Early metallurgy in SE Iberia. The workshop of Las Pilas (Mojácar, Almería, Spain)

Abstract

Big narratives on the role of metallurgy in social change and technological innovations are common in archaeology. However, informed discussion of these issues requires a contextualised characterisation of metallurgical technology at the local level in its specific social and technological contexts. This paper approaches early metallurgy in Iberia from a technological perspective. We focus on the site of Las Pilas in the Vera Basin (Mojácar, Almería, Spain), where the whole metallurgical chaîne opératoire has been documented in situ through archaeological excavation of a third millennium bc context. The study includes microstructural, mineralogical and chemical analyses of ores, slag, technical ceramics and finished artefacts, as well as domestic pottery used for comparative purposes. These results are discussed with reference to the archaeological context and evidence for other domestic activities and crafts. Our aim is to contribute to better characterise the early metallurgical tradition of Southeast Iberia, paying particular attention to specific technological tools, knowledge and recipes that may allow future comparative approaches to knowledge transmission or independent innovation debates. For this particular case, we demonstrate the direct production of arsenical copper in a low-scale, low-specialisation, low-efficiency set up that involved the crucible smelting of complex oxidic ores in a context that suggests associations with cereal roasting and, indirectly, with basket and pottery making.

Full text

ORIGINAL PAPER
Early metallurgy in SE Iberia. The workshop of Las Pilas
(Mojácar, Almería, Spain)
Mercedes Murillo-Barroso
1
&Marcos Martinón-Torres
1
&
Mª Dolores Camalich Massieu
2
&Dimas Martín Socas
2
&Fernando Molina González
3
Received: 29 July 2016 /Accepted: 6 December 2016
#Springer-Verlag Berlin Heidelberg 2017
Abstract Big narratives on the role of metallurgy in social
change and technological innovations are common in archae-
ology. However, informed discussion of these issues requires
a contextualised characterisation of metallurgical technology
at the local level in its specific social and technological con-
texts. This paper approaches early metallurgy in Iberia from a
technological perspective. We focus on the site of Las Pilas in
the Vera Basin (Mojácar, Almería, Spain), where the whole
metallurgical chaîne opératoire has been documented in situ
through archaeological excavation of a third millennium BC
context. The study includes microstructural, mineralogical
and chemical analyses of ores, slag, technical ceramics and
finished artefacts, as well as domestic pottery used for com-
parative purposes. These results are discussed with reference
to the archaeological context and evidence for other domestic
activities and crafts. Our aim is to contribute to better charac-
terise the early metallurgical tradition of Southeast Iberia, pay-
ing particular attention to specific technological tools, knowl-
edge and recipes that may allow future comparative ap-
proaches to knowledge transmission or independent innova-
tion debates. For this particular case, we demonstrate the di-
rect production of arsenical copper in a low-scale, low-spe-
cialisation, low-efficiency set up that involved the crucible
smelting of complex oxidic ores in a context that suggests
associations with cereal roasting and, indirectly, with basket
and pottery making.
Keywords Early metallurgy .Prehistoric technology .
Arsenical copper .Slag .Iberia .Copper age
Introduction
Technology has traditionally been considered of essential im-
portance in social change, given that key technological inno-
vations have the capacity to cause profound social transfor-
mations. However, we should be wary of possible assump-
tions implicit in this premise: there is a tendency to relate
technology to ‘progress’and from this surmise that more com-
plex technological systems equate to superior societies, which
are typically seen as the source of the knowledge transmitted
to ‘lower’societies.
In the realm of prehistoric metallurgy, V. Gordon Childe
created a persuasive model whereby metallurgy was seen as a
highly complex and socially transformative technology re-
quiring full-time specialists. Metallurgy was thought to have
been developed by the civilisations of the near east, from
Electronic supplementary material The online version of this article
(doi:10.1007/s12520-016-0451-8) contains supplementary material,
which is available to authorized users.
