
Acknowledgments
The authors would like to thank the University of Birmingham
and all the students that have taken part in the
field seasons; Harris Development Limited, in particular
Kenny and Duncan; The Heritage Lottery Fund; and
Members of the Harris Archaeology Group, in particular
George, John Goodwin, and Gillian. Special thanks as
always to Carol Knott for her support, motivation, and
enthusiasm for archaeology.
Literature Cited
Armit, I. 1994. Archaeological field survey of the Bhaltos
(Valtos) peninsula, Lewis. Proceedings of the Society
of Antiquitaries of Scotland 124:67–93.
Armit, I. 1996. The Archaeology of Skye and the Western
Isles. Edinburgh University Press, Edinburgh, UK.
Armit, I. 2006. Anatomy of an Iron Age Roundhouse:
The Cnip Wheelhouse Excavations, Lewis. Society of
Antiquaries of Scotland, Edinburgh, UK.
Bell, M. 2007. Prehistoric coastal communities: The mesolithic
in western Britain. CBA Research Report 149.
Council for British Archaeology, York, UK.
Bil, A. 1968. The Shieling 1600–1840. John Donald, Edinburgh,
UK.
Branigan, K. 2005. From Clan to Clearance: History and
Archaeology on the Isle of Barra c. 850–1850 AD.
Oxbow Books, Oxford, UK.
Collins, G. 1986. Geology of the Western Isles. Pp. 12–13,
In J. Barber (Ed.). Bronze Age farms and Iron Age
farm mounds of the Outer Hebrides. Scottish Archaeological
Internet Report 3. Available online at www.
sair.org.uk/sair3. Accessed 25 August 2006.
Colls, K., and J. Hunter. 2010. Defining the archaeological
resource on the Isle of Harris: An assessment of
the impact of environmental factors and topography
on the identification of buried remains. Shima: The
International Journal of Research into Island Cultures
4(2):15–40.
Evans, J.G. 1971. Habitat change on the calcareous soils
of Britain: The impact of Neolithic man. Pp. 27–74,
In D.D.A. Simpson (Ed.). Economy and Settlement in
Neolithic and Early Bronze Age Britain and Europe.
Leicester University Press, Leicester, UK.
Fenton, A. 1978. The Island Blackhouse: A Guide to the
Blackhouse at 42 Arnol, Lewis. Department of Environment,
Edinburgh, UK.
Fettes, D.J., D.I. Mendum, D.I. Smith, and J.V. Watson.
1992. Geology of the Outer Hebrides. British Geological
Survey, London, UK.
Goodenough, K., and J. Merritt. 2007. The Outer Hebrides:
A Landscape Fashioned by Geology. Scottish
Natural Heritage, Perth, UK.
Harding, D.W., and R. Dixon. 2000. Dun Bharabhat, Cnip:
An Iron Age Settlement in West Lewis, Vol. 1. Structures
and Material Culture, Calanais Research Series,
2. Edinburgh University Press, Edinburgh, UK.
mestic detritus (animal bone, pottery sherds, etc.) suggested
that the area had not been inhabited. Trenches
C and D demonstrated much the same. Trench C
measured 3 m x 1 m and was positioned over a specific
linear resistivity anomaly. On excavation, this feature
transpired to be the gully from a ridge and furrow system.
Trench D was located on the higher ground away
from the other trenches and measured 3.2 m x 1.2 m.
Like at Trench B, putative walling transpired to be
geological outcropping, and no positive evidence for
structural remains was discovered.
In many senses, the results were disappointing in
that the “lost” medieval settlement at Rodel was not
identified. However, the topographic turf features investigated
were clearly of anthropogenic origin and
seem to have incorporated natural geological features;
they perhaps related to cultivation or agricultural
practices. More optimistically, the finds from
the stone structure supported an 18th- or 19th-century
date appropriate for a blackhouse. Two fragments of
medieval pottery were also recovered; these at least
testify to a medieval presence in the area.
