Breckness Broch: The Excavation of an Iron Age well

Regular monitoring of the coastline north of Stromness, Orkney from the 1980s by the authors (Ballin Smith and Ballin) showed that erosion was slowly altering the cliffscape and that this erosion accelerated with the severe winter and spring storms during 1992-1993.

Comparison of photographs taken in the summer of 1993 with those of a year earlier indicated that about one metre of cliff had been taken by the sea in the vicinity of the Iron Age settlement and broch at Breckness.

A well, located below the fragmentary remains of the broch, was first revealed in the cliff section in 1992. One year later it was half removed by the sea and lay slightly seaward of the newly receding cliff line.

This paper examines the finds from the well, as the details of the excavation have been published elsewhere (Ballin Smith 2002).

Following the coastline approximately four kilometres north and west from Stromness, a small promontory can be found which contains evidence of occupation from two thousand years ago and burials from the late medieval period (Fig 1).

The broch at Breckness has been a settlement of antiquarian and archaeological interest since the latter half of the nineteenth century.

Prominently situated on the coast beneath the stone dykes enclosing the seventeenth century Bishops House, the partially surviving arc of Iron Age broch masonry sits on soft bedrock and faces the sea (Fig 2).

Until the middle 1980s the broch was reasonably well hidden beneath a turf covering, but a land slip allowed occasional human skulls and bones to be found on the beach by the general public and indicated the renewed erosion of the cliff section and settlement (unpublished report on the human remains by Daphne Home Lorimer), aided by the effects of rabbit burrowing.

Visits to the site by the authors in the summer of 1993 showed that features beneath and to either side of the broch had been newly exposed by the sea.

These were a well, situated beneath the broch, and cut into the bedrock (Fig 3); a flat bottomed ditch 27 metres to the west also cut into bedrock (Fig 4); and a contemporary or later Iron Age settlement with midden layers was exposed in the receding cliff line to the east.The details of these features can be found in Ballin Smith 2002.

The well had been uncovered during the erosion of the cliff and the removal of some of the interior deposits of the broch.

Some of the well fill and one of its sides had been partly taken by the sea and the feature lay somewhat seaward of a cliff face that had moved inland. Its situation and its remaining fill were clearly threatened by the next gale, if not the next high tide.

Some years earlier the well at Warebeth Broch located below the Stromness graveyard, on the same stretch of coastline, was exposed in similar circumstances. Its contents were excavated and its structure recorded in advance of coastal protection (Bell and Dickson 1989).

With the ever present threat of erosion by water and wind, and the potential loss of information on the Iron Age contained within the well, it was decided to excavate the structure.

If the feature was similar to that encountered at Warebeth Broch there were possible health and safety issues for members of the general public coming across it accidentally.

It was thought safer to remove the remaining contents of the well, to record its construction and report any safety issues to the local council, than to leave its contents and the structure partly exposed and in a possibly dangerous condition.

The excavation of the well took place over the course of one day on the 24 July 1993 (Fig 5).

The removal of one side by the sea allowed access into the feature. The top had also been removed by the waves, and the overlying stratigraphy from the broch interior ended abruptly at the landward well wall.

Approximately one metre of deposits survived in the bottom of the feature and these were excavated by hand.

As the excavation was unplanned the recording of the work was by written record, measured sketches and by photography. Finds and samples were collected.

The well was a rock cut parallelogram which measured 1.7 by 2.6 metres in plan (Fig 6).

It had been excavated out of soft laminated sandstone, and achieved a height of approximately two metres at its surviving north-west corner.

A small area of masonry lying above its north wall indicated that the bedrock had been partly built up at this point. This evidence also suggested that the well roof may have been corbelled.

During the removal of the lower levels of fill an almost intact square roofing slab, 0.95 by 0.95 metres was found together with other larger stones on the floor of the well (Fig 7).

One edge of this slab had been chipped to straighten it and two opposing notches, one at each side, had been cut close to one end (Fig 8). The evidence for the well access in the south-east had been removed by erosion.

The well, now resembling a walk-in cupboard with the door missing, was emptied of its remaining fill of earth and rubble which had already been disturbed by high tides.

The earth infill of the feature was loose at the top but became progressively heavier and stickier towards the base of the well. At the start of the excavation there was little indication of the depth of the feature and how its lower levels were constructed.

However, during the removal of its fill, the floor of the well was encountered on a level approximate with that of the beach. Beach stones, bedded in a thin layer of clean clay and small stone chips paved its floor, but no additional constructional complexities were found within it.

Extensive primary burning in the south-west corner of the well, including the floor, suggests that fire and water was probably used to remove the rock to create the cavity.

