Late finish to the excavation season at the Fiskavaig Rock Shelter site….

Posted by steven on 11/03/2009 at 10:37 AM

With poor weather towards the end of the 2008 excavation season, we were late finishing our fieldwork at the Fiskavaig rock shelter site. The remote setting and difficulty of access requires fairly stable weather conditions, so we selected the best of the days on offer completing the excavation of Trench 2 in November.

Plan of the rock shelter at Fiskavaig showing the location of Trenches 1 and 2

Although the weather conditions tested us at times, the excavations at the site were rewarding and we uncovered some exciting archaeology. In many ways, Trench 2 provided similar types of features to Trench 1 - in particular, we revealed and identified five new hearth settings. The images below show some of these features during excavation.

Hearth settings in Trench 2 associated with the final use of the rock shelter - scale = 0.5m

The slab-built hearth to the right in the image, along with its kerb of beach cobbles is a fine example and the last hearth to have been used in this part of the site. The hearth to the left in the image has been partially robbed, while the white deposit between the two is a concentration of burnt bone in a shallow scoop. Note also the deposits of fire-cracked pebble and fragments of animal bone. Little shellfish or fish bone was recovered from this trench, while the small quantities of these deposits generally at the site is intriguing, considering the location of the rock shelter adjacent to good fishing ground and a foreshore where limpets and periwinkles are numerous. We did identify a small oyster midden in the trench, which is also interesting considering the morphology of the foreshore at the site, which is not the type of ground where oysters naturally live. Therefore, as with the major part of the archaeological deposits identified at the rock shelter, the oysters would have been transported into the site from elsewhere.

A second view of the hearth settings from the west - scale = 1m

Plan of the surfaces in Trenches 1 and 2 relating to the final use of the rock shelter

The hearth setting below is of a type that has been identified in wheelhouses in the Western Isles of Scotland, where animal bone has been used as a form of organic kerb around the hearth. Again, a part of this hearth has been disturbed or robbed-out in the past, but the southern section of the feature had been preserved along with the animal bone. This hearth setting was deeply embedded in midden deposits and relates to the earlier use of the site. Although analysis and identification of the bone is required, it appears to comprise bone elements from red deer including limb bones and a lower mandible that have been placed end-to-end, forming a single line of bone. The hearth would have been a very important component of any dwelling, often placed centrally within structures and providing the focus of most day to day activities.

Hearth setting in Trench 2 with a kerb of animal bone - scale = 0.5m

Martin Wildgoose excavating the deposits around the hearth settings in Trench 2

The animal bone midden recovered from Trench 2 was extensive and possibly twice the quantity recovered from Trench 1. This may suggest that the major deposition of animal bone was taking place in this area of the site. With the excavation of the two trenches, we are starting to identify potential activity areas where different types of material are deposited. For example, in Trench 1 we uncovered a shellfish midden of limpet, periwinkle and some mussel shell, while we have uncovered a discrete oyster midden in Trench 2. Trench 2 has also produced the majority of the ceramics so far recovered from the site. However, we have found few other types of small find in this trench. In contrast, we have recovered a varied assemblage of small finds in Trench 1 and most of our evidence for metalworking.

Hearth setting, post and stake-holes in Trench 2 - scale = 1m

After removal of the archaeological deposits above, we finally came down to the primary occupation surface in Trench 2, which can be seen in the images above and below. This surface, which had been well trampled and partially scorched/burnt, was associated with two hearth settings. However, between the two hearths we uncovered evidence for ephemeral structures comprising four small post-holes forming a rough rectangle. Between the post-holes we found lines of smaller stake-holes, possibly hazel withies forming the uprights for a screen or partition.

Trench 2 from the north showing the two hearth settings (one hearth to the left in the baulk of the trench) and the arrangement of post and stake-holes. Note the reddened floor at the far end of the trench, which is possibly due to burning nearby - scale=1m

Structural evidence in rock shelters is quite rare, especially the remains of what would have been organic structures. The screens or partitions may have been used to shelter individual hearth settings, or may have been used for other activities, creating spatial complexity within the rock shelter such that identified in dwellings of the time such as roundhouses and wheelhouses. The drawing below shows a plan of the features in Trenches 1 and 2 relating to the first occupation of the rock shelter.

