Llwyn Bryn-dinas (SJ 172247; Clwyd Archaeological Record CAR 999) is the only hillfort of substance in this part of the Tanat Valley, lying almost midway between the major forts of Craig Rhiwarth and Llanymynech Hill (fig. 1). Craig Rhiwarth, enclosing an area of about 35 ha/86 acres, lies above Llangynog, about 12 km to the west, where the head of the valley cuts into the rugged uplands of southern Clwyd. The very large hillfort on Llanymynech Hill (56 ha/140 acres) lies about 9 km to the east, near the confluence of the Tanat and Vymwy, which feed into the Severn a little furtherdownstream. Two smaller neighbouring sites of later prehistoric type are known from aerial photography - a small double-ditched enclosure known as Plas Uchaf, on a hillock at the very foot of Llwyn Bryn-dinas (SJ179245; CAR 1002), and a square double-ditched enclosure on the banks of the Tanat about 1.5 km to the south-west (SJ 156245; CAR 1433); both enclose about 1 ha/2.5 acres. The site now lies within th~ community of Llangedwyn, Clwyd, but before Local Government re-organisation in 1974 it fell within Llanrhaeadr-ym-Mochnant, Denbighshire.

Several short seasons of excavation were carried out on the hillfort in the years 1954-56 by Mr L. M. Stephen, a schoolteacher, with the help of pupils from his school at Wolverley, Worcestershire. It appears that he may also have returned to the site in 1958, after he had taken up a new appointment at Dover College, Kent. The results were never published, but the work was mentioned briefly in print by Dr H. N Savory (Savory 1976, 260-1). Following the excavations by the Trust in 1983, Mrs Stephen kindly made available two notebooks, a series of photographs, correspondence, typescript notes, and a small box of finds from her late husband’s effects. Dr Savory also helped by providing various notes and correspondence relating to the earlier work.

The line chosen for the access track cut through the rampart at an angle of about 30° at a point where it had already been partially eroded by animals, just north of the entrance gully on the eastern side of the fort. The uneroded parts of the rampart here stood little more than 50 cm high above the interior, and even allowing for the accumulation of material behind the rampart it seemed unlikely that the total depth of core material would be more than about a metre. It was thought that a section about 3 m wide could be dug in about a fortnight with a small team seconded from work elsewhere, allowing the basic structural character of the rampart to be recorded, testing the underlying ground surface for the possibility of earlier (palisaded?) lines of defence, and seeking dating material from either the rampart core or the deposits behind or beneath it. The presence of the external ditch might also be tested, either in the excavation trench or in the subsequent bulldozing work for the access road.

See main report in printed text. The following sections give a full catalogue of metal-working debris from the site, together with descriptions and analyses of different categories of material.

See also descriptive text in printed report. The additonal notes which follow are of analysed items only.

21. Examination during conservation (conservation report on Lab. No. 3543 by Kate Hunter dated 8 March 1988, a copy of which is lodged with site archive) had suggested that the pin had been manufactured by fixing a flat, rimless iron disc several millimetres from one end of the shaft, pushing the decorative head onto the shaft as far as the iron disc and then slightly flattening the end of the shaft in order to secure the decorative head. Microscopic examination of the decorative head showed that it had no clearly defined internal structure, and that extensive decay had caused some lamination, particularly near the upper surfaces. Initial spot tests with the reagents cacothalene and potassium iodide gave negative results for tin but a good positive result for lead, and although these tests are not wholly reliable they suggested that the decorative head was of corroded lead-based enamel or glass. Spectrographs analysis, of which the results are given in the following table, shows that the decorative disc head is made of tin.

The numbering continues that given in the descriptions of illustrated crucibles and crucible fragments (fig. 4) in the printed text. See also full catalogue of metallurgical debris in Table 2.

32. Crucible body fragment, almost certainly from crucible no. 31 (6.00 g), showing relining; 25 x 22 x 20 mm. Context 08. Ref. LB-OX-125.

33. 1 body fragment from crucible no. 31 (2.2 g); thin vitrification only, outer layers perhaps having spalled off; 22 x 14 x 9 mm. Context 09. Ref. LB-OX-123.

