Ancient metalworking at Yodhawewa site, Sri Lanka: tracing the archaeological and geochemical relationship by analyzing soil and slags

The Yodhawewa archaeometallurgical site was the latest discovery in Sri Lanka in 2018. This study aimed to trace the archaeological and geochemical relationship by analyzing the soil and slags of the site dating back to the c. first to eighth centuries AD. The analysis was mainly based on the X-ray fluorescence (XRF) method. The soil chemistry resulted in some clues to special metal processing zones in the excavation-1 area. The cultural layers of that area (profiles 1 and 2) showed relatively high copper composition (max. 470 ppm), and crucible fragments containing copper particles were also detected in the same layers. The vertical distribution of soil elements and their correlations with TiO2 suggested that long-term metal activity may have affected changes in environmental soil chemistry. Relatively high phosphorus was indicated in the soil-related furnace wall, suggesting the artificial addition of organic matter to the clay body during the furnace wall construction to withstand the high temperatures. High iron slag (HIS) varies from 12.96 to 49.63 wt% of Fe2O3 and shows high MnO, P2O5, and V. Lightweight amorphous/glassy slags (LIS) associated with secondary refining have low iron content and a high CaO and Sr composition.


Introduction
Geochemical analysis has become increasingly recognizable in parallel with traditional archeological methods that provide interpretations of past cultural contexts (Bintliff & Degryse 2022;Chang et al. 2022;Columbu et al. 2019;Dias et al. 2019;Haslam & Tibbett 2004;Llull Estarellas et al. 2021;Pastor et al. 2016;Sánchez de la Torre et al. 2020;Trant et al. 2021;Walkington 2010;Warchulski et al. 2018). A record of human activities and management in a landscape provides the soil and a comprehensive picture of past human space used (Cook et al. 2006;Middleton 2004). Also, geochemical studies of the soil significantly improve the archeological interpretation of ancient metalworking habitations (Aston et al. 1998;Wilson et al. 2008;Oonk et al. 2009). Iron production by smelting ore was a revolutionary invention in pre-modern times. The ancient metalworkers developed the technology through long-term experience to separate "raw iron" from ores by the bloomery furnace at temperatures around 1200 °C or a blast furnace (indirect/cast iron followed by the decarburization process) temperature around 1400 °C (Liu et al. 2019;Pleiner 2000). However, the bloomery technique is a direct process and is more ancient than cast iron production (Li et al. 2022;Selskienė 2007). The main by-product of this smelting process is "slags" rich in iron oxide, formed by the reaction between iron oxide and silica; other contaminants in ore often include slag in the process (Blakelock et al. 2009;Dill et al. 2013;Tylecote 2002;Warchulski et al. 2018).

3
Variations in the slag morphology can be presented in iron smelting depending on the raw material, furnace structure, and other operating parameters (Charlton & Humphris 2019;Humphris et al. 2009). As a result, metal ore processing usually leads to multi-elemental soil contamination (Dudka & Adriano 1997;Ströbele et al. 2010). Reduction phase slag (tap slag, furnace-bottom slag, and adhering and entrapped bloom slags), and purification phase (reheating slag), can be used to make a fair judgment about past metal activities in a site (Blakelock et al. 2009;Coustures et al. 2003).
Anuradhapura citadel (c. 834-778 BC), Aligala near the Sigiriya, and Pomparippu megalithic burial site (c. 998-848 BC) revealed the earliest metalworking evidence in Sri Lanka (Begley et al. 1981;Deraniyagala 1992;Karunaratne & Adikari 1994). When considering the recent observations (after the 1980s), Seneviratne (1985Seneviratne ( , 1995's, Thantilage (2008)'s, and Anusha Kasthuri (2016)'s research led to significant advances in archaeometallurgical studies in Sri Lanka. Juleff (1990Juleff ( , 1996 and Solangaraarachchi (2011) provided detailed study records of iron and steel production in ancient Sri Lanka. Juleff (1996) describes the iron extraction of the Samanalawewa area through the west-facing furnaces activated by monsoon winds from the c. fifth century BC to the twelfth century AD. The research on iron extraction sites in the Sigiriya area (based on the Kiri-Oya basin archeological research 2004-2006 dated from the c. third century BC to the tenth century AD) performed XRD and electron microscopic analyses on selected iron ore and slag samples (Solangaraarachchi 2011). Furthermore, EPMA and EDAX bulk analysis has been accomplished to study metal artifacts unearthed during the 1980-1984 excavations in the Mannar ancient port-city area (Juleff 2013). This study aimed to trace the archaeological and geochemical relationship by analyzing the soil and slags unearthed from the Yodhawewa site. In achieving that aim, the objective was to interpret the acquired results by conducting a geochemical analysis of 46 pieces of slag, 33 soil samples, and two (2) furnace wall fragments collected from the premises in 2018.

