A Scottish hillfort and adjacent rock carving (Part 3/4)

We’re very pleased to be able to bring you, in serialised form, a paper courtesy of Dr Euan MacKie regarding the Sheep Hill hillfort and associated rock carvings, which are currently under a long-running threat of destruction. Please read the Introduction first. Part two discussed the Greenland rock carvings. We now move on to consider the Sheep Hill forts.

The Sheep Hill forts

Timber-framed hillforts tend to be situated on the north-east, south-east, south-west and western fringes of the Highland zone and Sheep Hill is one of the few close to the urban areas of Central Scotland (MacKie 1976, fig. 6). In addition an analysis of the strongholds by size shows that most are true hillforts, even if occasionally quite small (MacKie 1976, 233 ff). There are three situated within the highland zone, where arable land is in much smaller patches than in Aberdeenshire, which are of roundhouse size and enclose from about 100 to 200 square yards. The primary stronghold at Sheep Hill is not as small as that and encloses about 1350 square metres but it is better classed as a largish stone dun rather than a hillfort; it was probably roofed along the inside wall face and was doubtless suitable for the residence of a chiefly extended family and its followers. The destruction of this dun by fire and probably by later demolition was complete; a small length of vitrified wall core was found on the south-west margin of the area enclosed (marked ‘v’ in Illus. 5) and this gave a clue to how far the dun had extended in that direction; it seems only to have occupied the summit of the hill, an area marked in yellow on Illus. 5. A trench across the probable northern part of the wall, at the top of the north cliff of the hilltop, failed to find a clear occupation layer.

My limited excavations of 1967-69 revealed that a larger and more complex hillfort defended by ramparts, was built after the dún was destroyed and that masses of the heated stones from the original building were used in it. In 1969 there was no reason to study the geological origin of this debris but I recall that it was – as one would expect – mostly of basalt fragments. The later hillfort is not easy to date (as explained below two C14 dates for it probably relate to the first fort) but it may well have gone out of use when the Roman Army established itself on the north bank of the Clyde nearby. Finds were fewer than in the dún midden (below) but included a large number of fragments of shale bracelets, as well as many pieces of shale which showed that a bracelet workshop existed on the site – a very rare find. Shale ornaments can be given a black shiny surface by polishing and were popular in the Iron Age.

Illus. 5. Contour plan of Sheep Hill showing the positions of the fortifications and enclosures. The approximate outline of the Phase 1 vitrified (timber-framed) dun is in yellow and the ramparts of the Phase 2 hillfort are in blue. The southern entrance of the second fort is protected by a triangular annexe with its own entrance, which is itself protected by a short traverse. The excavated trenches are marked. The wall of both fort seems to have run along the same track at the top of the steep northern edge of the Hill.

The southernmost rampart of the hillfort was cut through and revealed a midden underneath its core, which evidently continued in use after vitrification, presumably by survivors who seem to have camped on the southern part of the hill top for a while. The reason for supposing this, and for not inferring that the midden was simply a refuse tip for the dun, is that a rim sherd of gritty Dunagoil ware was found in the midden which had itself been partly vitrified by intense heat; in other words the sherd had been heated when the dun was set on fire and partly vitrified and was then swept into the midden with other later material. The date of the midden was ascertained by two methods – a C14 date and dateable mould fragments.

Firstly another Dunagoil ware rim sherd was recovered of the kind of Early Iron Age carinated pottery that has been well dated to about 700-500 BC in Orkney (MacKie 2008, 265). Organic residue on this sherd gave a date of 2490 ± 30 bp, or 780-500 BC (SUERC-26969) at a 92.1% level of confidence).

Illus. 6: rim sherd of Dunagoil ware from the midden under the rampart on the southern edge of Fort 2. It has been severely burned and vitrified into a coke-like substance, presumably when the timber-framed dun was destroyed by fire.

This seems likely to reflect the destruction of the timber-framed dun. Secondly there were fragments of moulds for casting bronze tools and weapons in the midden, including types consistent with the Ewart Park phase of the Scottish Late Bronze Age metalwork sequence, which is dated to about 1000-800 BC. (Cowie forthcoming) (MacKie 1976, Fig. 3. This suggests that the bronze moulds date from a an earlier time than the carinated sherd – to one when the timber -framed dun was intact and occupied. The midden was therefore probably used both during and after the occupation of the dun.

That the actual construction of the timber-framed dun took place during the Ewart Park phase of Late Bronze Age metalworking is shown by one of two more C14 dates obtained in 2012 from charcoal which was actually found on the floor of the second hillfort, just within its southern main rampart. Since one date was slightly older than that for the stratigraphically earlier sherd in the sub-rampart midden it would seem to follow that large amounts of charcoal from the burnt wooden elements of the original dun were strewn about the site and became incorporated into the later floor levels. The charcoal could either be from wooden parts of the original building or from later constructions; the extensive timber-framing of the stone wall surely makes the former more likely (all date spans are expressed in terms of a confidence level of 93.5%). The dates are 2485 ± 30 bp, or 780-490 BC (SUERC-20968) and a more accurate one of 2605 ± 30 bp, or 830-760 BC (SUERC-26966). In addition a charcoal sample resting on rock, and which probably belongs to an earlier occupation (perhaps a palisaded enclosure ), gave a date of 3095 ± 30 bp, or 1440-1290 BC (SUERC-26967). All things considered a date for the construction of the timber-framed dun in about the 9th century BC, or perhaps the 10th, seems most probable.

Next: Were the two sites linked?