*Mercedes Murillo-Barroso
m.murillo-barroso@ucl.ac.uk
Marcos Martinón-Torres
m.martinon-torres@ucl.ac.uk
Mª Dolores Camalich Massieu
dmassieu@ull.edu.es
Dimas Martín Socas
dsocas@ull.edu.es
Fernando Molina González
molinag@ugr.es
1
UCL Institute of Archaeology, 31-34 Gordon Square,
London WC1H 0PY, UK
2
Departamento de Prehistoria, Arqueología, Antropología e Historia
Antigua, Universidad de La Laguna Campus de Guajara, 38205 La
Laguna, Spain
3
Departamento de Prehistoria y Arqueología, Universidad de Granada
Campus Universitario de Cartuja, 18071 Granada, Spain
Archaeol Anthropol Sci
DOI 10.1007/s12520-016-0451-8

where it would have spread into Europe. More recent work
has proposed that metallurgy could have been developed
independently in more than one location (e.g. Renfrew
1969; Ruiz-Taboada and Montero-Ruiz 1999; Höppner
et al. 2005; Radivojevićet al. 2010); it has questioned
its purported impact in social structures (e.g. Montero
Ruiz 1994; Bartelheim 2007;Kienlin2010,2016), and it
has emphasised that aesthetic rather than functional adap-
tations may have been major factors shaping metallurgical
traditions (e.g. Smith 1982; Aranda et al. 2012;Martinón-
Torres and Uribe-Villegas 2015).
Informed discussion of the above issues requires a
contextualised characterisation of metallurgical technology
at the local level. In our view, this approach is likely to allow
a better understanding of the social impact of metallurgy and
the organisation of production, besides providing more de-
tailed information to address issues of invention and knowl-
edge transmission. Southeast Iberia has been consistently fea-
tured on debates about the origins and social impact of metal-
lurgy, but contextual studies have been limited.
This paper approaches early metallurgy in Iberia from a
technological perspective. We focus on the site of Las Pilas
in the Vera Basin (Mojácar, Almería, Spain), where the whole
metallurgical chaîne opératoire has been documented in situ
through archaeological excavation. Metallurgical contexts are
defined and analytical results of archaeometallurgical remains
are presented. Our aim is to contribute to better define the
metallurgical tradition of Southeast Iberia, paying particular
attention to specific technological tools, knowledge and rec-
ipes that may allow future comparative approaches to knowl-
edge transmission or independent innovation debates. Beyond
strictly technical aspects, we discuss craft organisation and
aim to contribute to a contextual understanding of early met-
allurgy in Iberia.
Archaeological background
The Vera Basin, in the Iberian Southeast, is a large tertiary
basin spanning some 320 km
2
and traversed by three rivers:
Aguas, Antas and Almanzora. It is framed by the Almagro and
Almagrera mountain ranges in the north; Cabrera mountain
range in the south and Bédar and Lisbona ranges in the West
and the Mediterranean Sea in the East (Fig. 1).
Thanks to the extensive archaeological works that started
in the late nineteenth century by Henry and especially Louis
Siret, we know of the extensive occupation of the Vera Basin
during the late prehistory, with early metallurgical stages con-
stituting a decisive moment of occupation (Camalich Massieu
and Martín Socas 1999). With additional work at third millen-
nium BC sites such as Almizaraque, Zájara or Campos
(Cuevas del Almanzora) as well as at second millennium BC
sites such as El Argar (Antas), Fuente Álamo (Cuevas del
Almanzora) or Gatas (Turre), the Vera Basin has become a
priority study area, with much archaeological research espe-
cially devoted to investigate the role of early metallurgy in the
process of social stratification.
The site of Las Pilas/Huerta Seca (Mojácar), with an ap-
proximate surface area of 5 ha, is located in this broader con-
text (UTM 30S 601913, 4111690). It is placed on top of a
plateau 30 m above sea level, close to the estuary of the river
Aguas and flanked by two watercourses flowing into it.