Study area and the geoenvironmental setting
The island of Sri Lanka has a length of 432 km with a maximum width of 224 km, and the entire land of 65,610 km 2 is located in the Indian sea below Indian Peninsular (Fig. 1a). Yodhawewa site is located near the spill of the Yodhawewa reservoir (Giant's tank), 12 km southeast of the old port city of Mannar, in the Mannar District, Northern Province of Sri Lanka (Fig. 1b). The entire studied area in 2018 was 201,604 m 2 (Fig. 1c).
Sri Lanka has a tropical climatic condition that includes distinct wet and dry regimes (Elliot et al. 2003). The study area belongs to the dry-semi-arid zone and the North-western plains of the country (Pemadasa 1984). The peripheral area is richly endowed with natural and artificial wetlands such as rivers, coastal wetlands, and the canals associated with the irrigation tanks cascade system. Furthermore, subsurface geology comprises a Miocene limestone bed (Cooray and Katupotha 1991) (Fig. 1b). Three different soil types are distributed in the area: Reddish Brown Earth's undulating terrain developed on the underlying Red Yellow Latosol horizon, and the flat coastal terrain comprised of alluvial soils (Alwis & Panabokke 1972). A dry climate is typical, and the temperature ranges between 30 and 35 °C during the field sessions in March-April 2018. Due to the heavy rainfall in the Malwathu-Oya (river) during the northeast monsoon (December to February) of the North-Central part of the country, the Yodhawewa reservoir also overflows flooding the study area. The vegetation patterns signify the arid climate as tropical thorn forests and arid grasslands (Pemadasa 1984). Wind speeds in this region are 6.4-7.5 m/s (Elliot et al. 2003).

Yodhawewa ancient metallurgical settlement
In October 2017, a reconnaissance survey of ancient habitations near the Yodhawewa reservoir yielded the first evidence of the Yodhawewa metalworking settlement. Formal archaeological observations at the site were conducted in March-April 2018. Thirty-three soil samples, two furnace wall fragments (FWF), and 47 slag samples were collected from the Yodhawewa site in 2018 and geo-chemically analyzed.
The diversity of artifacts density was significant in identifying the two cultural layers of the entire premises; morphologically, the soil color presented primarily similar features. A total of 14,017 artifacts were collected from the entire Yodhawewa study, and the majority of them, 57.20% (n = 8,018), were collected from Ex.2. Other 16.25% (n = 2278), 16.15% (n = 2263), and 10.40% (n = 1458) were collected by exploration, from the first excavation pit, and six profiles, respectively. Quantitatively most slags, crucible fragments, furnace wall fragments (FWF), metal artifacts, or metal fragments have been acquired from Ex.2 area (Table 1). Five datings for five charcoal samples were obtained from the Yodhawewa (Ex.2) site, dated from c. first to eighth centuries AD (Fig. 2) (Wijepala et al. 2022b).
The "crucibles" found in the Ex.1 and Ex.2 areas varied morphologically and in use context. Regular-shaped crucibles were used for copper metalworks of the Ex.1 area, while elongated tube-shaped crucibles (found in the Ex.2 area) were used to produce crucible steel (carburized steel/ wootz) (Wijepala et al. 2022a). Factors related to copper metalwork in the Ex.1 area were also found with some copper inclusions in the slags. A significant discovery of Ex.2 was the bottom part of the lower half-spherical shaped (crucible shaped) furnace used for crucible steel production ( Fig. 3d, g). The slag pool (n = 6226) discovered from the  Ex.2 was also the most extensive artifacts collection in the entire research (Table 1). Although clear archaeological evidence related to iron and copper metalworks has been found from the Yodhawewa site, there were no raw materials (iron and copper ore) deposits nearby (Wijepala et al. 2022b).