Analyses of the following items are given in microfiche Table 6, where details of sample and context numbers are also listed. See also full catalogue of metallurgical debris in Table 2.

44. Irregular flat piece of bronze, 13x12 mm; as cast medium tin bronze; corroded away; dendrite arm spacing = 30 40μ, suggesting a moderate cooling rate; interdendritic distribution of sulphide clusters, all basically Cu2S, although some contain a few percent iron; massive surface corrosion with islands of α and δ; fragments of charcoal and small areas of typical melting slag adhering, some possibly having originally floated on the surface of a melt. It is reasonable to suppose that this piece could have been the residue left in a crucible after a pour and had set there and been knocked out later, with charcoal and dross still embedded in the metal. See also analysis of sulphide inclusions in microfiche Table 7.

Analysis of the following items is given in microfiche Table 9, which also gives details of sample and context numbers. See also full catalogue of metallurgical debris in Table 2.

50. Hammer scale (0.05 g). Microstructure: impossible to stabilise, therefore only a rmall amount polished. Massive oxide (mainly magnetite) and a Tayalitic glass* but containing Al and P as well as the elements expected of fayalite.

51. 5 flakes of hammer scale.

See analyses of the following items in microfiche Table 9, where details of sample number and context are also given. See also full catalogue of metallurgical debris in Table 2.

52. 5 fragments of lining material (18 g).

53. 8 fragments; probably pieces of vitrified lining material with Cu stains (8 .00 g); possibly just over-fired crucible.

54. 15 fragments of fuel ash slag (8.3 g), some copper stained. Analysis indicates copper-working slag.

The following slags are described separately: see also full catalogue in microfiche Table 2. See analyses of each of the following items (with the exo tptioi of nr 69), including occasional metallic inclusions, in microfiche Table 9, where details of sample number and conte... are also given.

55. Moderate-sized plano-convex,slag (type 4), with moderate porosity and few charcoal impressions (174 g). Microstructure: moderate-sized semi-dendritic iron-oxide FeO (40-60%) inhomogeneously distributed, in matrix of fayalite and fayalite plus iron oxide pseudo-eutectic. A little ternary glass+fayalite+iron oxide eutectic present. Some regions of pure fayalite; others with large laths of fayalite and iron oxides outlined by giass and iron oxide eutectic. A little metallic iron present.

56. Fragment of dense plano-convex slag (50.5 g). Patchily distributed iron oxide (FeO), locally massive (100%), and distorted. Dendrite arm size and spacing varies through fine-moderate-large arms. Matrix of fayalite and unresolved eutectic. A little metallic iron.

57. Piece of dense plano-convex slag (37 g). Moderate porosity. Metallic iron common in regions with less iron oxide. There are two regions: one with pools of massive semi-globular FeO with 90+% iron oxide (wustite), interspersed with regions with moderate or fine dendritic FeO in a matrix of unresolved material. Metallic iron contains several percent copper!

This limited study was principally concerned with the identification of wood charcoal from the metal-working floor and bowl-hearths, in order to identify the timber species exploited for fuel in the metal-working process, although other charred plant remains present in the samples were also identified.

Samples from site had previously been treated to extract the charred material and separate the cereal and weed seed remains from the charcoal present.2 A sample of plant remains taken from the sealed soil profile (38) beneath the rampart was also examined; it contained small-seeded legume species, cuscuta, grass rhizomes and fungal sclerotia and cleistothecia, which might be found in any soil.

Identification of a piece of wood charcoal requires its fracture in three planes (Leney and Casteel 1975). The fragments thus obtained were examined under an epi-illuminating microscope and identified by comparison with modern charred reference material and an atlas of wood anatomy (Schweingruber 1982). The minimum size of charcoal piece from which the three planes of fracture could be obtained was relatively large, so that many of the pieces in the samples were too small to be analysed. All samples contained at least twenty identifiable pieces, however, mainly of twig or branch wood.

The following radiocarbon determinations were measured in the Radiocarbon Dating laboratory of the Department of Geology, University of Wales College of Cardiff. Calibrations are derived from the University of Washington, Quaternary Isotope Lab., Radiocarbon Calibration Program 1987, Rev. 2.O.