Forsøg på rekonstruktion af fortidige smedeprocesser

Forfattere

  • Robert Thomsen

DOI:

https://doi.org/10.7146/kuml.v14i14.104249

Nøgleord:

prehistoric forging proces, forhistorisk smedeproces, rekonstruktion, reconstruction, experimental archaeology, eksperimental arkæologi

Resumé

Attempt at reconstruction of prehistoric forging processes

Chemical analysis and microscopic examination of prehistoric iron implements can give valuable information about the processes used in their production. But more precise information can be gained by attempting to reproduce the processes suggested by analysis.

Prehistoric iron contains considerable quantities of slag and of phosphorus, and these impurities cause difficulties in the use of processes employed in more modem times. It is therefore necessary to use in the experiments iron of "Iron-Age" quality. For the following experiments iron from a board of 20 iron bars found at Hedeby in 1964 was used as a basis of comparison, while the raw material for the reconstruction experiments was small pieces of iron from the iron extraction experiments carried out by Varde Steelworks (5).

Longitudinal and cross sections were sawn through the three bars from Hedeby, and the resultant surfaces polished and then etched with nitric acid to reveal the structure of the iron. From this it could be seen that all the bars consisted of many small pieces of iron welded together, some weighing less than 10 grams.

In addition to the normal microstructure of ferrite and perlite the iron contained large quantities of non-metallic inclusions the characteristic form of this slag being visible in Figs. 7 and 8.

The welding seams show a considerable charcoal content. For example at point b-b there is on either side of the seam a perlite area with high carbon content. At several points (Fig. 6) welding has been unsuccessful, and cracks can be seen full of a black substance which must be the charcoal which caused the carburization, while outside this carbonrich zone, at some places 1 mm. wide, the structure is purely ferritic. The section in Fig. 2 shows that the areas rich in carbon only occur at the surface.

A chemical analysis was made of the Hedeby iron, and tests of phosphorus content carried out at intervals of 50 mms. along the edge of the longest bar (cf. tables in Danish text). The considerable variations in phosphorus and carbon content are due to the lack of uniformity in structure, while the very low sulphur content is caused by the iron being produced with the help of charcoal, which contains much less sulphur than does hard coal. The phosphorus content is more significant; though high compared with modem iron, it only amounts to a tenth of that obtained in iron produced from bog ore. O. Arrhenius (6) has shown that such iron can only be produced from mined ore, and, as 32 % of the over 300 Swedish iron objects from all periods of the Iron Age analysed by Arrhenius contained less than 0.1 % phosphorus, it is possible that the Hedeby iron was imported from Sweden.

To sum up: the iron bars found at Hedeby are about a half meter long and weigh about 600 grams. They consist of a soft carbon-free central portion and harder carburised ends, made up of many small pieces welded together. They date to the early Middle Ages or the Viking Period, and cannot have been produced from the local bog ore.

Laboratory experiments

1. Carburized layers up to 1 mm. in thickness occur in the Hedeby iron. Surface carburization is widely used nowadays to give a hard surface to a soft tough core, and the object of the experiment was to determine the period of carburization required to give a layer of this thickness. An electrically heated laboratory furnace was used, with a temperature of 1150° C, the minimum temperature for forge-welding. Three pieces of iron were placed in crucibles with charcoal powder, and the crucibles heated to 1150 ° C over a period of three hours. One piece of iron was removed 15 minutes after this temperature was reached, the next after 75 minutes and the third after 105 minutes. The depth of carburization was 0.6, 1.4 and 2 mms. respectively. As a later experiment shows, the welding temperature must have been considerably higher, and the prehistoric iron can therefore only have been heated to this temperature for a very few minutes before achieving a carburization layer 1 mm. thick. The possibility must, however, be taken into account that the carburized zone may originally have been thicker, and have been wrought down to the thickness found.

2. One of the pieces from the former experiment was wrought down to a tenth of its original thickness. It appeared thereby that the hard carboniferous perlite layer was not appreciably deformed by the process, but instead split off, leaving the soft tough ferrite exposed.

3. Perlite areas are found on the surface of the iron bars, and can be interpreted as particles of steel with high carbon content welded deliberately to the surface. The possibility was also present, however, that these perlite spots occurred naturally at points where the oxide scale was thin or missing, and carburization occurred more rapidly. The oxide scale was therefore removed in three stripes on each of the three iron pieces before the carburation experiment. The resultant depth of carburisation was not greater at these points, nor could any structures be found resembling the perlite spots.

4. In order to estimate the welding temperature for carburized iron, two pieces of already wrought iron were surface cleaned and placed in a crucible with a thin layer of charcoal between. After healing to 1150 ° C the two pieces adhered together, but the temperature was too low for the pieces to be welded together.