Archaeological field methods
The fieldwork of the Yodhawewa research was mainly based on an archaeological survey and two excavations (Fig. 1c).
Surface sampling was carried out by exploring the distance of 1600 m on the right bank of the canal in 32 sample units of 50 m each. In order to identify the soil stratification and the artifacts scattering at the site, the soil profiles (P1-P6) were observed in parallel to the field survey ( Fig. 1c, 3 h). The two vertical excavations were (LMBA/YW/2018/Ex1 and LMBA/YW/2018/Ex2) of 9 m 2 each ( Fig. 3a, b).

Soil and slag sampling
The overall soil sampling was based on five prominent locations; first excavation (Ex.1), second excavation (Ex.2), and selected three profiles (P-1, P-2, and P-6). The topographical, geological, land use maps, and aerial photographs were used to determine corresponding locations. Thirty-three soil samples were collected using a "Marshalltown steel pointing hand trowel" to sample the soils after cleaning using alcohol. The average of 50-75 g samples was obtained by digging into the wall using a vertical linear sampling method. Plant residues were removed from the sample, packed in zip lock polythene bags, numbered outside each sample, and maintained the detailed records in a field notebook. Soil samples (n = 2) representing the first and second cultural layers were collected from Ex.1 (Fig. 3a, c). Twenty-one soil samples were collected from three locations representing the maximum soil layer details of the entire Ex.2. Samples 15 to 21 (Ex.2:15 to Ex.2:21) were collected based on the crosssection of the 9th context ( Fig. 3b, d, e), including residual soil accumulation connected to the furnace wall. The other 14 samples were unearthed from two vertical lines near the main furnace structures of contexts 5 and 9 ( Fig. 3b, f, g). The soil sampled as n = 3, n = 3, and n = 4 on P1, P2, and P6, respectively, from the profiles (Fig. 3h). Follow the Munsell Color (Firm) (2009) to designate the specific soil colors. Out of the 6814 slags acquired from the entire research area, 368 samples were discovered by the survey (Table 1). Selected 46 slag samples (displaying the slag sampled locations in Fig. 1c) were used for geochemical analysis. Overall research revealed 251 furnace wall fragments (FWF), and two samples collected from the survey were used for scientific analysis ( Table 2). The samples were sealed in polythene bags and transported to Japan. Geochemical analyses were carried out from January to March 2019 at Shimane University, Japan.

Analytical procedures
The corresponding soil and slag samples were analyzed using X-ray fluorescence spectroscopy (XRF) analyses. All samples were oven-dried at 120 °C for 24 h before being transported to Japan. Then, the samples were oven-dried at 160 °C for 48 h, with approximately 50 g of soil and slag samples before being crushed to a fine powder in an automatic agate mortar and pestle grinder. The powdered (< 63 µm) soil and slag samples were compressed into briquettes using a force of 200 KN for 60 s. The concentrations total of 22 major (TiO 2 , Fe 2 O 3 , MnO, CaO, and P 2 O 5 ) and trace elements (As, Pb, Zn, Cu, Ni, Cr, V, Sr, Y, Nb, Zr, Th, Sc, F, Br, I) were then determined by X-ray fluorescence spectrometry using a Rigaku RIX-2000 spectrometer equipped with an Rh-anode tube at Shimane University. The XRF data for oxygen and carbon does not represent this research because they are light elements that cannot be detected with XRF (Ravansari et al. 2020). Analytical methods, instrumental conditions, and calibrations of Kimura and Yamada (1996) have been followed. Analytical results for GSJ (Geological Survey of Japan) standard JSl-1 were acceptable compared to the proposed values (Potts et al. 1992). Average errors for all elements are ± 10% relative. A summary of the geochemical results related to the soil, slag, and furnace wall fragments is presented in Table 2.