Conclusion
Research on Harris has now demonstrated the
survival of significant archaeological deposits dating
from prehistory through to the 20th century. Their
previous invisibility has been partly a consequence
of submergence under sand and peat, and partly due
to the extreme effects of the 19th-century clearances
in molding the landscape. It may also be argued
that the intensive research over the last 50 years on
the other Hebridean islands succeeded in highlighting
the profile of archaeology in these areas, but of
course this scenario never really developed to such
an extent on Harris. Detailed fieldwork as part of
this continuing project, both non-invasive and invasive,
has demonstrated that the level of monument
survival is no less than that on other parts of the archipelago;
in fact, given the range of landforms that
Harris offers, the diversity of monuments is arguably
greater than elsewhere. The survival of shielings
is a case in point, as is the presence of the fishing
landscape on the eastern coast and the unique status
area within the rich soils around Rodel; the machair,
as on Lewis and the Uists, conceals a rich and remarkably
intact settlement heritage from prehistoric
times. This new knowledge is the product of systematic
seasonal surveys and exploratory excavations
spread across seven years. Without it, Harris would
still remain an almost blank and barren area on the
archaeological map of Scotland.
Journal of the North Atlantic
124
K. Colls and J. Hunter
2015 Special Volume 9
Housley, R.A., and G. Coles. (Eds.). 2004. Atlantic Connections
and Adaptations: Economies, Environments,
and Subsistence in Lands Bordering the North Atlantic.
Symposia of the Association for Environmental
Archaeology 21. Oxbow, Oxford, UK.
Hunter, J.R. 2007. Investigations in Sanday, Orkney. Vol
1: Excavations at Pool, Sanday. The Orcadian and
Historic Scotland, Kirkwall, UK.
Jordan, J.T. 2004. Holocene coastal change in Lewis and
Harris, Scottish Outer Hebrides. Unpublished Ph.D.
Thesis. Coventry University, Coventry, UK.
Lawson, B. 2002. Harris in History and Legend. John
Donald, Edinburgh, UK.
Neighbour, T., and C. Burgess. 1996. Traigh Bostadh.
Discovery and Excavation in Scotland, 1996:113–114.
Council for Scottish Archaeology, Edinburgh, UK.
Parker Pearson, M., N. Sharples, and J. Symonds. 2004.
South Uist, Archaeology and History of a Hebridean
Island. Sutton Publishing, Stroud, UK.
Royal Commission on the Ancient and Historical Monuments
of Scotland (RCAHMS). 1928. Ninth report
and inventory of monuments and constructions in the
Outer Hebrides, Skye, and the Small Isles. RCAHMS
Edinburgh, UK.
Rennell, R. 2010 Islands, Islets, Experience, and Identity
in the Outer Hebridean Iron Age. Shima: The
International Journal of Research into Island Cultures
4(1):47–64.
Ritchie, A. 1977. Excavation of Pictish and Viking-age
farmsteads at Buckquoy, Orkney, Proceedings of
the Society of Antiquaries of Scotland 108(1976–
1977):172–227.
Sharples, N. 2005. A Norse Farmstaed in the Outer Hebrides:
Excavations at Mound 3, Bornais, South Uist.
Oxbow, Oxford, UK.
Simpson, D.D.A. 1965. Northton, Isle of Harris, kitchen
midden. Discovery and Excavation in Scotland
1965:22. Council for Scottish Archaeology, Edinburgh,
UK.
Simpson, D.D.A., E.M. Murphy, and R. Gregory. 2006.
The Prehistoric Settlement of Northton, Harris (BAR
British Series). Archaeopress, Oxford, UK.
Sinclair, J. (Ed.). Reprinted 1978. The Statistical Account
of Scotland 1791–1799, Volume XX, The Western
Isles. EP Publishing, Wakefield.
Walker, B., and C. McGregor. (Eds.). 1996. The Hebridean
Blackhouse: A Guide to Materials, Construction,
and Maintenance. Historic Scotland, Edinburgh, UK.
Wickham-Jones, C.R., and S. Dawson. 2006. The scope
of Strategic Environmental Assessment of North Sea
Area SEA7 with regard to prehistoric and early historic
archaeological remains. Strategic Environmental
Assessment Programme report. UK Department of
Trade and Industry, London, UK.