S5003131

8

Experlments with the larger shaft furnace were still unaucceaaful even after using two tuyeres and a total airflow around 500 litres/raln (Tylecote & Boydell 1978). However, results were beglnning to improve when t his progranune was dlscontlnued. A smali quantlty of molten slag was reported to have leaked through the front wali of the furnace durlng the last smeltlng attempt. A sultable tap hole and procedurę were still requ-lred. It was suggested slag tapplng could not be accompllshed away from a tuyere. Problems existed with burnlng rates and heat losses. The furnace bottom was too cold to maintain molten copper and slag. Much morę work was required with the shaft furnace. Successful operatlon demanded Solutions vcry different from copper smeltlng in a bowl furnace.

In morę recent years several other new attempts at ancient copper smeltlng techniques have been undertaken. A single heating experiment was conducted at Tlmna using a smali shaft furnace and natural draft (Bachmann & Rothenburg 1980). The reconstruction was pear-shaped and built of clay. No ore or flux was charged. The point of the experiment was solely to determine internal furnace temperatures which reached 1260*C.

Based on recent archaeological evidence in China, Hua and Lu (1981) reported on their experiments with copper smeltlng. A shaft furnace was built of clay and used to reconstruct the process. Few other details are presently avallable (Fig.4).

(b) Iron

Perhaps because the bloomery process was still being carried on in the Pyrenees in 1840 (Percy 1864) and in New York State with hot blast in 1889, the early experimenters could hardly claim that the bloomery process was extinct. But by the 1930s metallurgists examining the products of the Roman bloomery process some nineteen hundred years earlier were well aware that they were dealing with something that was not the product of the iron blast furnace and was clearly obsolete.

In order to understand the process producing these materials it was necessary to do simulation experiments. Perhaps the first were carried out by Gilles in Germany, who, being convinced that the earliest furnaces were blown by induced draught, erected a shaft furnace 1.72 m high with a diameter of 0.9 m. There was a single (30 cm diameter) entrance for air at the bottom which could be controlled. These experiments were continued on shaft and hearth furnaces and were reported in 1958 and 1960. Fayalite slags containing iron lnclusions were produced. The maximum temperaturę reached was 1400*C.

The Iow bloomery hearth was based on that shown by Agricola and blown with a blast of air through a 32 mm diameter tuyere. It was rela-tively unsuccessful when worked with a mixed charge of charcoal and ore.

In fact at this time two groups of workers, Gilles (1958 and 1960) in Germany and Wynne and Tylecote (1958) in Brltaln, were coming to much the same conclusions that a Iow shaft furnace had either to be worked on the Catalan hearth prlnciple with ore going in one side and fuel down the other (tuyere) side (Flg.5),or 'manipulated' in such a way as to keep the reduced iron from being reoxidised by the blast from the tuyere.

The only alternative, as Wynne and Tylecote (1958) and Gilles (1958 and 1960) found, was to convert the Iow hearth into a shaft furnace by puttlng on top a reservolr for the twin purposes of holding the charge and keeplng the heat from being lost by radlation through the top.

Durlng this tine Profesaor H J 0'Kelly who had been excavating iron smeltlng aitea in Ireland had been carrying out sonę snelting experinenta which he reported at the Hamburg Conference in 1958. He concluded that the bowl furnace with a depth of 20-50 cm and a diameter of 20 cm at the aurface was the commonest type in Britain and Ireland (0'Kelly 1961).

The word 'bowl* used in connectlon with iron snelting has been generally accepted aa applying to a Iow hearth where the height/wldth ratio doea not exceed —1.0. But many people have serious doubts about the use of this type of iron snelting furnace in spite of its latter use in the form of the Catalan hearth. This latter furnace was believed to be a developed bowl hearth and, in the form shown by Percy (1864), its height/ width ratio exceeds one and it is worked ln a special way with the ore going down one side and the charcoal the other.

A recent description of the operation of a bowl furnace in Burundi, East Africa, by J-P Chretien (1982) renlnded us of sinilar descriptions of Scandinavian hearths (Evenstad 1968, Busch 1972). The Burundi type furnace operated for about 2000 years up until the 1930s in the south of the country, to the northeast of Lakę Tanganyika. In the 1970s J-P Chretien managed to get together enougb people who knew how to work the process.

The rich ore contained 70% ferric oxide and was roasted. The furnace was about 0.5-1.0 m diameter with two opposed tuyeres supplied by double bellows. The two tuyeres were directed downwards towards the botton where the charcoal was consumed, causing an upward current of carbon nono-xide to reduce the ore above. When the reduced ore gets too Iow and is in danger of oxidation, it is pushed away by a stick to one side and replaced by further supplies of charcoal and ore. Thus it is 'manipulated' and this movement leaves no tracę in the archaeological record.

The Scandinavian process reported by Busch (1972) at Nornas ln Sweden seems to adopt much the same principle (Fig.6). The 'bowl* is 50 cm square and 40 cm deep with a single horizontal tuyere; thus it has a height/width ratio ~ 1.0 and can be sald to be a true bowl by our defini-tion although it has a sloping funnel-like top.

As in the furnace described by Evenstad (1968), the charcoal is madę in the furnace so that no heat is lost. After filling in this way and adding additional charcoal, the carefully blended ore was charged over the whole surface of the bowl, first at the edges and then ln the centre. An air-vent hole was kept open in the middle and charcoal was added as it burnt away near the tuyere. Fuel ash was also added (as flux7).

Then the ore from near the walls was pushed towards the blast while the bowl was topped up with charcoal. At this point slag could be seen to be flowing, presumably away from the tuyere, and the finished bloom was found in the centre of the bowl weighing about 7 kg.

Clearly this is a highly skilled process and manipulatlon of the con-tents is the key to success. Evenstad describes a sinilar process ln use in Norway in 1782. The furnace was a truncated cone (Evenstad 1968) about 50 cm diameter at the bottom and 1 m diameter at the top; its height was about 80 cm. Its mean height/width ratio is therefore 1.15 and there is no doubt that it is a true bowl furnace and, although it is circular, has much in common with the furnace at Nornas. There is a single horizontal tuyere. Just below the tuyere there is a 'piąte' of stone. When