Plan of Trenches 1 and 2 showing the primary occupation surface at the rock shelter site

The plan above shows the hearth setting kerbed with animal bone and associated post and stake-holes in Trench 2, while in Trench 1 two hearths (one partially robbed-out) and a pit feature can be seen. In 2009, we will be excavating the area connecting these two trenches, which will allow us to interpret with more certainty the main activity and depositional areas in the rock shelter. We will also extend Trench 1 in a southerly direction to investigate further the concentration of metalworking residues identified in this area.

We have now received some of our specialists reports resulting from analysis of materials recovered from the Fiskavaig rock shelter site. These are included below.

Small Finds Analysis


Dawn McLaren & Fraser Hunter


This small assemblage of 25 objects from the first phase of work on the site comprises bone, ceramic, copper alloy, iron and vitrified material.  A summary of the assemblage is presented in the table below.

Vitrified material - 1 Stone with adhering glassy slag
Stone - 1 Natural crystal fragment, no evidence of working
Bone - 4 Needle or pin with perforated head
Roughouts -
Working debris -
Iron - 1 Unidentified
CuA - 8 Needle
Ingot - 1
Sheet, strip & rod fragsments
Casting debris
Ceramic - 10 Mostly body sherds of hand-made coarse vessel, some rim sherds

Table 1: summary of small finds assemblage

This material was recovered in 2006 and 2007 during preliminary investigations of a later prehistoric midden within a rock shelter at Uamh an Eich Bhric. The site is being actively eroded by the sea, resulting in instability and collapse of sections of the deposit. These finds were recovered from secure positions within the exposed midden face.

This mixed assemblage comprises finds of stone, bone, metal and ceramic. The single stone fragment, a crystal flake, has no evidence of working but would merit examination by a geologist. Examination of the bone objects by an osteologist to determine species and skeletal element is recommended.

The metal objects are badly degraded and require stabilisation. The iron object is unidentifiable in its current condition and requires X-radiography and perhaps conservation to allow identification. The copper alloy fragments are actively corroding and require stabilisation and conservation. Non-destructive XRF analysis is recommended for all the copper alloy fragments. 



This small assemblage of 176 objects from the 2008 phase of excavation on the site (Trench 1) comprises bone, ceramic, copper alloy, iron, stone and vitrified material.  A summary of the assemblage is presented in the table below.

Bone/antler - 14 Includes needle fragments, a skin scraper, pin/points, a bead, a spindle whorl and roughouts
Ceramics - 125 124 pottery sherds consisting mainly of body sherds. Rim, base and decorated sherds also present
Unusual perforated and decorated ceramic disc - possible lid?
Copper alloy - 12 Includes disc-headed rivet, cast projecting ring-headed pin, fragmentary fine ring and several sheet
and strip fragments
Iron - 7 Two possible blades, a ring, bar fragments and several unidentified fragmentary objects
Slag - 3 Unclassified iron slag
Stone - 12 Range of cobble tools including a burnisher, grinder, possible hammerstone and smoother. Three
joining fragments of a stone palette and an unfinished/re-used stone pendant in unusual stone
Vitrified ceramic - 3 Includes possible mould fragment

Table 3: summary of small finds assemblage

This small finds assemblage was recovered during the 2008 excavations of a later prehistoric site situated within a rock shelter at Uamh an Eich Bhric, Skye. The site is being actively eroded by the sea, resulting in instability and collapse of sections of the deposit. These finds have been recovered from secure positions within the exposed midden face as well as two trenches which were excavated further into the interior of the rock shelter, exposing several features including hearths, possible floor levels, pits and postholes.

Preliminary investigations in 2006 and 2007 recovered a small quantity of artefacts which included evidence of non-ferrous metal working in the form of casting debris, and an ingot, and bone/antler working. There were also ten sherds of hand-made pottery (see Table 1 above). The 2008 assemblage is consistent with the finds from the previous seasons of evaluation with further evidence of bone/antler working in the form of roughouts and unfinished objects. A greater quantity and range of copper alloy objects were recovered in 2008 with an emphasis on finished objects, such as a small cast projecting ring-headed pin and disc-headed rivet, rather than evidence of production, although small fragments of vitrified ceramic are present which may be mould or crucible fragments. In addition to non-ferrous metalworking, limited evidence of ironworking is now present in the form of a small quantity of unclassified iron slag. Whether this represents evidence of smelting or smithing is unclear due to the small size and fractured condition of the fragments. The preparation and working of leather is also represented by a bone skin-scraper and at least one possible stone smoother, while textile production is hinted at by the presence of an unfinished bone spindle whorl.