The severe lamination of the well floor and sides may have been problematic and been the cause of the laying down of the 50 mm thick stone paving set in clay. The floor sloped gently from east to west.

The finds that were retrieved during the excavation were located throughout the well fill. They

included bones of bird, land and sea mammals with some human remains, stone and bone artefacts and sherds of pottery (Ballin Smith and Ballin 1993 DES).

All the finds had become incorporated into the well fill as a result of the well roof collapse, and from possible subsequent slippage and erosion of stratigraphy within the broch and from deposits resting above its exposed walls.

In the cliff west of the broch, the profile of a U-shaped, rock cut ditch was also exposed by the gales (Fig 4).

This feature is interpreted as part of the Iron Age settlement defences and was also recorded at the time of the excavation of the well, but not excavated. Finds were retrieved both from the exposed ditch section and from midden layers to the south-east of the broch uncovered in the cliff face (see below).

A total of 14 small finds and 22 samples were recovered from the fill of the well. In addition there were four small finds from the ditch and one find from the midden east of the broch. Table 1 indicates the range of material that was found during the excavation (a fuller description of the finds is given in the Archive Catalogue here).

small find nr	category	location
		well	ditch	midden east of broch
SF 1	Pottery		2 joining body sherds
SF 2				3 body sherds
SF 3		1 large body sherd
SF 4		2 joining body sherds
SF 5		1 body sherd
SF 6		1 body sherd
SF 7		1 base edge sherd
SF 8		1 body sherd
SF 9		1 large body sherd
SF 41			2 body sherds & 1 rim
SF 44			3 body sherds
SF 10	Bone artefacts	Fragmentary point
SF 11		Borer
SF 12		Polished borer
SF 43			Worked point
SF 13	Stone artefacts	Pot lid
SF 14		Hammerstone/rubber
SF 15		Hammerstone/pounder
SF 42		Small bead
SF 16	Charcoal	Unidentified
SF 17	Marine shell	Cowrie shell
SF 18		Fragment of dog whelk
SF 19		Fragment of lobster/crab
SF 46		Coral fragments
SF20-28, 30-40	Animal remains	Bones & antler
SF 29, 45	Human remains	Bones & teeth

Pottery

Eight sherds of coarse pottery were found in the fill of the well, with an additional seven sherds from the ditch and three sherds from the midden east of the broch. These sherds closely resemble each other and also assemblages from other broch sites on Orkney, such as Howe (Ross 1994).

The sherds are relatively small and none weighs more than 53 g: the total assemblage weighs 303.5 g. All the sherds have large, heavy temper that shows through a thick slip that was applied mainly to the outer surfaces of the vessels (Figs 9a & 9b). There are no fine wares representing pots from later periods, as the average thickness of the sherds is c 9 mm. Several fragments show evidence of sooting or carbon deposits, indicating their use as cooking pots, or vessels used near a hearth.

There are few diagnostic sherds as only one fragmentary base-edge sherd came from the well and a partial rim sherd from the ditch fill. No decorated sherds were recovered. All are typical Middle Iron Age wares that would have been in use during the time of the broch and during the filling in of the ditch.

The vessels would have been flat-bottomed pots with quite wide mouths tapering gently to their bases (see vessels on display in The Orkney Museum, Tankerness House).

The sherds that came from the well fill are likely to have derived from either the floor of the broch or from deposits immediately above it.

The lack of evidence for later wares reinforces the hypothesis that the well roof collapsed during the main occupation of the broch.

It is interesting to note that three bone points or borers (from antler and other bone) were found in the well together with pieces of uncut and cut red deer antler fragments. These artefacts indicate the remains of activities which may have taken place within the broch at the time of the collapse of the well roof, such as leather working and bone or antler working.

Two of the borers were complete hand tools surviving over 110 mm long, with high polish on their points indicating a long period of use (Figs 10, 11a & 11b).

These may have been everyday tools used for making holes in, for example, skins and leather. A sliver of bone, also used as a bone point was found in the well and another from the ditch fill (Fig 12). Both these tools were cut pieces of bone that had been roughly shaped by a knife, and were perhaps used for the purpose intended and then discarded.

Bone points and borers were commonly made from the long bones of sheep and similar tools are frequently found on broch sites. They are however, not indicative of any one particular period, as the same types of tools were also used during the Neolithic at settlements such as Skara Brae.

The occurrence of antler fragments (Fig 13) and butchered bones implies that red deer were still present on Orkney and that they were either hunted or managed for their resources of meat, hides and antler (Smith 1994, 139-153).

Antler, as seen at other broch sites, was a versatile material used for handles for iron knives, for comb making, weaving combs etc (see Bone Artefacts in Ballin Smith 1994, 168-185).