The site was discovered in 1989 and it was subjected to
three archaeological campaigns in 1990, 1991 and 1994
(Alcaraz Hernández 1992; Camalich Massieu and Martín
Socas 1999; Rovira Buendía 2007). In the last campaign, an
area of 40 m
2
was excavated. The occupation of the site was
structured in a sequence of ten phases, based on documented
stages of the restructuration and reorganisation of the
inhabited area. Based on the absolute dates, this occupation
took place during the third millennium BC, with the most re-
cent phase (phase 10) defined by the occurrence of material
culture associated to the Bell Beaker horizon.
In terms of constructive features, the settlement is
characterised by round huts, some of them partially excavated
on the ground, generally built on a stone and clay plinth, with
rammed-earth walls. They usually have a central post to hold
the conical roof made of vegetal lattices waterproofed with
clay. Several negative structures related to grain and water
storage are associated to these huts. Some large walls have
been documented too, possibly related to the demarcation or
defence of this sector, which clearly underwent frequent func-
tional reorganisation of the domestic and craft activity areas.
The last excavation campaign yielded a substantial
archaeometallurgical assemblage. Notwithstanding one cop-
per awl and one copper mineral fragment recovered in phases
1 and 2, respectively, metal-working debris (i.e., slag and a
copper lump) appeared from phase 5 and reached the highest
frequency in phase 9. Phase 5 is defined by three negative
storage structures associated to a round hut partially excavated
on the ground. After their useful life, these structures were
intentionally filled up with waste derived from household
and craft activities. Medium-sized stones resulting from the
dismantling of previous stone plinths due to a restructuration
of the inhabited space have also been documented as filling
materials. Two small copper mineral fragments, one slag frag-
ment as well as two corroded fragments of copper were found
in two of these structures. Four Hordeum vulgare nudum
seeds from this phase were dated by accelerator mass spec-
trometry (AMS) (Table 1
1
), three of them from the same de-
positional contexts than the metallurgical remains. They show
that the first metallurgical activity documented in this sector of
1
Three samples of charcoal entrapped in slag fragments were also dated and
results areincluded in Table 1. However, dates discussed in the text are based
on short lived samples to avoid the old wood effect.
Archaeol Anthropol Sci

the site occurred during the first quarter of the third millenni-
um BC. If we consider the chronocultural sequence proposed
for the Iberian southeast, this would be associated to the sec-
ond phase of the Formative Period (Castro et al. 1996)or
Early Copper Age (Molina et al. 2004).
However, the most complex and complete evidence of met-
allurgical activity was recovered from phase 9. This phase is
characterised by the building of a 60–70-cm-wide wall demar-
cating an area where mainly two activities were carried out:
cereal processing and metal and melting. This area, of circular
or oval tendency, is c. 6 m in diameter and is delimited by a
ditch, 30-cm wide and 15-cm deep. In this area, three post
holes and two combustion structures lined with clay were
identified (Fig. 2). The central one was primarily used for
cereal roasting while the second one, in a peripheral area,
was connected to metal smelting and melting. This structure,
only partially preserved (60–70 cm), had a circular/oval shape
and is delimited by adobe bricks, with the ground completely
vitrified by high temperatures. A minimum of five ceramic
blowpipe nozzles were found in this structure (Fig. 2) together
with a complete crucible. Copper droplets as well as slag and
crucible fragments were also recovered in the interior of this
structure, as well as in a border area where most of the metal-
lurgical waste was discarded. Remains associated to the com-
plete metallurgical sequence have been documented in this
area: from mining (one stone hammer, two grinding stones
and several ore fragments), through smelting and melting
(slag, crucibles, blowpipe nozzles and copper droplets) to fin-
ished objects.
