The production of metal artefacts in Southern Etruria (Central Italy): case studies from copper to Iron Age

ABSTRACT An analytical study is presented, aimed to determine the elemental composition of copper-based artefacts dated back from Copper Age to Early Iron Age (mid-fourth millennium to the VIIIth century B.C.), found on the Tyrrhenian side of the peninsula, corresponding to the Lazio region. The objects belong to different archaeological contexts and had various functions. They were analysed by the X-ray fluorescence technique. The results highlight the experimental character of Copper Age metallurgy, which will later evolve in the established use of copper-tin alloys. Regarding the Bronze Age, despite the typological and functional heterogeneity of the artefacts and the wide chronological range, the alloys are relatively homogeneous in composition, with regular changes that appear related to chronology, according to what is already known for the Italian peninsula. Such changes are supposedly due to variations in the availability of tin, which was not locally mined. Early Iron Age metallurgy is represented by the Selvicciola Hoard solely, which restricts the possibility of generalizing the conclusions. A striking feature of the alloys is the great compositional difference between the complete and the fragmented artefacts. The formers are made of tin bronze, whereas in the latter tin is replaced by antimony and/or lead. The use of such unusual alloys is unlikely due to lack of metallurgical knowledge. Considering the urbanized communities that arose in the Middle-Tyrrhenian area during the Early Iron Age, we suppose that such variability in a single context might be related to a production system capable of using alloys of different quality and value to satisfy a diversified demand.


Introduction
The first studies concerning the elemental composition of prehistoric metal artefacts date back to 50s and the 60s of the past century (Otto and Witter 1952;Junghans, Sangmeister, and Schröder 1960, 1968, 1974. Although they included Italian findings too, this approach to the study of early metallurgy (Dolfini and Giardino 2015) played a back-line role among Italian scholars. Only in recent decades, it gave rise to systematic research projects like, for instance, the examination of the prehistorical bronzes at the Museo Civico di Storia Naturale of Verona (Aspes 2011).
The present paper reports the firstpartial and provisionaloutcomes of a research carried out in the framework of the Memorandum of Understanding signed in 2017 between the Soprintendenza Archeologia per il Lazio ed Etruria Meridionale and the Istituto Nazionale di Fisica Nucleare-Laboratori Nazionali di Frascati (INFN-LNF), which aimed to share research, scientific advice and training related to prehistoric metallurgy.
The research regards the systematic analysis of preand proto-historic metallic findings coming from areas in the provinces of Rome and Viterbo, in Central Tyrrhenian Italy. The typology of the items and the size of the groups is affected by the vicissitudes of archaeological research and by events that sometimes brought to spread unitary contexts in different museums. Thus, the artefacts are not homogeneously distributed from a geographical point of view: at present, most objects come from an area between the Middle Valley of River Fiora and the lake Bolsena, for the north, and an area among the Tyrrhenian Coast, the Tolfa mountains and the lake Bracciano, for the south (Appendix and Figure 1).
The objects so far examined cover a chronological range which conventionally spreads from the beginning of the Eneolithic (mid-fourth millennium B.C.) to the Early Iron Age (XIth-VIIIth century B.C.). It is not aim of the present paper to discuss the chronological attribution of the artefacts. It is taken from the cited literature and assumed correct. The items come from different types of archaeological contexts and many are isolated findings. Their function is not always unambiguously recognized: they might have been for daily use or be valuable and ceremonial artefacts. At the end of the research, it will be possible to use the function as a variable together with geographic distribution and chronology, since it might have affected the choice of the alloy and the production technique.

Experimental method
Analyses of the artefacts were performed by X-ray fluorescence (XRF) (Ferretti 2014), on surfaces from which corrosion products had been previously removed. The experimental conditions are summarized in Table 1. They are established to provide both efficient excitation and well-resolved detection of such key role elements as Ag, Sn and Sb. In such conditions, detection limits calculated with the single standard method (Jenkins, Gould, and Gedke 1995) are approximately 400 mg/kg for Fe and Ni, 250 mg/kg for Zn, 50 mg/kg for Ag, Sn, Sb and 300 mg/kg for Pb. The overall relative accuracy is 5-10% for all elements. Accuracy is estimated by calibration standards as the mean deviation between the measured and the nominal concentration. Quantification was carried out by the combined use of calibration standards and the PyMCA Fundamental Parameters software package (Solé et al. 2007). The only exception is for As, whose quantification relies on Fundamental Parameters solely. Whenever possible, multiple measurements were performed on different parts of the object, depending on the size and conservation conditions. All the charts shown in the subsequent sections refer to mean values calculated over all the measurements performed on the object.

