Title: Long-term developments in the technology of iron craft production: a case study in the territory of Sagalassos
Other Titles: Ontwikkelingen op lange termijn omtrent de technologie van Antieke ambachtelijke ijzerproductie: een casestudy in het Sagalassos territorium
Authors: Eekelers, Kim
Issue Date: 24-Oct-2016
Abstract: Abstract

Today iron is one of the most important metals in our daily life. The first iron objects resulting from smelting iron ores was dated to the 3rd millennium BC. Iron smelting was developed in Anatolia (Turkey) and spread fast across the Mediterranean. Initially iron was considered a valuable metal, but by the 13th century BC, it had evolved to a metal for daily use, in tools and weaponry.

Iron slag, the metallurgical waste product of smelting and smithing activities, are often found on sites dating already from the Neolithic. Studying slag over metal object has the advantage that the record of slag is more complete (slag are rusty lumps not worth plundering), destructive analysis can be performed (since no one really cares about slag, except a select group of archaeometallurgist) and corrosion has little effect on the texture of the slag. Slag are one of the most underestimated find on archaeological sites, often classify in boxes as production waste material and discarded for decades.

Slag found during almost three decades of excavations in Sagalassos, a Hellenistic to Early-Byzanine site in the SW of Turkey, underwent the same faith. The iron slag found in Sagalassos, were first described by Kucha et al. (1995). These authors attempted a first technological reconstruction of the iron production. Afterwards, slag were used for several smaller, mostly provenance studies. This thesis fills the gaps in the first studies and has two main aims, employing the framework of “chaîne opératoire”. Firstly, the local iron technology is reconstructed using chemical, mineralogical and statistical techniques. This includes among others the determination of the used raw materials as well as deducing the thermochemical history of the slag. Secondly, the “chaîne opératoire” offers a framework to put this technology in a broader, archaeological context. This includes determining the potential exchange of knowledge and goods based on provenance and spatial analysis, as well as determining the scale of the production, potential trade and the sustainability employing mass balance models.

Iron slag are a complex amalgam of iron silicates, iron oxides and amorphous phases. This makes standard chemical and mineralogical analysis need adaptation specific to slag material.. The chemical composition of the iron slag is determined by ICP-OES, applying an adapted LiBO2-fusion. Due to a change in legislation, sampling was unfortunately no longer allowed in Sagalassos during the course of this thesis. A new methodology, using HH-pXRF was developed to characterize iron slag in the field. For this study, a calibration was developed in cooperation with dr. R. Scott (KU Leuven), to obtain semi-quantitative results. Mineralogical characterization of the slag is acquired by optical microscopy and XRD-analysis. The minerals are quantified using the Rietveld method. Also here, adjustments specific for iron slag needed to be made. SEM was further used to refine the Rietveld implementation code for magnetite and ulvöspinel quantification in XRD. Interpretation of the chemical and mineralogical data is aided by applying statistical analysis, such as box- and kernel density plots, PCA, correlation tests, regression analysis and ANOVA. To obtain insight in the used raw materials and yield of the iron production process, mass balance calculations are necessary. Different models are tested and compared. It could be concluded that none of the model are applicable for the Sagalassos iron slag. The models require a thorough knowledge of the used raw materials, which is not the case for the Sagalassos sites. To estimate the scale of the production and ecological impact, this thesis needed to rely on ratio’s obtained from experimental archaeometallurgy.

Three, potentially four smelting sites are identified in the Sagalassos territory: Bereket (dating inconclusive), Tekkeli Tepe (6th-7th century AD), Dereköy (6th-7th century AD) and potentially Düzen Tepe (5th-2nd century BC). There are remarkable differences between the different sites in the use of raw material and the applied technology. At Bereket, a site 25 km SW of the administrative centre, a limonite ore in a chert matrix was exploitated for iron smelting. A clear ore dressing site, with chopped off chert, could be recognized. A slag heap of around 110 m² is found next to the ore dressing site. The main type of slag found at this site are tap slag, indicating the use of tap slagging furnace. The slag (and ore) are characterized by elevated Cr-values. CaO is clearly used as a flux. The iron smelting process must have been very efficient, since no free iron oxides are present and the main mineralogical constituent is hedenbergite and tephroite.