Only three stone artefacts were found during the excavation of the well. These were a small circular pot lid (Fig 14), and two hammerstones made from beach cobbles.

One hammerstone had also been used as a grinder (Figs 15a & 15b), the other a pounder (Figs 16a & 16b), perhaps used in food preparation. These tools are typical finds from broch sites but are not datable in themselves.

A single, round and flat bead (Fig 17), identified as limestone or marble, was found in the midden east of the broch. Its size and form are similar to glass beads found on other Iron Age sites (see Henderson 1994, 235).

However, stone beads, usually larger in size but of various shapes and raw materials are commonly found on Iron Age sites (see Stone Artefacts in Ballin Smith 1994, 192).

The eroded well at Breckness Broch contained a number of animal bones in a very good state of preservation. The mammal species identified from its fill were: cattle, sheep, pig, red deer, dog, cat, fox, seal species and rabbit. Bones of birds were also found: domestic fowl (Gallus gallus), rock dove (Columba livia), gannet (Sula bassana) and guillemot (Uria aalge). Four fish bones were also recovered.

The total numbers of bones and an estimate of the minimum numbers of individuals from each species are shown below in Table 2. In terms of fragment count, bones of sheep were the most numerous (38 fragments which were definitely from sheep, plus 137 ribs and vertebrae which were designated as small ungulate but were more likely to be from sheep than any other species).

Cattle bones were only poorly represented, accounting for only three definitely attributable bones. Vertebrae and ribs designated large ungulate were more likely to have come from red deer than cattle, since 55 bones were identified as coming from the former species. The red deer bones are from large animals, similar to those found at the Iron Age broch at Howe, and are probably of early date.

Butchery marks were found on the bones of cattle, red deer and domestic fowl only, indicating that the meat from these animals had most likely been used as food. However, in the case of the sheep, none of the bones showed any evidence of having been butchered, and this, together with the presence of complete small ungulate/sheep ribs and vertebrae would seem to indicate that whole carcasses had been dumped down the well. This may also have been true for the cat and fox bones.

All the wild bird species could have been eaten, but there is no proof that this was so at Breckness, since there was no evidence of cut marks on any of the bones.

The presence of rabbit bones in the well is problematic, since the species was not introduced to the British mainland until the medieval period. They are thus likely to be modern intrusions, particularly since no butchery marks were observed on them, derived from burrows in the cliff section.

The single bone of a pig was in a markedly poorer condition than the rest of the collection and may also have been of a more modern date.

As regards the age of the animals, six out of seven of the sheep appear to have died before the age of two years, on the basis of both dental and epiphysial fusion evidence; one animal died at a very young age, possibly between two to six months.

As at Howe, very young red deer were present as well as adults; at least half were juvenile or juvenile/immature.

One sheep mandible showed evidence of infection associated with heavy calculus deposits; another mandible thought to pair with this specimen was also affected by calculus, although to a lesser degree.

The anterior medial malleolar facet of a red deer tibia was pitted and eroded, perhaps because of an arthritic process.

Species	Number of bones	Minimum nr of animals
Cattle	3	1
Sheep	38	7
Pig	1	1
Red deer	55	4
Dog	2	1
Cat	14	2
Fox	29	1
Seal species	1	1
Rabbit	17	4
Large ungulate	36
Small ungulate	137
Indeterminate mammal	30
Domestic fowl	4	1
Rock dove	7	1
Gannet	2	1
Guillemot	1	1
Fish	4
Total	381

Fragments of eight bones from a young adult were found in the upper layers of the well fill.

Other remains found in the well fill include small fragments of charcoal (unidentified), possibly derived from the broch interior, and fragments of sea shells and lobster or crab. Due to the coastal location of the broch these latter remains are likely to be modern intrusions due to the action of the sea and erosion of the well.

The well was located centrally beneath the surviving remains of the broch tower floor and had been presumably served by stone steps situated at its eastern side with access most likely gained from the broch floor itself (Fig 18).

The top of the bedrock into which it was cut was levelled by the addition of masonry. This stonework, which was likely to have been corbelled and then capped by the single large flagstone found in the base of the well, formed its roof. The square roof stone was notched, at one or both sides, indicating it may have been held securely across the aperture leading down into the well. The lack of wear on either face of the stone suggests that it was lifted away from the aperture rather than dragged to one side.

The orientation of the large flagstone in the well fill indicated that its roof had collapsed inwards, most likely from a south-easterly direction. Deposits of in situ hearth material on the stone indicated that part of the broch floor deposits had subsided into the well along with its roof. From this evidence it is assumed that the collapse occurred during the occupation of the broch tower. The occurrence of stone and bone artefacts and some sherds of pottery are likely to have derived from the floor surrounding the well aperture.