Experimental results and discussion
A detailed description of the investigated artefacts, the normalized composition for each measurement (Table A1) and the mean composition of each artefact (Table A2) are given in the Appendix.
This work refers to copper-based alloys solely, whereas objects containing silver, antimony and lead as main components are not included. It is immediately apparent that the observed compositional differences are correlated to the development of metallurgical know-how and, therefore, to the chronology of the artefacts. This aspect is examined and discussed widely  in the literature for other contexts of the same period, in different areas of the Italian peninsula.

Eneolithic
The most ancient period here considered corresponds to the chronological range of Eneolithic (in Italy between the mid-fourth and the end of the IIIrd millennium B.C.). The objects are axes, dagger blades and awls; halberd blades were not considered since this kind of artefacts is missing in the local museums involved in this project.
A preliminary analysis of the data set shows differences in the contents of Co, Zn, As and Sb seemingly related to the area they come from. Figures 2 and 3 show the Zn-Co and Sb-Zn scatterplots, respectively.
Although the groups are not equally numerous, which affects the significance of the comparison, they suggest the use of differnt ore sources. Even though Co concentrations are close to the detection limit, it can be observed that artefacts from the northern area tend to be higher in Co and lower in Zn, respect to  the southern one. Moreover, the artefacts from the northern area tend to be high in Sb, except for the dagger blade Selvicciola necropolis 3. Figure 4 shows the Sb-As scatterplot for axes, daggers and awls solely. It appears that different alloys were used for axes and daggers, with the latter characterized by a variable but detectable presence of As and Sb. Three different groups can be distinguished: the first group has relatively high As and low Sb; the second group has high As and high Sb and the third group, consisting in a single item, has low As and high Sb. Figure 4 also shows that As content of the daggers is significantly higher than 1 wt%, which leads assuming that it was deliberately alloyed with copper and that this was a common practice in both areas. Figure 5 goes deeper in detail and compares the compositional profiles of the dagger blades. The bars account for the mean concentration calculated over all the measurements performed on the item.  The daggers Poggio Spolverino 1, Selvicciola necropolis 1 and Selvicciola necropolis 2 (see Appendix) are made of arsenical copper: the content of As is high, while Zn and Ag are low and comparable; two dagger blades from the northern area (Selvicciola necropolis 3 and Ortaccia 2) show significant amounts of As and Sb, which are supposedly due to the use of fahlerz minerals. The last dagger blade of the group (Poggio degli Spiriti 1) has a compositional profile completely different from the others and characterized by relatively high amounts of Ni, Ag, Sb and by low As. The latter item seems compositionally similar to artefacts from northern Italy, dated at the beginning of the Early Bronze Age, for which it was recognized the use of fahlerz minerals (De Marinis 2006). Figure 4 shows that axes have low concentrations of As and Sb compared to daggers. In particular, from the compositional profiles of axes alone ( Figure 6), it appears higher heterogeneity respect to the daggers ( Figure 5): the use of fahlerz minerals for the axe Grotta Fichina 1 and of arsenical copper for the axe Monte S. Angelo 1 is recognizable. For the other artefacts, ore source is more uncertain. The Zn content is around the detection limit in all axes.
Awls, which are the most widely distributed items all over the Italian territory, at the moment do not provide significant information.
It is to be noted that the use of fahlerz minerals that we observe here in this period, does not occur in the Late Eneolithic in northern Italy (De Marinis 2006).

Bronze Age
In the investigated areas, metallic findings belonging to the initial phase of the Early Bronze Age are missing, opposite to what happens in northern Italy. Here, for the facies of Polada, an initial horizon is documented (Early Bronze Age IA) corresponding to the Middle-European phase A1 (De Marinis 2006). It is known that the copper used in this phase was smelted from fahlerz minerals, with high concentrations of As, Sb, Ag and Ni, whereas Sn, Pb, Bi, Fe at low or trace levels. The artefacts investigated in this work pertain to a chronologically and technologically more advanced phase of the Early Bronze Age when the use of binary Cu-Sn alloys was established. Figure 7, where the artefacts are presented in chronological order, shows that, for the Early Bronze Age, Sn concentrations reach a peak approximately corresponding to the so-called "horizon 3rd of the hoards", followed by a decrease extending to the whole Middle Bronze Age and by a new increase in the Late Bronze Age.
Further analysis of the data set shows that the compositional changes are correlated to chronology more than to geography, Zn, Ag and Pb being the most effective discriminators. Figure 8 shows the Ag-Zn scatterplot, with different colors used to distinguish among different periods of the Bronze Age. It is observed that both Zn and Ag are below 1 wt%. The former is relatively high in the Early Bronze Age, then it tends to stabilize at lower levels. The latter tends to increase and reaches the highest levels in the Final Bronze Age. Figure 9 shows thePb-Zn scatterplot for the Bronze Age artefacts. Both elements are relatively high in the Early Bronze Age and tend to decrease in later phases.
An anomalous presence of Fe, especially concentrated in the group of artefacts coming from lake Mezzano, is apparent. Such behavior might be explained, at least in part, by assuming the use of slagging in the smelting process (Craddock and Meeks 1987;Garagnani, Spinedi, and Baffetti 1993). Figure 10 shows the bar-chart of Fe concentration for all the artefacts, including the Eneolithic and Early Iron Age ones. The increase in Fe content at the transition from Eneolithic to Early Bronze Age is consistent with Craddock and Meeks 1987 if one supposes that the use of slagging was introduced at that time. The question is complex since no slags or furnaces were found in the investigated areas.
The slagging process, however, does not explain the high (up to 20 wt%) and extremely variable Fe concentrations observed in some artefacts from Lake Mezzano, for which intentional Fe-Cu alloys cannot be excluded. Further research including microstructural analysis will be carried out to clear this point.
Finally, it might be necessary to consider the effect of the ferrous underwater sources present in Lake Mezzano, although surface iron deposits should have been removed when the artefact was prepared for analysis.