Based on six AMS dates, five of them on Triticum aestivum
durum seeds and one on H. vulgare nudum (Table 1), this
phase developed during the second half of the third millenni-
um BC (2578–2276 2σcal. BC), that is to say in an advanced
period of the Iberian Chalcolithic when, according to the
periodisation proposed for the area, the first evidence for
Bell Beakers are documented [Late Beaker Chalcolithic
(Castro et al. 1996) or Late Copper Age (Molina et al.
2004)]. This area was affected by a fire that caused its col-
lapse, and the space was then reorganised with the building of
a round hut over it which corresponds to the Bell Beaker phase
(phase 10).
Materials and methods
More than 240 archaeometallurgical finds were recovered dur-
ing the archaeological excavation of the site. The collection
includes 70 ore fragments (combined weight 383 g), 93 slag
fragments (377 g), 24 slagged crucible sherds, 1 complete
crucible, 13 fragments of ceramic blowpipe nozzles (MNI
five), 42 corroded copper droplets or lumps (14.7 g), 2 copper
awls (2.7 and 1.8 g, respectively), 1 possible fragmented burin
(3.5 g), one stone hammer and two grinding stones (Fig. 3). In
order to reconstruct the whole metallurgical chaîne
opératoire, selected materials were sampled for further analy-
ses at the UCL Institute of Archaeology’sWolfson
Archaeological Science Laboratories. We employed a strati-
graphic sampling frame that considered archaeological and
Fig. 1 Location of the site in
relation to the main mining
districts in the area and other sites
mentioned in the text
Archaeol Anthropol Sci

typological information, as well as the results of screening
analyses of 130 objects by portable x-ray fluorescence
(pXRF) using an Olympus Innov-X Systems Delta
Premium. Thirty-one samples were selected from across the
compositional groups qualitatively identified by pXRF, and
including all archaeological phases with archaeometallurgical
remains. The sample set included 11 ore fragments, 9 slag
fragments, 9 crucible sherds, 1 blowpipe nozzle, and 2 awls
(Table 2). Six additional samples of domestic pottery were
also studied for comparison with technical ceramics.
Samples were mounted in epoxy resin and polished to
0.25 μm. Optical microscopy under both plane polarised
light (PPL) and cross polarised light (XPL) was used to
identify areas of interest for further analyses by scanning
electron microscopy with energy-dispersive spectrometry
(SEM-EDS), which were performed with a Philips XL30
with an Oxford Instruments x-sight EDS. The SEM-EDS
system used an accelerating voltage of 20 kV, a working
distance of 10 mm, a spot size of 5.3, and a process time
of 5, corresponding to a dead time of c. 30 %; acquisition
time was 100 s. The certified arsenic copper standard
BCR 691-C from the European Commission was used to
monitor the reliability of the analyses. Data was processed
by INCA spectrometer software, outputting data as ele-
ments for metal phases, and adding oxygen by stoichiom-