Some of the rubble surviving in the base of the well cavity may also have been the result of collapse of its access, presumably a staircase. It is not known whether the well collapse was the result of its own structural failure or that of part of the broch itself. Whatever the cause, no indication of further occupation above the well was noted, nor was there an attempt to clean it out or repair its roof. Indeed there is evidence from the analysis of the animal bones that sheep carcasses and red deer bones, as well as cat and fox bones may have been dumped in the well sometime after its collapse. Unfortunately the full sequence of events concerning the deposition of material within the well cavity is limited by the removal of some of the deposits by the sea.

The collapse of the well and its being filled with evidence of domestic occupation may have been the result of a subsidence within the broch. This collapse may have been so severe that it probably led to the broch tower’s abandonment. Watt (1905) recorded that the walls of the broch were exposed but there was no evidence of it ever having been excavated. It is possible that part of the seaward arc of the broch slipped off its foundations leaving only the landward remnant secure. There is no record of any previously located finds from the broch or its settlement seen in the cliff sections to either side. The lack of rubble within the remaining arc of broch tower masonry may not be solely due to erosion, but to an early abandonment and possibly stone robbing. The evidence points to the broch tower having had a limited use with the debris of later settlement not accumulating within it. However, the entrance to the well may have remained partly open allowing for the later accumulation of bones of pig, birds and rabbits. The integration of human remains, into the upper well fill, is likely to have occurred from the erosion of later medieval graves dug into what is now the top of the cliff. Bones will have tumbled down the cliff slope and come to rest in hollows and cavities. The erosion of the graves is not entirely due to wind and wave action, but also to the burrowing of rabbits.

The well fill also included finds of pottery, bone and stone which derived from the floor of the broch and typify some of the work and activities that were carried on there. Meat from cattle, red deer and domestic fowl was eaten; bones from butchered animals were used for tools for working hides; and cobbles picked up from the beach were used either for dehusking grain for food preparation, or for the roughening of worn querns, and for repairs to the stonework of the broch and its other buildings. The rare bit of charcoal indicates wood was gathered for fuelling fires, if not for other domestic purposes. The few finds retrieved from the exposed section of broch ditch to the west are similar to those found in the well. The small pieces of evidence together add up to a picture of life within the broch that was not dissimilar to other Iron Age settlements along the Orcadian west coast.

The relationship of the ditch and the broch can be complex as seen at Howe (Ballin Smith 1994), and the reasons for, and methods of, the infilling of the ditch have not yet been determined. The relationship of the settlement lying to either side of the broch, to the tower itself, is also not fully understood. The number of structures and midden layers, as well as the few finds, suggest a contemporary and later use of the site. But the extent and date of that settlement remain untested.

One question not posed by this publication, but discussed previously (in Ballin Smith 2002), is whether the well was in fact a well. Was it originally a simple wet cavity beneath the centre of the broch or did it have an entirely different function as for example the souterrain at Howe? The occurrence of structures beneath broch floors such as at Gurness and more recently in the mound at Mine Howe, suggests that not all Iron Age subterranean chambers were wells. The position of the Breckness feature and its simple construction, suggests that it was a well, like the one at Warebeth (Cemetery) Broch.

What is important are the series of events that led to the collapse of the well, and presumably the abandonment of the broch. Breckness Broch has been actively eroded away by the sea, but did part of it collapse during its occupation? Structural failures of brochs have been recorded from Gurness, Midhowe and Howe among others, and they may have been a common occurrence. The lack of deposits within the broch at Breckness recorded in the nineteenth century by Watt may indicate that its occupation was not long. It may not have been used for later settlement or rubbish dumping as at Howe. The extensive remains of what are presumed to be contemporary and later settlement at Breckness, now visible in the eroding cliff faces, suggest that the site was not entirely abandoned and that life carried on well into the first millennium AD. Indeed by the medieval period the broch had largely disappeared under accumulated deposits of occupation and sand, and that the area above it was eventually chosen for a burial ground and the location for Bishop Graham’s house in the seventeenth century (Fig 2).

This paper is dedicated to Daphne Home Lorimer who has always taken an active interest in the human remains found at Breckness. A picture remains in the mind, of Daphne some twenty years ago eagerly trying to locate the graves in the unstable cliff face above the broch. Barbed wire fences and stone dykes did not curb her enthusiasm. Daphne has on various occasions analysed the human remains from this site, but her reports are unpublished. Further archaeological investigations are required at this area for her results to be put into context.