Early Iron Age
The Early Iron Age is a complex period on the Middle-Tyrrhenian side of the Italian peninsula. The historical processes that started in the Middle Bronze Age lead to complex urbanization phenomena. The production of metal objects was subjected to further increase, as proved by the grave goods found in the necropolises situated around the centers that, at the end of the VIIIth century B.C., became the Etruscan cities.
All the Early Iron Age artefacts are from the Selvicciola hoard. This hoard, accidentally discovered in 1957, was soon dispersed. F. Rittatore Vonwiller could recover and return to the Superintendence only a small part of it. The loss of information resulting from the incompleteness of the hoard is, even more, penalizing if one considers that several items have unusual compositions and need deeper investigation.  An important distinction concerns the conditions of the artefacts: few are complete, whereas the majority is fragmented. Their function, whether axe, ring, bracelet or ingot, is however recognizable. Table 2 shows the mutual correlation coefficients of the elements concentration. It appears that some elements, i.e. Co, Ni, As, Ag and Sb, are positively correlated to one another, which leads assuming that they came into the alloy altogether in the (Sb-rich) ore. This point will be addressed further below. All these elements are low in the complete items and high in the fragmented ones. Figure 11 provides an example of the correlation for the pair Sb-As.
Another effective element to distinguish between the two groups is Zn, which is high in the complete items and low in the fragmented ones, as shown in Figure 12.
The most striking feature, however, are the elements intentionally alloyed with copper, provided one considers 1 wt% as the borderline of intentional addition. Figures 13-15 show the Sn-Sb, Sn-Pb and Sb-Pb scatterplots. In the complete artefacts, Cu is alloyed with Sn, whereas Sb and Pb are below 1 wt %. Conversely, in most fragmented objects Cu is alloyed with Sb and Pb (Sn is below 1 wt%), except two axe-blades, which contain Pb but no Sb and two fragments, which contain Sb, but no Pb.
The concentration of As, in the range of 1-2.5 wt%, is on the borderline of intentional addition. However, its correlation with Sb ( Figure 11), which is assumed to be the intentional alloying element, leads to conclude that it was not deliberately added.
It appears that the elemental composition of the hoard items is strongly correlated to the conditions of the object, whether it is complete or fragmented. Taking into account the processes that lead to the formation of a hoard and given the chronological homogeneity of all the artefacts, it is clear that the observed compositional differences do not depend on chronology, nor on lack of metallurgical knowledge. The analysis of the characteristics of Early Iron Age societies, which in the Middle Tyrrhenian Italy of the 8th century B.C. are highly Figure 11. Sb-As scatterplot for the Early Iron Age artefacts; circles and diamonds refer to the fragmented and complete objects, respectively. urbanized, brings us to suppose that such differences might be related to a diversified production system, capable of using alloys of different quality and value to meet a diversified demand.