Cultural and natural formation of the study area
The dark brown surface soil layer, granular and crumb structure, and fine and medium particles were the majority in the whole area. The first cultural layer (dark reddishbrown) of Ex.1 comprised diversified artifacts of pottery fragments, crucible fragments, slags, and beads. The layer was slightly thicker than the surface layer and mainly consisted of fine, medium, and coarse particles in the soil. Significant crucible fragments (n = 26) were obtained from parallel layers of nearest profiles (P1 and P2). The second excavation also had similar soil composition to the first cultural layer; however, artifacts density was higher than the first cultural layer of the Ex.1 area. The artifacts pool  of the first cultural layer of Ex.2 consists of slags, ceramic fragments, crucible (including lid) fragments, weathered metal pieces, furnace wall fragments, beads, and some glass pieces.
The dark reddish-brown second cultural layer (almost similar to the first cultural layer) of Ex.1 and Ex.2 commonly consists of medium and coarse particles. The layer was almost 30 cm in Ex.1 and about 40 cm in Ex.2 in height ( Table 3). The same artifact types of the first cultural layer could be discovered in different contents in this wide layer. The crucible fragments in the Ex.1 and Ex.2 areas show morphological differences. The natural reddish-brown layer, discovered after the second culture layer, was slightly solid, consisting of fine, medium, coarse, and slightly very coarse particles. The minimum to maximum values of each layer's top and bottom levels and soil samples related to soil layers are represented in Table 3.

Geochemical characteristics of the soil and furnace wall fragments
A summary of the major and trace element concentrations of soil in the Ex.1 and Ex.2 areas with mean, minimum to maximum range, and the upper continental crust (UCC) is given in Table 2. The elemental concentrations of the cultural and natural soil layers of both areas showed slight variation against UCC (mean values in Table 2).
Major element concentrations are mentioned here as oxides (wt%), following the common convention for signifying major-element bulk chemical composition, chemical concentrations of crystalline oxides, and silicates (Piatak et al. 2015). Average abundance of major elements TiO 2 (cultural layers 1.68 wt%, natural layers 1.82 wt%, and furnace wall fragments 1.26 wt%), Fe 2 O 3 (6.97, 6.55, and 4.36 wt% respectively), MnO (0.21, 0.23, and 0.13 wt%), and MnO percentage concentrations were higher in the natural layers than in the others, while CaO and P 2 O 5 percentages were higher in the FWF ( Table 2). The abundance of heavy metals (As, Pb, Cu, Ni, Cr, and V) in both areas' cultural and natural layers was higher than UCC values. However, In the FWF, the concentrations of As, Pb, and Cu were lower than in UCC. In contrast, the zinc concentration in 94% of all samples was lower than the UCC. The cultural and natural layers of P1 and P2 (Ex.1 area) had a high percentage of copper, with a maximum value (470 ppm) nearly 17 times higher than relative to UCC.
The average abundance of lithophile and high field strength elements (HFSE) such as Sr, Y, Nb, Zr, Th, Sc, and halogens like bromine and iodine does not differ significantly between the two areas. However, the elements Sr and F average was lower than UCC and had a higher value of Zr, while the concentration of the other elements was relatively similar to the UCC (Table 2).

Visual examination of slags morphology
Almost all the slags found in this study were dislocated and broken. Thus, two kinds of slags, (a) high-iron metal typed slags (HIS) and (b) low-iron amorphous/glassy-typed slags (LIS), were collected from the site. Morphology comprises the shape parameters and surface roughness. Almost all slag samples were angular and irregular in shape and had porous (vesicular) structures. Most of the metal-typed slags were dark than amorphous slags in color. Among the metallic slag were parts of tap slag of various sizes. Glassy-typed slags denoted various ranges of blue, green, and white colors. Oxidized conditions were observed in many metal slags due to deposition after metal extraction and prolonged exposure to oxy-hydroxide.