The worked bone/antler from the 2008 excavations complements that recovered in the preliminary investigations, with a greater range of unfinished and finished objects present. These include needle fragments, pins/points and a suite of unfinished objects and roughouts that indicate bone, antler and possibly cetacean bone was being worked on the site.

A limited range of stone cobble tools were recovered including possible hammerstones, grinders, smoothers and multi-function tools. Many of these general tools may have been used to process foodstuffs but could equally have been used to prepare clay for potting, grind pigments or crush ore for metalworking. More unusual stone items are also present, including three joining fragments of a highly polished palette and an unfinished or re-used pendant of an unusual stone. The worked stone objects would benefit from examination by a geologist to identify the lithologies present.

A small number of badly corroded iron objects are present, including two possible blade fragments and an iron ring. All will require x-rays and conservation prior to further examination.


The copper alloy and iron objects are badly degraded and require stabilisation. The iron objects are friable and many cannot be identified in their current condition. The iron requires X-radiography and cleaning to stabilise and to allow identification. The copper alloy fragments are actively corroding and require stabilisation and conservation. A stone palette, of which three joining fragments are present, would benefit from being rejoined. A full list of items requiring conservation is included in Appendix 1.

Examination of the bone objects by an osteologist to determine species and skeletal element is recommended and identification of the lithologies of the worked stone by a geologist would be greatly beneficial.

Non-destructive XRF analysis is recommended for all the copper alloy fragments. 


The 2008 small assemblage of finds from the rock shelter at Uamh an Eich Bhric compliments those found in the 2006-07 preliminary investigations and enhances our understanding of the activities taking place within the rock shelter during the late Iron Age. Further evidence of non-ferrous metalworking and bone/antler working is present as well as new evidence for leather-working, textile production, food processing/storage and limited residues from iron-working. Well-stratified animal bone from the midden has been dated to 60-240 AD, making this one of the few recently excavated and securely dated Iron Age sites with evidence of metalworking in Skye and western Scotland: it is a highly significant site. The evidence of non-ferrous and ferrous metalworking here complements that from High Pasture Cave, Skye, where crucible fragments have been recovered in association with an earlier Iron Age structure.

Full study and publication of this small find assemblage is recommended as it has the potential to afford a greater understanding of craft activities within Iron Age society and the role of the site in the wider Iron Age landscape.

Animal Bone Analysis

Trench 1 Fiskavaig, Skye: Animal bone assessment of potential

Jacqui Mulville 2009


This report is an assessment of the animal remains recovered from Trench 1 at Uamh an Eich Bhric (the Cave of the Speckled Horses), Fiskavaig.  This site is a rock shelter on the Isle of Skye containing deep deposits of archaeological midden material interleaved with rock falls, all covered over by a depth of overburden. The site is suffering from erosion by the sea and Historic Scotland has funded excavations to rescue the most vulnerable parts of the site from destruction. 

The main bulk of the deposits cover an area approximately 15m x 7m and to date two trenches have been excavated in 2008.  Trench 1 measures around 4m x 3m and Trench 2 measures around 4.5m x 3.5m.  A number of carefully constructed hearths have been revealed within these trenches, along with metal working debris (copper-alloy deposits and a clay mould fragment) and quantities of charcoal.  Other finds include iron objects, stone tools, pottery, fire cracked pebbles and substantial quantities of animal bone.  The evidence suggests the cave was occupied, possibly seasonally, with metal working occurring on site.  Three boxes of bone from Trench 1 have been received for evaluation, a greater quantity of material has been recovered from Trench 2 and excavations are continuing in 2009.  Material was recovered by hand, with only selected fragments considered useful for identification recovered for analysis.

Assessment Methodology

All contexts were scanned and the fragments identifiable to species and element quantified.  Condition of the bone was noted for each context on a scale of one to five, where one is very poor with fragments not identifiable beyond ‘bone’ and five is excellent with surface features such as cut marks, gnaw marks and pathology well-preserved.  Bones and teeth for which measurements could be taken, and/or the age ascertained, were noted along with any modifications that were evident.