Conclusions
As already pointed out, this paper aims at presenting the first outcomes of an ongoing project. At this stage, it is, therefore, more appropriate to formulate hypotheses than draw ultimate conclusions. Concerning the Eneolithic, the analysis carried out on three different types of items, i.e. axes, daggers and awls, confirms what is known for the whole Italian peninsula, i.e. alloys of "deliberate different composition" were used for the production of axes and dagger blades (De Marinis 2006). In the case of daggers, the difference between copper rich in As and copper rich  in both As and Sb is supposed to be connected to the chronological development of Eneolithic, namely to the beginning and to the advanced phase, respectively, of Central Italy Eneolithic. Indeed, this hypothesis may be affected by the formation process of an archaeological deposit inside a common burial (De Marinis 2006), from which most of the analysed artefacts come.
Compositional differences between axes and dagger blades are known to be part of the experimental character of Eneolithic metallurgy. Search for standardized solutions can explain the higher compositional variability of Eneolithic artefacts respect to Bronze Age ones.
An overall view of Eneolithic metallurgy requires to mention, besides the copper-based alloys discussed in this paper, Cu-Ag alloys as well as the frequent and documented use of Sb alone, in particular for ornaments. Whereas Sb artefacts are in general extremely  rare in the Italian and European settlements, they are relatively common in Central Tyrrhenian Italy, with special regard for the middle valley of River Fiora. Moreover, in the Italian Eneolithic, objects made of Cu-Ag alloys can be found. Another extraordinary finding is a fragment of an iron awl, discovered in tomb 21 of the Selvicciola necropolis. This find does not imply to backdate the beginning of iron metallurgy from the 1st to the 4th millennium B.C., but it documents a phase of alloying and process experiments (Grazzi et al. 2012).
Compared to Eneolithic, the Bronze Age shows an extraordinary increase consisting not only in larger numbers but also in new types of objects now made of metal.
Copper-based alloys have now reached a standard composition (at least from the qualitative point of view) involving the use of Sn as the main alloying element. Its concentration has relatively regular variations versus time. According to literature, it was not mined locally, therefore we hypothesize that such variations are due to corresponding variations in Sn availability.
Changes in the content of Ag, Zn, and Pb, all considered unintentional, are possibly related to changes in the smelting technology. Conversely, changes in ore supply are not likely since, for both the northern and the southern area, mineral sources were locally available (Tanelli 1985;Zifferero 1992).
Regarding the Early Iron Age, the data on the Selvicciola Hoard should be interpreted as resulting from a commercial strategy, rather than a technological regression. At the moment we have found no comparisons in the literature. In the case of S. Francesco hoard (Antonacci Sanpaolo, Canziani Ricci, and Follo 1992), for example, there seems to be no difference in composition between fragments and complete artefacts. Thus, at the present stage of the research, the results obtained for the Selvicciola hoard must be considered strictly limited to the area of Vulci.

Disclosure statement
No potential conflict of interest was reported by the authors.

Notes on contributors
Adolfo Esposito is a Radiation Protection Expert of the National Institute of Nuclear Physics (INFN) at National Laboratories of Frascati (LNF) in Italy. His main expertise is on the Radiation Measurements field (X, gamma and neutron spectrometry). A part of his research work regards the application of XRF technique to cultural heritage artifacts. He is co-author of publications on journal, books and congress proceeding.
Patrizia Petitti is an archaeologist specialized in prehistory and proto-history. She worked at the Ministry for Heritage and Cultural Activities until 2018. Her interests concern, besides ancient metallurgy, also the Copper Age and underwater archaeology. She is the author or co-author of many works, published on journals, books and congress proceedings.
Marco Ferretti is a senior research scientist at the National Research Council of Italy (CNR). Most of his research work regards the technical examination of ancient metal artefacts and the design of portable X-ray fluorescence equipment. He is the author or co-author of more than 100 works, published on journals, books and congress proceedings.
Astrik Gorghinian is a high school teacher and scientist collaborator at the Institute of Nuclear Physics (INFN)-LNF. She has acquired experience on the X-ray fluorescence technique analysis directed to Cultural Heritage Goods. She is coauthor of publications on journals and books, regarding ancient metals, gemstones and coins.
Fabio Rossi is an archaeologist and director of the Museum of Prehistory of the Tuscia and of the Rocca Farnese of Valentano (VT). Most of his research work focuses on the prehistory and proto-history of the central Italy, with particular regard to metal artefacts. He is the author or coauthor of many works, published on journals, books and congress proceedings. (1) Dagger blade from tomb 15: probably it is an autonomous type; in association to a primary deposition, (1) San Francesco type winged axe, variety B, dated back to the second half of the 8th century B.C. (Carancini 1984, tav. 85-88); (2) Winged axe; dated back to the half of the 8th century B.C. (Carancini 1984, tav. 8-11); (3) Vetulonia type winged axe; dated back to the second half of the 9th century B.C. (Carancini 1984, tav. 1-4); (4) 4 fragments of winged axe, not definable; (5) 6 fragments of axe blade; (6) Fragmented axe wing; (7) Gouge-awl; (8) Fragment of a ring with circular section; (9) Fragment of a bracelet with rhomboidal section; (10) Ingot fragment with flat-convex section; (11) Ingot fragment with trapezoidal section.  Petitti et al. 2009, 22, fig. 12 (1) Sickle similar to the Poggio Berni type, second-third horizon of the Final Bronze Age hoards (Carancini 1979a, 635, fig. 2, 20;Carancini and Peroni 1999, tav. 32 (1) Naviglione type flat axe (Carancini 1993 , fig 2,  (1) Complete pin with biconical end round off on the top;