etry in ceramics and slag. Chemical compositions of slag
Tabl e 1 Conventional radiocarbon age and calibrated AMS results of the metallurgical phases (OxCal 4.2 software; Intcal13 calibration curve). The
highest probability is highlighted in bold
Site Location Phase Laboratory Material Date BP Cal. BC 1σ(68.2 %) Cal. BC 2σ(95.4 %) C13/C12
Las Pilas Mojácar, Almería 5 Beta-408051 H. vulgare nudum 4220 ± 30 2894–2866 (34.5%)
2804–2762 (33.7%)
2905–2853 (42.8 %)
2813–2743 (42.8%)
2727–2696 (9.9%)
−21.5 o/oo
Las Pilas Mojácar, Almería 5 Beta-408053 H.vulgare nudum 4210 ± 30 2890–2864 (26.4 %)
2806–2760 (39.7 %)
2716–2713 (2.1 %)
2900–2848 (33.3 %)
2814–2737 (47.7 %)
2731–2679 (14.5 %)
−22.9 o/oo
Las Pilas Mojácar, Almería 5 Beta-408054 H.vulgare nudum 4200 ± 30 2886–2861 (20.0 %)
2808–2757 (40.5 %)
2718–2706 (7.7 %)
2894–2841 (27.1 %)
2814–2678 (68.3 %)
−22.7 o/oo
Las Pilas Mojácar, Almería 5 Beta-408055 H.vulgare nudum 4120 ± 30 2856–2811 (21.3 %)
2747–2724 (10.2 %)
2698–2624 (36.7 %)
2866–2804 (25.1 %)
2777–2579 (70.3 %)
−24.5 o/oo
Las Pilas Mojácar, Almería 9 Beta-403262 Charcoal within slag 4130 ± 30 2858–2831 (13.5 %)
2821–2809 (5.5 %)
2753–2721 (15.3 %)
2702–2631 (33.9 %)
2872–2798 (27.0 %)
2794–2786 (1.1 %)
2780–2617 (62.9 %)
2610–2582 (4.5 %)
−21.9 o/oo
Las Pilas Mojácar, Almería 9 Beta-408061 T. aestivum durum 3980 ± 30 2565–2526 (36.8 %)
2496–2468 (31.4 %)
2578–2457 (95.4 %) −21.8 o/oo
Las Pilas Mojácar, Almería 9 Beta-408063 T.aestivum durum 3950 ± 30 2562–2535 (14.6 %)
2492–2454 (38.4 %)
2418–2407 (5.5 %)
2376–2351 (9.7 %)
2568–2521 (19.7 %)
2499–2346 (75.7 %)
−22.9 o/oo
Las Pilas Mojácar, Almería 9 Beta-408062 T.aestivum durum 3880 ± 30 2455–2418 (20.7 %)
2408–2336 (39.4 %)
2323–2308 (8.2 %)
2467–2286 (94.2 %)
2247–2236 (1.2 %)
−19.3 o/oo
Las Pilas Mojácar, Almería 9 Beta-408060 T.aestivum durum 3870 ± 30 2454–2418 (18.0 %)
2407–2376 (16.9 %)
2350–2293 (33.3 %)
2465–2278 (89.7 %)
2251–2229 (4.3 %)
2220–2211 (1.4 %)
−21.4 o/oo
Las Pilas Mojácar, Almería 9 Beta-408064 H.vulgare nudum 3860 ± 30 2454–2418 (14.8 %)
2406–2376 (14.8 %)
2350–2286 (38.6 %)
2461–2276 (84.3 %)
2254–2210 (11.1 %)
−19.9 o/oo
Las Pilas Mojácar, Almería 9 Beta-408065 T.aestivum durum 3860 ± 30 2454–2418 (14.8 %)
2406–2376 (14.8 %)
2350–2286 (38.6 %)
2461–2276 (84.3 %)
2254–2210 (11.1 %)
−20.9 o/oo
Las Pilas Mojácar, Almería 9 Beta-403257 Charcoal within slag 3820 ± 30 2299–2202 (68.2 %) 2448–2446 (0.2 %)
2436–2420 (1.4 %)
2405–2378 (3.5 %)
2350–2193 (84.9 %)
2176–2144 (5.3 %)
−21.6 o/oo
Las Pilas Mojácar, Almería 10 Ua-48819 Charcoal within slag 4155 ± 43 2871–2837 (20.2 %)
2815–2799 (9.5 %)
2793–2785 (4.8 %)
2780–2673 (65.3 %)
2882–2619 (98.6 %) −17.0 o/oo
Archaeol Anthropol Sci

samples are averages from several analyses trying to
avoid large inclusions and corroded areas. For ceramics,
we analysed both large areas to obtain an estimate of the
‘bulk’composition, and smaller areas focused on the ma-
trix and devoid of large inclusions. All data have been
normalised to 100 % but analytical totals are provided.