A total of 1823 bone elements were recorded.  These were derived from 6 different contexts, contained within the ten squares (numbers 1 to 9 and 6a).  The majority of the material is derived from context 1.02. 

Of the total number of bones examined over 25% can be identified to species and element at the assessment stage.  The Number of Identified Specimens (NISP) is at 436, above the recommended minima for analysis of 300 specimens (Hambleton 2000).  None of the bone was in poor condition and the majority was in excellent condition with intact surfaces and sharp edges to breaks. There were a number of burnt fragments and a small amount of the bone had been subject to light gnawing by a large canid and by rodents.  The material contained numerous instances of articulating bones and also un-fused shafts with accompanying epiphyses. These demonstrate the integrity of the assemblage with material being recovered from the position of its initial deposition. 

Species Representation

The majority of the identified bone was cattle. They make up 87% of the three main food species, with smaller and similar quantities of sheep and pig. The latter two were represented by a few mandibles (these mandibles were ageable; for example the pig mandible was derived from an elderly individual) and long bone fragments. 


Body Parts

Cattle is represented by all the main elements of the skeleton, from the skull to the phalanges, many of which were relatively complete.  The number of measureable bones would have been much higher if not for the numerous butchery events which had damaged the articular ends. 


The butchery at the site is unusual with the position of cut marks paying little attention to the traditional points of carcass division at the joints.  These bones have been chopped at and through in a random manner - sometimes leaving deep gouges in the bone (e.g. in the middle of cattle ribs). There is little evidence of bone being smashed to extract marrow. 


Fusion evidence indicates that the majority of the bone came from animals between one and three years old.  Only a single neonatal animal was present and few younger animals are represented.


One mature (over four years old) pathological individual was noted in context 3, square 9, with a pair of pelves and associated femoral heads showing eburnation (changes associated with the femur rubbing in the pelvis socket).  In the same unit was a pathological first phalanx, probably from the same individual.  Other articulating material was noted in this context and square which contained the remains of at least two cattle.  Elsewhere another cattle pelvis showing eburnation was recorded. 

Minor species

Red deer is represented by one element and fragments of antler.  Wolf has been identified from a fragment of left maxilla/zygomatic which contains the carnassial and the two molars.  Positive identification of wolf can be problematic, however the teeth present are exactly those necessary to determine wolf or dog by using the measurements of these three teeth. In wolves the length of the carnassial is greater than the length of the first and second molar combined (Clutton-Brock 1963) and this is the case in this example.  The length of the carnassial itself is longer than reported Neolithic Dogs in Britain (Clark 2006, Table 4.1). There is also a possible small cut mark across the zygomatic consistent with the removal of the skin - these marks need further examination under a microscope.  The excavator reports little in the way of bird or fish remains.

Worked Bone

There is one example of worked bone, what appears to be a bird bone has been shaped into a narrow shaft (possibly from a bone pin).  The shaft has been broken at both ends.


There are few Iron Age cave assemblages recovered from Atlantic Scotland, the exception being High Pasture Cave, which is also on Skye.  This site is very unusual with evidence for metal working outside the cave and for special deposits of animal remains within the cave found in the form of a number of complete cattle, and a high prevalence of selected pig elements (Drew 2005).  This rock shelter is following this trend for unusual cave deposits with an exceptionally high percentage of cattle remains and non-traditional processing patterns. Within Atlantic Scotland there are no sites with cattle present at this level – although sites on Skye do have higher percentages than the majority of sites in the Atlantic Iron Age (see figure 1 below); for example the site of Dun Ardtrek on Skye (Mackie 2000) which is in relatively close proximity to the Fiskavaig rock shelter, while there are also large numbers of cattle from the earlier deposits (600-400BC) at High Pasture Cave (Drew 2005). 

Elsewhere higher levels of cattle exploitation can be found at a few sites, but none reach the levels recorded from Fiskavaig.  Cattle make up 50% or more of the cattle, sheep, pig and deer remains at Crosskirk Broch (McCarthney 1984) on the Scottish mainland and in later deposits on from the Vallum at Iona (McCormick 1981); on the Western Isles at the broch at Beirgh on Lewis (Thoms 2003 - although the domestic assemblage is dominated by cattle, overall red deer are most numerous) and in the small assemblage from Sollas.  On the Northern Isles this percentage of cattle is only found at Early Iron Age Howe and Mine Howe on the island of Orkney.  This unusual predominance of cattle at Fiskavaig may be a function of the small size of the assemblage but is more likely to demonstrate the deliberate selection of beef as the appropriate food for consumption at the site.