5. Six small pieces of iron, containing slag and not previously wrought, were heated to 1150 ° C surrounded by charcoal powder. After an hour at this temperature so much slag had run off that it was clear that the iron must have held more than 50 % slag, and the pieces were cintered into a single lump. This was heated to 1250° C in a forge hearth, kneaded and worked with smith's tongs to remove more slag, and finally wrought to a thin bar of almost slag-free iron.

6. The greater part of the iron from the extraction experiments is in the form of sponge­iron, and it could be expected that such sponge-iron, heated with carbon and thereafter welded together, would form perlite structures, as the wall thickness of the individual cells would be less than the carburization depth of 1 mm. A piece of sponge-iron, already containing charcoal particles from the extraction process, was accordingly heated to 1150° C, but was later wrought after healing in a forge hearth to 1250 ° C. The structure was not the expected perlite, but ferrite with same "islands" of perlite corresponding exactly to the structure in Fig. 12.

7. While pieces of iron weighing over 50 grams are easy to forge together to larger units, providing the temperature is high enough, smaller pieces are more difficult, because: (a) they are lost among the forge coals, (b) they bum up because of their spongy surface, (c) they lose their heat before reaching the anvil, (d) it is difficult to remove the oxide scale, and (e) the presence of slag prevents metallic contact between the pieces. The loss of heat can, however, be avoided by welding a smaller piece to a larger, using the heat capacity of the larger piece as in the Japanese Shinto-smithy. Accordingly a hollow was wrought in the larger piece, in which the smaller could lie. The welding was successful, and the result resembled the two areas in the upper left-hand corner of Figs. 2 and 3. A similar process has been employed with bar 3 (Fig. 13 ), where a little wedge of iron has been wrought into a fork formed by two other pieces. A similar technique is illustrated in hand­books for smiths from the present century.

8. The perlite strips at c-c in Fig. 2 may belong to a carburized welding seam, but may also belong to a carburized surface zone. If so the ferrite layer closest to the surface must have arisen from a later decarburization. A considerable thickness of decarburization is obtained by healing steel in an oxidizing atmosphere. Accordingly the piece of iron which had been carburized for 75 minutes in the laboratory furnace was heated three times to 1250° C in the forge hearth. After each heating it was hammered with light strokes, as in forge welding, and thereby wrought down to half thickness. Laboratory tests showed no decarburization, perhaps because slag, continuously hammered out of the iron, protected the surface against oxidization. The piece was wrought three more times, to a tenth of its original thickness, but the microscope still showed no decarburized portions.

9.The zones with high carbon content along a welding line must have arisen by two pieces of iron being welded together with a layer of charcoal or other organic matter between. The addition of carbon may have been motivated by: (a) the fact that the welding temperature is lower for carbon-rich steel than for carbon-free iron, (b) the fact that carbon hinders the formation of oxide scale which prevents metallic contact between the pieces and forms one of the greatest difficulties in welding, or (c) a desire to convert the iron to steel. (That this carburization process was known even in the Early Iron Age is shown by the discovery of knives where the cutting edge has been carburized, while the back consists of carbon-free iron (1)). Accordingly an attempt was made to weld two pieces of iron together with a charcoal layer between. The experiment appears to eliminate (a), as, even when the charcoal did not fall out from between the pieces, it proved to be considerably more difficult to weld carburized than carbon-free iron, as the surfaces slide the one upon the other even at lowered temperatures. It did, however, prove possible to get the two pieces to adhere, and a carburization did take place (Fig. 18), though with a lower carbon content than in the prehistoric welding seams.

10.The difficulties of welding together small pieces of iron might also be reduced by either packet-welding or welding in crucible. Both methods were probably used. Packet­welding involves hammering pieces of iron thin, which tends to remove much of the slag, and then welding them together in a pack, and the fact that the Hedeby iron contains less slag than the iron we produce suggests that at least bar 1 (Fig. 2) was produced by packet­welding. For the experiment in packet-welding iron from the crucible-welding experiment (below) was used. The pieces were wrought down from 8 mms. to 1-2 mms. thickness, the oxide scale removed from the surfaces, and the sheets packed as shown in Fig. 19. Charcoal powder was introduced between the sheets in order to try once more to produce carburized zones at the welding points. The packet was heated slowly in the forge, to allow the single layers to reach welding temperature simultaneously. The hearth coals had been burnt to coke previously to remove the sulphur, which would have ruined the experiment. After several heatings and hammerings the packet was welded into a compact mass apart from the outhermost sheet which would not adhere on account of formation of oxide scale. However, after a little charcoal powder was scattered under the sheet this too adhered satisfactorily. Figs. 19, 20 and 21 show that by packet-welding the same appearance and microstructure was achieved as is seen in Fig. 2.