We acknowledge that SEM-EDS area analyses of porous
and coarse-grained materials such as those reported here
are more prone to mineralogical effects that affect the
accuracy of quantitative values, as reflected in the vari-
able analytical totals for our results. However, this tech-
nique remains the most cost-effective for the study of slagged
crucibles, as it provides the combination of microstructural
and chemical information necessary for a technological study.
Fig. 2 Archaeological context in
which most of the
archaeometallurgical remains
were found. On the right,details
of the two combustion structures:
top the central fire and the
complete crucible found in situ;
below the blowpipe nozzles found
in situ
Fig. 3 Some of the
archaeometallurgical materials
found in the site: astone hammer;
b,cmills; dore fragments; e
blowpipe nozzles; fthermally
altered bricks on which blowing
pipes were recovered; gslag
fragments; ha complete crucible;
icrucible sherds; jmetal objects
Archaeol Anthropol Sci

X-ray diffraction (XRD) was conducted on 10 ore samples
to complement the identification of mineral species. XRD was
performed using a Rigaku MiniFlex 600 x-ray diffractometer
with a Cu(Ka) target, a tube voltage of 40 kV and a tube
current of 15 mA.
Trace element analysis of ore samples were done at the
Geochronology and Geochemistry SgIker-Facility of the
University of the Basque Country UPV/EHU (Spain) (see
Supplementary Material 1for analytical procedures).
Results
Ores
Ore fragments are mostly ofsmall size with an average weight
of5.5g,andthebiggestfragmentsreachingupto79g.
Malachite and azurite are the main copper mineral phases,
within two maintypes of gangue macroscopically identifiable:
either dark ferrous or whitish calcareous/dolomitic.
Tabl e 2 List of samples analysed Type Phase Sample ID OM SEM-EDS XRD ICP-MS
Ore 7 5-8479 X X X X
Ore 8 5-6726 X X X X
Ore 9 5-6740 X X
Ore 9 5-7706 X X X X
Ore 9 5-6972 X X X X
Ore 9 5-6491 X X X X
Ore 9 5-6826 X X X
Ore 9 5-6902 X X X
Ore 9 5-6597 X X X
Ore 9 5-7422 X X X
Ore 9 5-6732 X X X
Slag 5 5-8815 X X
Slag 9 5-6816 X X
Slag 9 5-6825 X X
Slag 9 5-6855 X X
Slag 9 5-6918 X X
Slag 9 5-6912 X X
Slag 9 5-6916 X X
Slag 9 5-6933 X X
Slag 9 5-6606 X X
Slagged crucible 9 5-6727 X X
Slagged crucible 9 5-7479 X X
Slagged crucible 9 5-6815 X X
Slagged crucible 9 5-6966 X X
Slagged crucible 9 5-6974 X X
Slagged crucible 9 5-6608 X X
Slagged crucible 9 5-6616 X X
Slagged crucible 9 5-6618 X X
Slagged crucible 9 5-6735 X X
Blowing pipe nozzle 9 5-7250 X X
Domestic vessel 5-1158 X X
Domestic vessel 5-2312-1 X X
Domestic vessel 5-4128-2 X X
Domestic vessel 5-5007-2 X X
Domestic vessel 5-4145-19 X X
Domestic vessel 5-4087-2 X X
Copper awl 1 5-11,454 X X
Copper lump 5 5-8862 X X
Copper awl 9 5-2997 X X
Archaeol Anthropol Sci

Qualitative pXRF analyses were conducted on all ore samples
recovered. Most of them turned out to be complex copper
minerals bearing high levels of arsenic, zinc and lead. Four
qualitative compositional groups were identified on the basis
of pXRF: mainly Cu, CuAs, CuAsZn and CuAsZnPb without
any clear correlation between these groups and the two types
of gangue identified. Six samples were then selected for ICP-
MS and ten samples for XRD, optical microscopy and SEM-
EDS, which allowed the identification of complex copper ox-
ides, carbonates and arseniates such as calcio-duftite,
conichalcite or olivenite (Tables 3and 4). Most of the samples
bear relatively large amounts of zinc; this element appears
usually combined with arsenic and copper forming complex
phases such as cuprian adamite or zincolivenite, but it can also
occur as a silicate (willemite) or carbonate (smithsonite) as
well as enrichment in malachite. Sulphides are comparatively
scarce, although some galena inclusions were documented.
