Jernudvinding i Danmark i forhistorisk tid

Forfattere

  • Olfert Voss

DOI:

https://doi.org/10.7146/kuml.v12i12.103924

Nøgleord:

Iron smelting, jernudvinding, forhistorisk, prehistoric, slag pits, slagge grube, Drengsted, Snorup, Scharmbeck, yield, udbytte, dating, datering

Resumé

Prehistoric Iron Smelting in Denmark

Of the various arrangements for iron smelting which have been employed in Denmark up to c. 1600, only those with slag pits will be discussed here. Usually, only the slag from these is preserved, most often fused into one large lump, which reproduces a portion of the pit shape, fig. 2.

Nearly all earlier attempts at reconstruction are based on the assumption that the iron particles which are formed during extraction sink down through the fluid slag mass and collect at its base 1), but as examination of such accumulations "in situ" has never yielded traces of the extracted iron (bloom), this supposition is hardly acceptable. Both from the appearance of the slag block and from technical probing and experiment, as well as from ethnographical and historical sources, it can be assumed that a superstructure of clay has been erected over the slag pit. In this shaft the actual extraction has taken place, the iron being retained whilst the slag has run into a pit in the ground, where it has often solidified into one large Jump, the slag block. The construction of the superstructure is reproduced in a furnace shaft which was found at Scharmbeck near Hamburg in the 1950s 2). The necessary air supply was maintained through four holes and the chimney effect of the shaft presumably ensured a sufficiently powerful draught. The slag pit was probably quite empty apart from a plug of straw which closed the hole under the superstructure. This plug perhaps continued to the bottom of the pit as a cylinder, or was merely held in place by thin twigs. The slag first collected at the bottom of the superstructure and after a certain length of time the straw plug was perhaps sufficiently burned to allow the slag to run out, or was pushed down into the pit by means of a stick through one of the air holes.

The metallic iron formed in the superstructure had a very low carbon content and consequently aggregated into a spongy mass which easily stuck to the sides of the furnace or perhaps fell to the bottom. When the slag had run away, the bloom was broken out of the furnace, which was subsequently repaired with clay and placed over an empty slag pit and a new smelting commenced.

The new reconstruction of an iron smelting furnace with a slag pit, shown in fig. 1, is primarily based on observations made in 1961 during the investigation of 5 slag pits at Drengsted 4) in the south-western part of Sønderjylland. They were all confined to an area of less than 1 acre on a settlement site from the 5th century A. D. A radio-carbon dating of one of the pits to 210 ± 100 A. D. 5) indicates, however, that the smelting in all probability ante­dates the settlement. Only two of these pits, BV and EL, were entirely undisturbed; in the remainder the slag was entirely or partially taken up and it was not possible to ascertain whether this had happened in prehistoric times. The subsoil all over the site was glacifluvial sand, and there was no special lining of the pits.

Slag pit BV (figs. 4-6) was first exposed from above, but in order to obtain the clearest information on the structure of the slag block, the ultimate examination was made from the side, as much of the loose soil in and near the pit being removed as possible without dislodging any of the slag, which was finally sectioned through the middle, roughly east-west, fig. 6. The depth of the pit under the present surface was 90 cm and the greatest width 105 cm. It can clearly be seen on the cleaned slag block, figs. 4-6, that the slag has run into the pit over a limited area, corresponding to a ring 30 cm in diameter roughly in the middle where the slag is highest. It is also clear that the slag has not filled the pit in one go, but that the process has been prolonged, sand slipping into the pit so that it has become filled with alternating layers of slag and sand. For the slag in the pit to have the structure and shape it has, there must have been a superstructure where the actual reduction of the iron oxide and melting of the slag has taken place. The temperature in the pit has at no stage been so high that the slag could melt there; this could be seen, for instance, by the fact that the sand on the inside of the pit was nowhere burned red. On top of the slag lay a quantity of red-burned clay with a slag-covered inner surface. Most of it had a loose structure but a few pieces up to 15 X 20 cm were found, probably deriving from the superstructure.

In slag pit EL, which was c. 80 cm deep, only a median section was made, and as the slag was less coherent there, the photograph (fig. 7) does not so clearly demonstrate that the structure is identical with that of the block from pit BV; the amount of intruded sand has merely been somewhat less, so that the slag block resembles more closely a normal slag block (fig. 2). A different structure was observed down the middle of the central portion of the slag mass, where solidification had apparently not taken place in a rapid descent. On removing the slag a distinctive structure was observed in the middle (fig. 8), which must be the impression of straw, not loosely packed but compressed into a cylinder with a diameter of 20 cm or more. Centrally, at the bottom of the pit, was a thick layer of carbonised straw, just under 20 cm in diameter and 2-4 cm thick, which is no doubt connected with the straw impression on the underside of the slag. Both these phenomena are known from other slag pits and can be explained as evidence of a straw plug in the connecting pipe from the superstructure.

In the straw from pit EL a score of seeds were found, identified by Hans Helbæk 6) as predominantly barley with a few seeds of wild oats. The radio-carbon dating of 210±100 A. D. was made on straw from this pit, and there is thus an established agreement in time between the use of the furnace and the basis of the radio-carbon dating, namely the harvesting of the corn.

Most of the slag in pit EM had an unusual structure, the surface being lumpy and uneven (fig. 9), in contrast to the slag in the pits mentioned before, which was dense with a smooth surface. The cause of difference cannot be determined without practical experiment, but it is conceivable that the extraction in this case has completely failed. The weight of slag in this pit is also much less. In the middle of the pit bottom a thick layer of carbon was found and at the bottom of that, a layer of compressed straw, 20-25 cm in diameter, corresponding to the straw in pit EL.