While it is possible that bar 1 was produced from a single piece of iron, hammered thin and then folded together several times, the appearance of bar 2 (Fig. 3) shows that it was welded from small solid pieces. As it is practically impossible to weld such small pieces in an open forge it is tempting to imagine them welded together in a crucible. African tribes use this method, but there was no evidence of this method being used in prehistoric Denmark until Olfert Voss, excavating slag-pits at the iron-producing site at Drengsted in the autumn of 1964, found a little lump of slag in the topsoil which can only have come from a crucible-welding process.

The lump of slag (Figs. 22, 23 and 24) was almost circular, with a greatest diameter of about 80 mms., and weighed 290 grams. The whole upper surface was covered with a thick layer of cintered sand. This sand must have been added at a late point in the smelting, after the slag had run off, in order to remove the oxide scale by the well-tried method of converting the iron oxide of the scale, with its high melting point, to iron silicate, which has a low melting point.

Whether a crucible heated from outside was used, or an internally heated "crucible-forge", could not be immediately determined. Our iron-extraction experiments had shown that welding temperatures could easily be reached with an outside-heated crucible, while the amount of charcoal which a crucible 80 mms. in diameter and 150 mms. high could contain also appeared probably to be sufficient to reach the required temperature.

11.The first attempt at welding small pieces of iron in a crucible heated on a forge failed. A thin-walled pottery vessel was used as crucible, and melted after less than an hour.

The thermoelectric pyrometer among the iron pieces showed at that point 1020° C, but the temperature of the crucible walls must have been considerably higher. A chamotte crucible with walls 10 mms. thick was then employed. The lower quarter was filled with charcoal, which would give place to the slag as it melted down. Above this were pieces of iron with slag content, and above this again a layer of charcoal. The crucible had a lid, and a thermo-element of type Pt/Pt-Rh was set in the middle of the iron. The crucible was heated slowly on the forge without forced draught until the meter showed 400 ° C, when forced draught was applied and in 40 minutes the temperature reached 1280° C. More iron and charcoal was added, as the contents had sunk. After a further 45 minutes the temperature had reached 1320° C and the experiment was terminated. The 85 pieces of iron, originally weighing 755 grams, had coalesced to a large lump of sponge-iron weighing 260 grams. The slag weighed 490 grams. Judged from these figures and the weight of the Drengsted slag, the crucible used at Drengsted was about 80 mms. in diameter and about 80 mms. high.

Accordingly a crucible was made to these dimensions and with walls 30 mms. thick (Fig. 27), and the experiment repeated. The resultant sponge-iron was squeezed with tongs at 1300° C into a vertical square-section bar and was wrought further in the forge. Here it broke into several pieces, partly through inexperience. It is undoubtedly possible for an experienced smith to weld small pieces of iron with slag-content in an exterior-heated crucible.

12. The final experiment was to test welding in a crucible-forge, but as no more iron of "Iron-Age quality" was available it was only possible to test whether temperatures sufficient to melt slag could be reached. Such a forge has an entrance for draught, and as the Drengsted slag shows no sign of this it is probable that it is from an exterior-heated crucible. However, while the experiment was in process Olfert Voss found a lump of slag in a smithy at Trans church, dating to the 12th century and about the same size as the Drengsted lump, but with marks from the draught and of the sloping shape which results from the heat being greatest at the point of draught entrance.

For the experiment a chamotte crucible was made 70 mms. in interior diameter and 150 mms. high, and an 8 mm. pipe was introduced 30 mms. above the crucible bottom tangential to the interior surface. A bellows was attached and the crucible filled with charcoal. By use of the bellows the temperature was raised in the course of an hour to 1100° C. 400 grams of slag pieces were placed at the top of the crucible and when the charcoal was burnt out there were 60 grams of melted slag in the bottom, a cintered lump of slag above the draught-hole, and loose pieces of slag at the top. The partial failure was due to insufficient draught. The experiment was therefore repeated with a larger crucible and two bellows. This time the temperature reached 1280° C and there were no loose pieces of slag left in the forge. The experiment thus showed that crucible-welding could also be carried out in a crucible-forge, and they will be continued when more extracted bog-iron is available.

One thing which the series of experiments makes very obvious is the great care and patience expended on the production of these iron bars. It required undoubtedly a full day's work to weld 100 grams of pieces of iron into one piece. This fact serves to give a picture of the value of iron in the Viking Period.

To one in the trade it is particularly impressive to see that much of the technique of the smith used today has been known for a thousand years. It is as though the Viking Period is not so many years distant.

Robert Thomsen

Downloads

Publiceret

1964-02-13

Citation/Eksport

Thomsen, R. (1964). Forsøg på rekonstruktion af fortidige smedeprocesser. Kuml, 14(14), 62–85. https://doi.org/10.7146/kuml.v14i14.104249

Nummer

Sektion

Artikler