Geochemical characteristics of slags
The summarized geochemical abundances of HIS and LIS are mentioned in Table 2 with mean, minimum to maximum range, and UCC values. The mean value of the major elements (TiO 2 , Fe 2 O 3 , MnO, CaO, and P 2 O 5 ) of the HIS was indicated as 1.01, 22.49, 0.47, 4.40, and 0.64 wt%, respectively. They were signified as 1.03, 7.02, 0.12, 9.66, and 0.57 wt% in the LIS means. The Fe 2 O 3 content in the HIS varies between 2.92 and 49.63 wt% (Table 2, Fig. 4d) and c. 2.28 and 26.19 wt% were in LIS. The Fe 2 O 3 average abundance shown in HIS was approximately 5 times higher than UCC. It was shown to be around 5 times MnO and 4 times P 2 O 5 than UCC. The average value of TiO 2 in HIS and LIS was nearly 2 times higher than in UCC.

Vertical distribution of elements Ex.1 and Ex.2 areas
More detailed geochemical signatures were studied on selected elements (As, Pb, Cu, Zn, Cr, Ni, Sr, V, Zr, Nb, Fe 2 O 3 , TiO 2 , MnO, and P 2 O 5 ) in the vertical soil layers of Ex.1 and Ex.2 areas (Figs. 5 and 6). Although the cultural and natural stratification of both areas is similar, the artifacts show the difference in territorial and cultural activities. The first and second cultural layers of Ex.1 do not indicate a drastic change in elemental composition, and such differences can be seen in the vertical layers of P1 and P2. Overall, Cu stands out among the elemental contents, suggesting significant differences in the cultural and natural layers of P1 and P2 (Fig. 5). The maximum copper value represents P1 and P2 anthropogenic layers and suggests the copper is exotic to the context, not expressing such a high composition from the nearest Ex.1 pit or entire Ex.2 area (Figs. 5, 6 and 7a). They are more likely to accumulate in the soil due to copperrelated production activities. Manganese is often associated with metals such as iron, copper, nickel, and cobalt, and the concentration may vary during extraction or raw material used in the premises (Fuerstenau & Han 1983). Therefore, changing the MnO of that area's cultural layers may result from copper metal works. The P 2 O 5 and Zn enrichment can be highly correlated with human waste and wood ash inputs (Aston et al. 1998). Thus, prolonged metallurgical activity over a long period may have influenced changes in other mineral compositions indicated by the vertical distribution ( Fig. 5a, b, c) of the Ex.1 area. Twenty-five samples from Ex.2(A) (n = 6), Ex.2(B) (n = 8), Ex.2(C) (n = 7), and P6 (n = 4) were studied to analyze the vertical concentration pattern of elements in Ex.2 area (Fig. 6). Significant variations in major oxides and heavy metals can be identified in the cultural layers of this region. It has been identified from much research in different areas that the composition of heavy metals and major oxides in soils varies with industrial mining and smelting (Aston et al. 1998;Gautam et al. 2016). However, low Cu content could be detected in the natural and cultural layers of the Ex.2 region.
Significant changes in the vertical deposition pattern of the elements in this premise were identified in the Ex.2(C) samples. The second cultural layer shows significant changes in elements such as Zn, Cr, Ni, V, MnO, P 2 O 5 , Sr, and Zr ( Figs. 6c and 7d, e, f, h). The vertical arrangement of elements such as As, Pb, Cu, Cr, Ti, V, and Nb remains stable, confirming that none of the region's significant cultural or natural activities affected those elements (Fig. 7g). Juleff (1996) states that the clay was prepared to construct the Samanalawewa experimental furnace based on indigenous knowledge of the area (by mixing black paddy-field mud, termite-mound earth, river sand, red alluvial gravel, fresh paddy husk, charred paddy husk, and cut paddy straw), which is durable and resistant to high temperatures. The use of such unique raw materials in the design of the Yodhawewa furnace wall may be reflected in the different elemental arrangements of Ex.02(c). The presence of phosphorus concentrations in Ex.02(c) samples (Fig. 7e), which are not present in the soil samples of the entire Yodhawewa research, also confirms the use of organic matter in the preparation of these clays.
The red color of the reddish earth results from the long-term release of free iron oxide from the ferromagnesian minerals (Reuter et al. 2020). As Koralegedara et al. (2021) point out, magnetite (~ 1 wt%) and ilmenite (~ 1 wt%) are the major Fe-containing minerals in the red soil of the Mannar region. Thus, it is confirmed that this region's earth contains a certain percentage of iron. However, the Fe content of the soil in the entire Yodhawewa area did not exceed twice the UCC value (5.04 wt%), and the maximum was 9.45 wt% ( Table 2 and Fig. 7b). In the Fe extraction process, it was common practice to extract near the ore deposit; however, when the soil composition Table 4 Elemental concentrations of selected slag samples related to  of the premises was examined, it was revealed that no such raw material had been obtained from the premises and that the metal processing ore had been imported from outside. The evaluation of soil and slag correlations against Tio 2 also does not show similar correlations between them (Table 5). Thus, Juleff (2013) points out significant magnetite/chalcopyrite deposits of Arippu and Kollan-Kulam areas, located quite nearby Yodhawewa site. It can be assumed that iron ore was brought from such an external deposit and extracted here.