At Tekkeli Tepe, iron slag are spread across a hill site. Magnetic images reveal the presence of two, maybe more, furnaces. A potential magnetite-spinel vein was exploitated at this site. The slag here are furnaced cooled slag, implying that the slag solidify at the bottom of the furnace. The slag are rich in Ti, V and Zr. CaO was clearly used as flux. The furnace efficiency is remarkable lower than that of Bereket. Euhedral magnetite, interpreted as unreacted magnetite ore, is the main constituent in the iron slag. Also other free iron oxides, such as wüstite, are present in the slag. At Dereköy, a similar slag type is recognized. Here, a magnetite-spinel placer deposit is the most likely candididate to be exploitated as iron ore. The slag are also rich in Ti, V and Zr and also here CaO was used as a flux. Also in the case of Dereköy the furnace efficiency is lower than that of Bereket, though a slight improvement compared to Tekkeli Tepe could be noticed. Magnetite is still present as remnant of the ore, but it has clearly reacted since hopper-shapes are the dominant texture in the slag.

The reason for the efficiency difference between Bereket on the one hand and Tekkeli Tepe and Dereköy on the other, is related to the used ore. Ti will elevate the smelting temperature, making it more difficult to tap. Moreover, magnetite is more difficult to reduce than limonite.

There are also some resemblances between the three sites. They are all three build on a hill site, probably exploiting the wind for their furnaces. The sites are also build in the proximities of their ores (least effort principle), away from the settlements. Finally, the applied furnace temperature falls in the same range of 900-1250°C.

It is unclear whether Düzen Tepe is a smelting or smithing site. Primary and secondary smiting slag are identified on the site. The presence of a valuable iron ore source, namely a magnetite placer deposit, and the possible identification of an ore dressing site, suggest however that also smelting activities occurred at Düzen Tepe. This implies, in comparison to Bereket, Tekkeli Tepe and Dereköy, that both smelting and smithing occurred within the city walls. The thermomechanical characteristics of the smithing slag indicate the use of a single, mechanically handled forging technique. This means that the smithed object is heated at constant temperature, after which it is hammered (mechanically deformed). This process is repeated over and over again and will result in a simple utility object.

At the administrative centre of Sagalassos, the main activity is smithing. Three potential workshop dating Early Roman to Early Byzantine, are recognized, though most iron slag are found in secondary contexts (dump fill, road leveler). Based on the textural analysis of the slag, two main smithing techniques can be recognized. The dominant technique is the mechanically handled technique similar to Düzen Tepe. The second technique that emerges, is a temperature handled technique. Here different temperature regimes are recognized in the slag, indicating the forging of complex objects. Moreover, CaO is used as a protective agent, which can be indicative of the forging of cutting-edge objects.

After the technological reconstruction, the data are further used to unravel the “chaîne opératore” of the site. The archaeological implications, mainly economical and politically, are further described. First of all, there does not seem to be an exchange of knowledge between the smelting sites. This can be related to a potential time lag or the distance between sites, but is most likely related to the different use of ores. Based on the provenance analysis, there does not seem to be an exchange of goods. The scale of the iron smelting sites is estimated based on literature of experimental archaeology. Though the sites of Bereket and Tekkeli Tepe are, in comparison to Düzen Tepe, considered industrial scaled iron production sites, compared to other Roman sites worldwide, the production is small. This implies that the Sagalassos sites are mainly self-sufficient, answering to local the local market and no export occurred. This however also implies that the ecological impact of the iron production was minor, surely compared to other wood consuming activities in the territory. The main demand of iron comes from civil demand, opposed to military demand. There does not seem to be a link between the type of smithing techniques and periods of conflict or supply to the Roman army.

In general, this thesis showed that the iron production in the Sagalassos territory is small, mainly focusing on local civil (urban and rural) demand. There was no or limited trade. The impact on the local environment would have been minor.

This thesis whipped off the dust that covered the iron slag of Sagalassos after being stored as production waste for almost 20 years and reconstructs the technology and associated “chaîne opératoire” of the iron production in the territory. It offered the opportunity to estimate the importance of Sagalassos as an iron production centre. Moreover, this thesis offers future archaeometallurgist a workflow applicable to smithing and smelting slag. This thesis contributes to the field of archaeometallurgy in general by offering a methodology and showing the importance of studying smithing slag, which are often discarded in literature. By studying SHB, the researcher obtains insight in the applied smithing techniques and thus the forged objects. Finally, this thesis also stresses the importance of mineralogical analysis, more important the use of simple optical microscopy to study the thermomechanical history of smelting and smithing slag.
Publication status: accepted
KU Leuven publication type: TH
Appears in Collections:Division of Geology
Administrative and Support Services, Faculty of Science

Files in This Item:
File Status SizeFormat
Thesis_Kim_Eekelers_11092016.pdf Accepted 19714KbAdobe PDFView/Open Request a copy

These files are only available to some KU Leuven Association staff members


All items in Lirias are protected by copyright, with all rights reserved.