The range of elements recorded, from the skull down to the toes; indicate the presence of entire animals at the site.  Detailed analysis of the relative abundance of elements will help to clarify if any parts of the animals are under-represented (from the assessment it appears that toes may be under-represented).  Also evidence for the selection of particular sides of elements must be sought; at High Pasture Cave and at Llanmaes (Madgwick pers comm) right hand side elements were favoured.  If there is evidence for entire cattle being bought to the cave it is interesting to speculate how they reached the cave and why they, and not the more portable sheep, were favoured.

The cattle butchery evidence is also perplexing. The lack of precision in the location of butchery marks, the prevalence of heavy chops and repeated knife cuts as well as the large number of relatively complete bones hints at unusual practises.  This could suggest unskilled and/or hurried butchery was taking place at the cave; alternatively this pattern of butchery could be a deliberate disregard of the normal practises associated with food preparation. There is no attempt to carefully extract maximum nutritional value by careful division and filleting, preparation of meat for roasting and/or cooking within vessels and marrow cracking. The rough and rapid division of animals, the presence of articulating bones and adjoining epiphyses suggest the disposal of bones with meat/ligaments/cartilage still present. This suggests an abundance of meat and an impetus to quickly process, consume and dispose of the remains.

Semi-articulated cattle vertebrae exposed in the section of the midden

Future work

This material, and that from the other trenches, deserves full and detailed analysis.  The sieving of all material through a 1cm mesh would standardise recovery and sub-samples should be removed and sieved through finer mesh for the recovery of small mammal, amphibian, bird and fish remains.  The logistics of recovery are problematic within this location – however 1cm sieving should not be too onerous (though the recovery of the excavated material from the site maybe).

This assessment is concerned with only the contents of one trench and it is imperative to analyse material from the other excavated areas to identify spatial difference.  Initial work by the excavator has identified structured deposits in the form of hearth surrounds as found at other Atlantic Iron Age sites (Mulville et al 2003). 

This work can be undertaken at Cardiff University providing access to resources, equipment and knowledge.  Analysis will be able to shed further light on how the cattle were bought to the site, processed, consumed and disposed off. Recommended analyses are as follows.

1. Detailed analysis of species frequencies, with particular emphasis on the identification and quantification of individual animals (NISP and minimum number of individuals).  Comparisons with other island assemblages. 

2. Detailed comparison of age representation with comparable sites of the Atlantic Iron Age. The age and sex profiles will allow detailed reconstruction of which animals were selected to be bought to the cave.  This may give indications of seasonal exploitation strategies, the economy and reveal whether domestic mammals show similar age profiles as other assemblages. 

3. Detailed analysis of anatomical distribution between sites, site sub-divisions and periods in order to ascertain exploitation strategies and possible areas of industry/activity.

4. Health indicators of domesticates should be analysed, particularly with reference to the recorded occurrence of articulating pathological specimens.

5. The material appears to be deposited in-situ and detailed zooarchaeological analysis (identifying articulating elements of individual animals) linked to spatial analysis will provide details on the ‘chaine-operatoire’ of cattle processing within the cave. 

6. Biometrical analysis to provide detail on the size, sex and types of animals present.

7. Stable isotope analysis to identify the source, diet and management of the species recovered from the site.

8. Stable isotope analysis to identify the source and diet of the wolf (the potential of the remains for DNA will be explored as cave environments are ideal preservational environments).

9. Detailed microscopic analysis of the bone preservation to provide information on the state of the carcasses when deposited – en-fleshed bones have a different preservation pattern to those stripped of meat. 