In the last two pits, EO and EP, the slag must have been taken up, and this may have happened either recently, because ploughing was hampered, or in antiquity, because large lumps of iron had fallen into the slag and had to be hacked free. In the lowest, undisturbed part of EO lay a 10-15 cm layer of thick pieces of charcoal (fig. 10), and in the middle, at the bottom of the pit, a lump of carbonised straw. In the disturbed part of the pit, apart from pieces of slag, lay a number of cakes of burned clay, with a slag-covered inner surface. They were up to 4 cm thick, and from the position of the slag drops on the inside, it was obvious that the pieces must have stood vertically. In one of them, the underside of which was apparently undisturbed, fig. 11, there were traces of a circular hole c. 5 cm in diameter, which was partially filled by slag. This was presumably the lowest portion of the shaft, and the curvature suggests a diameter of roughly 50 cm.

The slag blocks of Jutland have a characteristic shape, fig. 2, which is connected with that of the pit. One slag block (fig. 12) of 171 kg, from Snorup, Tistrup parish, Ribe county, gives us an almost perfect cast of such a pit, as the slag has been hot enough to reach the bottom of the pit. Only the middle of the underside is missing, and here straw impressions can be seen, which can be related to the temporary bung. The slag pit in the reconstruction, fig. 1, has the shape of the slag block from Snorup. The majority of slag blocks from Jutland have a slightly different shape, but this is because the slag has rarely reached the bottom of the pit, before solidifying. The lower portion is thus lacking, the slag reaching correspondingly higher up in the cylindrical portion of the pit.

None of the Drengsted pits contained a complete block as in fig. 2, but it is nevertheless reasonable to consider them as special variants of this type. The differences can be explained by the slag in the Drengsted pits not having had a sufficiently high temperature to remain fluid in the pits for even a short period, and in the case of pit BV there has moreover been an intrusion of sand as the slag has run in.

Of the superstructure -the shaft- such characteristic parts are preserved, fig. 11, that one may assume that it has been of the same type as the shafts found at Scharmbeck near Hamburg, fig 13, which can be assigned to the 2nd century A. D. 2). From the slag-covered interior of these it is seen that the temperature up to 30 cm above the tuyeres has been higher than the melting point of slag.

An estimation of consumption and yield in iron extraction has been based on Fr. Hupfeld's description of iron extraction in Togo in West Africa 13), J. W. Gilles' 1957 experiments 11) and Wynne and Tylecote's experiments 12) of the same year, and on this information the yield in furnaces with slag pits is fixed at 1/4 to 1/3 of the iron content of the ore. With 70% Fe2O3 in the ore and 60% FeO in the slag the yield of iron will be 15.5-23 kg per 100 kg ore, with an ore consumption of 127-143 kg. Charcoal consumption is put at at least 10 kg per kg iron.

An archaeological dating of iron smelting sites is as a rule impossible, dateable artefacts lacking completely. True, potsherds have been found in connection with a few sites, but it is not established that these derive from pottery used in the smelting and they must therefore be regarded as secondary. The only dating method which can be employed here is the radio­carbon method, as sufficient carbonised straw or charcoal for such a procedure is present in most pits. At present, only one slag pit, EL from Drengsted, has been dated in this way, and the result was 210± 100 A. D. 5). This did not correspond with the age expected, as it was supposed that the slag pits belonged to the 5th century settlement investigated at the same place. This is a clear indication that slag pits cannot be dated by the pottery found on the site, consequently throwing doubt on many of the dates arrived at in the early literature. Before a comprehensive series of radio-carbon datings of different iron smelting sites has been undertaken, it is impossible to establish when these furnaces were first used in this country, and how long the technique survived.

Slag blocks of the type described here are found in a large quantity in West and Central Jutland 14), but are at present not recorded from the rest of Denmark or from Scandinavia. This type can, however, be traced down into Central Europe along the rivers Elbe, Oder and Weichsel, in Germany, Poland and Czechoslovakia (fig. 20). The distribution area covers a broad belt outside the frontiers of the Roman empire and partially corresponds with the distribution of Roman imports, but the presence of iron-smelting furnaces can hardly be a direct result of Roman influence, as the Roman furnaces were of a quite different nature, without slag pits 15). As long as datings are so few and uncertain, it is difficult to have any well-founded views on the origin of the furnace with slag pit, but it is most probable that the technique has spread from the south-east towards the north-west.

It is most likely that production has covered home consumption, but it is also possible, as suggested by Wielowiejski 16), that iron was to a large extent exported to the Roman empire. For the Danish finds to support such a hypothesis, a greater density of Roman import articles ought to be expected in the iron area of Jutland than in the rest of Denmark, where an equivalent iron production has yet to be discovered. If anything, it looks as if the distribution of slag blocks emphasises the connections demonstrated earlier between the Oder-Weichsel region and Jutland, in the centuries around the birth of Christ 17), but before these interesting cultural-historical problems can be illuminated further, it will be necessary to undertake complete investigations into a number of iron-smelting sites in Jutland and to carry out so many radio-carbon datings that the technique can be accurately placed in time. Further excavation of iron-smelting furnaces will, it is hoped, also contribute to a verification of the reconstruction offered here, the efficiency of which it is intended to prove by practical experiment.

Olfert Voss

Downloads

Publiceret

1962-02-05

Citation/Eksport

Voss, O. (1962). Jernudvinding i Danmark i forhistorisk tid. Kuml, 12(12), 7–32. https://doi.org/10.7146/kuml.v12i12.103924

Nummer

Sektion

Artikler