Inter-element relationship of slags
The Yodhawewa geochemical data of HIS and LIS (n = 45) was subjected to correlation analysis to explore possible associations between variables (Tables 5). Titanium (TiO 2 ) indicated correlations with other HIS and LIS elements; their reduction or conversion during smelting is relatively low (Takeda et al. 2020). Analyzing the correlations between HIS and LIS in the slag sample revealed evident discrepancies. Coefficient correlations resulted under four   (Table 5). Zink (r = 0.95), Cr (r = 0.82), and V (r = 0.84) of HIS showed strong positive correlations against TiO 2 ( Fig. 8a and c). Among LISs, only Th presented a strong positive correlation (r = 0.92) (Fig. 8b). The overall slag analysis observed no strong negative correlations of elements; however, Cu represented weak negative correlations (r = − 0.30 and − 0.40) of HIS and LIS (Fig. 8e). Some elements show strong or moderate positive correlations in one slag type (Zn, V, Nb, and Zr) and do not show in another (Table 5 and Fig. 8c). The elements As, Ni, Sr, Y, F, Fe 2 O 3 , and MnO do not significantly correlate with both types of slags (Fig. 8d, f). Basal linear tendencies are rare in the elemental deposition pattern against TiO 2 and often have scatterings of one or both types of slag ( Fig. 8a-i).
In general, the stability and content of the major and trace elements change during a metal extraction process. Also, the metal ore elements at the extraction end can be integrated with the raw metal (result) in various quantities to strengthen or weaken the metal, mix with slags, or be lost during combustion (Coustures et al. 2003;Ettler et al. 2015). During metal extraction, refining, or steel-making stages, the correlation levels between the elements may also change due to the reduction or conversion of the essential elemental components (Coustures et al. 2003). The correlations between TiO 2 -Fe 2 O 3 , CaO-P 2 O 5 , Fe 2 O 3 -Cu, and Fe 2 O 3 -MnO (Fig. 8f-i) are always presented separately in HIS and LIS, suggesting that the results from two or more different processes. Figure 8i shows that a significant number of samples of HIS had a considerable amount of MnO (0.5-3 wt%). In a typical iron extraction process, elements such as nickel or copper combine with iron, and elements such as manganese and zinc are mixed into the slag (Tylecote 2002;Maslak and Skiba 2015). It allows defining that HIS is the waste of iron extraction.
The P 2 O 5 in HIS samples increases while CaO in LISs gradually increases. The high or low P 2 O 5 composition of the slag is mainly based on combustion nutrients (charcoal) used in the extraction process. Furthermore, phosphorus is reduced under conditions very close to iron, and the decisive factor is that charcoal behaves as essential to reducing conditions during iron smelting (Craddock 2000;Morel & Serneels 2021). However, in crucible steel production, a minimal amount of dried wood is incorporated into the crucible, which does not remove the slag as it absorbs enough of the steel ingot (Juleff 1996;Srinivasan & Ranganathan 2004). Charcoal contains about 5wt.% of ashes, including CaO (Morel & Serneels 2021). Also, due to the reduction properties of CaO, refining slags may contain a higher percentage of CaO than extraction slags. Some evidence use of various carbonates (such as oyster shells) in wootz steel production has been found in historical crucible-steel production sites of Southern Iran (Feuerbach 2002). Therefore, the relative CaO composition of LIS can be expected to be higher than that of HISs.