Barber, J. (1981) Excavations on Iona, 1979. Proceedings of the Society of Antiquaries of Scotland 111, pp. 27-52.
Clark, K. 2006. Dogs and Wolves in the Neolithic of Britain.  Ed D. Serjeanston and D. Fields Animals in the Neolithic of Britain and Europe.  Oxbow Books.
Clutton-Brock, Juliet. 1963. “The origins of the dog,” in Science in archaeology. Edited by D. Brothwell and E. Higgs, pp. 269–74. London: Thames and Hudson
Drew, 2005. Specialist Report 2005 - The Mammal Bone Assemblage. Accessed 2/2/09
Fairhurst, H. (1984) Excavations at Crosskirk Broch, Caithness. Edinburgh, Society of Antiquaries of Scotland Monograph 3.
Hambleton, E. (1999) Animal husbandry regimes in Iron Age Britain: a comparative study of faunal assemblages from British Iron Age Sites. Oxford: Archaeopress, British Archaeological Reports (British Series 282).
MacKie, E. W. (2000) Excavations at Dun Ardtreck, Skye in 1964 and 1965. Proceedings of the Society of Antiquaries of Scotland 130, 301-411.
McCarthney, E. 1984. Analysis of Faunal Remains. In Fairhurst, H. (ed.) Excavations at Crosskirk Broch Caithness. Society of Antiquaries of Scotland Monograph Series 3:Edinburgh
McCornick, F.1981. The animal bones from Ditch 1. In Barber, J. (ed.) Excavations in Iona. Proceeding of the Society of Antiquaries of Scotland.(111) 282-380
Mulville, J. Parker Pearson, M., Sharples, N, Smith, H. and Chamberlain, A.  2003 Quarters, Arcs and Squares: Human and Animal Remains in the Hebridean Late Iron Age. eds. J. Downes and A. Ritchie Sea Change: Orkney and Northern Europe in the later Iron Age AD 300-800 The pinkfoot press, Balgavies, Angus
Noddle, B.(1978-80) Animal bones from Dun Cul Bhuirg, Iona. In Ritchie, J. N. G. and Lane, A. M. (eds.) Dun Cul Bhuirg, Iona, Argyll. Proceeding of the Society of Antiquaries of Scotland:110:209-29
Thoms, J.E. 2003 Aspects of economy and environment of north west Lewis in the first millennium AD: the non-marine faunal evidence form Bostadh and Beirgh considered within the framework of north Atlantic Scotland.  Unpublished PhD University of Edinburgh

General Discussion

The excavations in 2008 were successful in reducing the risk of erosion to the seaward face of the archaeological site and gathering data relating to site context and function. In particular, we uncovered a wide range of features relating to the use of the site during the Middle Iron Age including at least ten individual hearth settings (five hearths within each trench). The hearths show a wide range of types including kerbed, kerbed and cobbled, paved and kerbed with animal bone (see drawings below).

Hearth types identified in Trenches 1 and 2

Of particular interest with regards to the different types of hearth identified at the site is their potential function and chronology. The hearths excavated have different types of associated residues (fuel types and residues from activities at the hearths), which may indicate that differing types may have served a specific function at the site - metalworking, general fires, cooking hearths, and hearths associated with other craft skills undertaken at the site. Armit, Sharples and Parker-Pearson have all commented on the role of the hearth, especially in domestic structures such as brochs, duns and wheelhouses.

Map showing location of the Fiskavaig rock shelter

The map above shows the location of the rock shelter at Fiskavaig (black square) and other known structures of possible contemporary age. The area around Loch Bracadale contains a significant number of Atlantic Roundhouses including the broch at Dun Beag, while closer at hand is Dun Ardtreck, which was excavated by Euan MacKie in the 1960’s. Also marked is the souterrain at Cnoc Ullinish (yellow marker) and the Iron Age farmstead and associated souterrain at Tungadale (green marker) excavated by Roger Miket and Martin Wildgoose in the 1990’s. A detailed survey of the landscape surrounding the Fiskavaig site is planned this year, so we can set the site within its wider prehistoric landscape.

Excavations at the Fiskavaig rock shelter site will resume in late March and we look forward to sharing our discoveries with you on-line as they take place. So, visit the website again soon for all of the latest updates.

A view north towards the rock shelter showing the surviving portion of the talus (the grass-covered mound in this image) and the potential landing beach at bottom left

Next entry: Fiskavaig Rock Shelter 2009 - Excavations Underway.....

Previous entry: Excavations Resume at Cave of the Speckled Horses, Fiskavaig


Add Your Comment Here

Commenting is not available in this section entry.