|
The stated aims of the 2002 excavations were threefold:
The results of the excavations have addressed these items successfully:
The interpretation of the linear magnetic anomlaies in the upper part of the site was addressed by Trench 4, which proved a substantial (2m wide by 1.5m deep) ditch at the location of the anomaly. Magnetic susceptibility measurements confirmed that the ditch silts (particularly the upper silts at about 240 SI units) had a substantially elevated magnetic susceptibility compared with background (up to 120 SI units). In addition the upper stoney fills of the ditch contained substantial quantities of iron slag, which may also have had a significant contribution to the observed anomaly.
The interpretation of the major magnetic anomalies on the lower hillsides as furnaces has not been borne out by the results of Trench 2. The intense magnetic anomaly (with a peak close to 2000nT!) observed in the area of the lower part of Trench 2 appears to be associated with a dump of gravelly oxidised ore and slag, with a measured magnetic susceptibility locally of over 3000 SI units. The origin of this curious material is currently unknown, but it may be debris from ore roasting. The gravelly deposit appears to rest, at least in part, in a negative feature of burnt clay, but whether this is an ore roaster, or an earlier furnace, remains unclear at the present level of investigation.
The state of preservation of the features appears variable. On the top of the hill the ploughsoil (around 23cm) appears to have slightly truncated the archaeology, but the siting of this trench deliberately avoided areas where structures are believed to be present. In Trench 2 on the slopes, a similar depth of ploughsoil only truncated the ancient archaeological deposits at the base of the slope, where over-compaction and linear clay-filled gouges appeared to mark the flattening of the 1950s. On the higher slopes the modern ploughsoil overlay a considerable thickness of probable slope-wash deposits. These were marked by very mottled textures, particularly towards their base and this is interpreted as representing the effects of tree roots. The archaeological deposits below the slope-wash may be slightly truncated by the slope, but suggest accumulation on a much flatter buried terrace than had been expected. The deposits include large amounts of stratified debris from iron-making, but in several areas appear to rest on burnt clay surfaces, suggesting the possible preservation of earlier structures beneath.
The dating evidence obtained during the excavation largely comprised ceramics (both pottery and building materials) and appears on current evidence to be Roman. Pottery includes Samian ware and amphora sherds, indicating both an earlier date (1st - 2nd century) and a slightly higher status assemblage, than might have been expected. There was no ceramic evidence to support the expectation that the lower part of the site would be of Tudor age. Samples of stratified charcoal have been retained in order to pursue a radiocarbon date for the waste piles in Trench 2, but it seems unlikely that such a consistently Roman ceramic assemblage could be merely residual.
The redating of the lower part of the site as Roman forces a drastic reconsideration of the site as a whole. Although the earlier interpretation of the lower part of the site as Tudor seemed the most likely age, particularly in view of the large scale of the operation, there was a problem in that the documentary evidence from the 1540s (particularly PRO C78/1/74) seems to refer to only a single iron mill within the area covered by the lease (3 miles around Clun Park). The site at Mwyndy shows a distinctly different atyle of smelting (on the basis of the size and chemistry of the slag cakes, and supported by mass-balance calculations) from both the slags from Caergwanaf uchaf and those from School Road. The slag samples analyzed from the upper and lower parts of the field at Caergwanaf uchaf, were also not chemically distinguishable - a surprising fact when interpreted as differing by over a thousand years. The new evidence helps build a case for Mwyndy as the 1540s iron mill, perhaps with the mill site surveyed on HS2314 in 2002 near Pont-y-parc Farm being associated with the 1530s operation.
The Caergwanaf uchaf site seems to be emerging as a major Roman iron-making settlement. There are no other sites of this type in Wales, but there are major smelting complexes just across the border in the Forest of Dean. Here the evidence is patchy, but it would appear that early Roman (up to about AD200) iron-making may have been focused on moderately large settlements (Newent, Dymock, Ariconium, Monmouth...), but much of the ore was transported for smelting up to 50km from Dean, and iron-making may have been an integral, if small scale, part of rural activities for farms and villa estates over a very wide area. Later Roman iron-making in the area seems to have been mainly dispersed. However, the evidence for this interpretation is not particularly good. In the Weald of SE England (Cleere & Crossley 1995), Roman iron-making seems to have been dominated by large sites, often with connections to, or operated by, the fleet (Classis Britannica). Sites such as Bardown (Cleere & Crossley 1995, Fig 23) are of a similar size and structure to that at Caergwanaf uchaf. In the third great Roman iron-making area of the East Midlands there are again sites with a similar size, and with valley floor slag dumps (e.g. Crew 1998).
The setting of the Caergwanaf uchaf site is worthy of comment. It lies on the SW side of the Ely, on the opposite bank from the likely ore sources, but on the same side as the closest Roman town, at Cowbridge, some 7km to the SW. Recent excavations in Cowbridge (Parkhouse & Evans 1996) have revealed substantial evidence for 2nd-4th century iron-working (800kg of smithing slag from the Bear Field site). A connection between the large amount of urban (or semi-urban) iron-working and the large iron-producing site at Caergwanaf uchaf is possible.