Upper continental crust-normalized geochemical characteristics of slags
Concerning the normalization of Yodhawewa slags, the average compositions of HIS and LIS types represent high, parallel, and departure patterns against the UCC (Fig. 9). Almost all heavy metals are shown at higher and lower levels of HIS above UCC, and Pb and Zn are aligned parallel to the regression line. Considering the average of major oxides, CaO was moderately high in LIS, while Fe 2 O 3 and MnO were high than UCC in HIS. However, TiO 2 and P 2 O 5 coincided with UCC levels. The lithophile element was almost all aligned with the reaction line, and only the Sr average of the LIS was high. All heavy metals are absorbed into HIS with a higher composition than LIS, and V reflects the highest density. However, it should be noted that some heavy metals, such as zinc (Zn) or arsenic oxide (As 2 O 3 ), may be volatilized partially or entirely during the extraction process (Morel & Serneels 2021). Since the melting point of vanadium is 1929 °C, it is more likely to combine with slag during the iron extraction process before homogenizing with the metallic phase (Morel & Serneels 2021;Moskalyk & Alfantazi 2003).
Furthermore, researchers predict that the lithophile elements (Sr, Y, Nb, Zr, Th, Sc, F, and Br) are also more likely to be combined with slags during the iron smelting process (Brauns et al. 2013;Coustures et al. 2003). Only Sr and CaO are higher in LIS than HIS among normalized lithophile elements and major oxides. Strontium (Sr) is an alkaline earth element whose properties closely follow the CaO origins of plagioclase, calcite, and feldspars (Simmons 1999). Thus, the composition of Sr with the raw material used for CaO requirement is also more likely to establish in LIS.

Conclusions
Archaeological excavations of the whole area show some key facts; (a) slags and crucibles confirm that copper work has occurred in the Ex.1 area. (b) Furnace factors, slags, and crucible fragments were confirmed that crucible steel was produced using lower half-spherical furnaces in the Ex.2 area. (c) Furnace wall fragments, tap-slag, and tuyere fragments confirm that iron extraction occurred parallel to other metalworks from first to the eighth century AD at the Yodhawewa metalworking site.
Analysis of the topography of the entire Yodhawewa site expresses the similarities in the cultural and natural stratification. The geochemical signatures of soil in the Ex.1 area ensure that the zone has long been used for copperbased metallurgical activities. The vertical distribution of soil elements and their correlations with TiO 2 suggest that historical metal activities have impacted the environment. Geochemical results indicate no evidence of iron-rich soil (iron ore) on the premises. Accordingly, source materials are more likely to be transported from other places. The high phosphorus content of the furnace walls ensures that organic matter is artificially added to the clay body during furnace construction to withstand high temperatures.
The correlations confirmed the high concentrations of Fe 2 O 3 , MnO, and P 2 O 5 in the high content of iron slags (HIS) acquired from the Yodhawewa site. Amorphous/glassy textured slag (LIS), correlation patterns, and normalized values are incompatible with HIS. Calcium and Sr, which show significant growth in LIS, reflect an artificial substitute for refining operations rather than a metal extraction process. Analyzing the inter-elemental relationship between HIS and LIS against TiO 2 does not reveal many strong positive or negative correlations and similar elemental correlations. It also confirms that these HIS and LIS originate from different production or refining methods.