The workshop was a rectangular, open-sided, post-built structure. The visual appearance of the building was originally designed to blend with that of the adjacent Celtic Village, with its reproduction Iron Age roundhouses. No Iron Age smelting site has yet been excavated in South Wales, but Roman iron making buildings have been investigated in the Forest of Dean. Of these, the best known is at Woolaston (Fulford, M.G. and Allen J.R.L. 1992. Iron-making at the Chesters Villa, Woolaston, Gloucestershire: Survey and Excavation 1987-1991, Britannia, 23 , 159-215), and in Phase 3 (c. 250-370?) at least, the iron-making activity there was within a 6-bay padstone building (16.5 x 8.2 m). This building contained evidence for two sets of smelting furnaces at either end, together with charcoal storage, and a central ore-crushing area. This provided a model for our much more modest 6 x 4m, 2 bay structure, housing a single smelting furnace at one end, and a smithing hearth at the other. The original (1998) posts decayed and the rather unattractive red tarpaulin roof was later carried on scaffolding. It is hoped to rehouse the smelting facility in the not-too-distant future.

Smelting furnace: the smelting furnace was based on the design of furnaces operated with great success by Peter Crew (Crew, P. 1991. The Experimental Production of Prehistoric Bar Iron, Historical Metallurgy, 25, 21-36). It is a low shaft furnace and, as originally built, had 20 cm thick walls, a bed depth of 50 cm, an internal diameter of 40 cm, and a blowhole with a gentle inward inclination, opening 23cm above the base of the bowl. A tapping arch lies at 90° to the blowhole. The furnace was subsequently modified on several occasions, most notably during the winter of 2002/3. The final build of the furnace had internal diameter of approximately 34cm, a wall thickness of greater than 30cm, and a bed depth of approximately 1m.

Smithing hearth: The original hearth was formed by a horseshoe shaped clay wall, rising to a height of about 20 cm at the closed end, with a blowhole slightly inclined inward at 10 cm above the hearth floor. This hearth proved too insubstantial for bloom reheating and a new hearth was built in autumn 2003, with a more enclosed shape and higher walls.

This second hearth (seen here) was subcircular, with the walls slightly higher at the blowhole side. The internal structure of the hearth was strengthened by including bricks within the clay wall, intended to enable it to withstand more wear-and-tear. The blowhole was formed using a section of steel scaffold tube, which did not reach the inner face of the wall. The hearth has been used to reheat a substantial (2.5kg) billet of iron, and was capable of bringing this to welding heat using the larger of our pear-shaped smith's bellows.

Air supply: we purchased two pairs of old pear-shaped smith's bellows at the start of the experiments. These bellows are of the double bellows design, certainly in existence by the 16th century (Georgius Agricola, 1556. De Re Metallica). One has boards 54 cm across, the other 71 cm (upper image to right). Our working estimates suggest that the volume of the lower bellows of these, through which the air is drawn, are 33 and 66 litres respectively. By late 1998 it became apparent that the air flow from these bellows was too low, and a large set of single action bellows was designed. A mock-up of these was employed for smelt 18 (1998) and an authentic pair (lower image to right) produced for smelt 19 (1999). In 2003 the decision was made to employ electric blowing to allow experimentation with higher blowing rates. A bouncy castle blower was hired, and following favourable results in smelt 23, one has now been purchased. The larger smith's bellows are still employed for smithing work and it is hoped that once appropriate blowing rates have been established, then we will build additional single action bellows to allow "authentic" blowing during smelting too.