Some samples also contain phases with significant levels of
cobalt and nickel, especially sample 6975 (Table 4), as well as
small inclusions of perroudite, a sulpho-halide of mercury and
silver (Fig. 4).
The chemistry and mineralogy of the specimens described
are consistent with those of ores documented in the surround-
ing mining districts (Montero Ruiz 1994; Favreau et al. 2013).
In particular, three copper deposits may have been the source
of the ores identified in Las Pilas: Pinar de Bédar, Herrerías/
Almagrera and Cerro Minado, respectively, c. 10, 20 and
30 km from the site, as the crow flies (Fig. 1).
The copper mineralisations from Pinar de Bédar are con-
sistent withthose from Las Pilas asthey have arsenic and zinc
and lead as major elements accessory to copper (Montero Ruiz
1994, p. 177). Although there is no evidence of prehistoric
exploitation in Pinar de Bédar, this could have been obliterat-
ed by intense works in modern times. Thus, we consider it a
possible ore source area due to proximity to the site and con-
cordance in composition.
Cerro Minado has some similarities with the ores document-
ed in Las Pilas too, and it is likely to have been exploited in
prehistory. Domergue (1987) identified some stone mining
tools and classified it as a Bronze Age mine, and recent surveys
have documented stone peaks and hammers inside the mine
(Delgado Raak et al. 2014). An absolute date, MAMS-18508
3905 ± 21 BP (Delgado Raak et al. 2014, p. 30) confirms its
exploitation during the Copper Age contemporaneous with the
occupation of several sites in the area such as Almizaraque,
Campos or Las Pilas. The elemental composition of some geo-
logical samples reported high levels of arsenic (up to 42 %) and
the frequent occurrence of sulphidic phases, but the most char-
acteristic feature of copper minerals from Cerro Minado is the
typicallyhighcobalt(upto0.9%)andnickel(upto1.2%)
(Favreau et al. 2013; Delgado Raak et al. 2014,p.19)—atrait
also shared with some samples from Las Pilas. This is due to the
co-existence of erythrite [Co
3
(AsO
4
)
2
·8H
2
O] and annabergite
[Ni
3
(AsO
4
)
2
·8H
2
O] with copper minerals and arsenates. Native
silver and Hg-bearing silver have also been documented in
Cerro Minado (Bertran-Oller et al. 2012,p.247;Favreau
et al. 2013,p.36)—which is again consistent with the micro-
analysis of some Las Pilas samples, and perroudite is known to
occur in the Almeria province in the mines of Las Cocotas and
Rodalquilar (Favreau et al. 2013,p.37);however,themain
native silver deposit in the area is Herrerías (c. 20 km from
the site), where some specimens reach up to c. 13 % Hg
(Bartelheim et al. 2012, p. 200; Murillo-Barroso et al. 2014).
Copper minerals from Sierra Almagrera/Herrerías also bear
high levels of arsenic (Montero Ruiz 1994, p. 177), and two
Cu–Zn and Cu–Zn–Fe metallogenic phases have been docu-
mented in Sierra Almagrera (Martínez Frías et al. 1989, p. 265).
Thus, there are several potential ore sources for Las Pilas, and it
is possible that more than one were exploited. An ongoing lead
isotope analysis programme in the area will hopefully shed
more light on copper sourcing and will allow us to characterise
the structure and organisation of copper mining in the area.
Technical ceramics
Three types of technical ceramics have been found in Las Pilas:
the blowpipe nozzles and two types of crucibles. The blowpipe
nozzles are cylindrical and have a maximum length of 13.2 cm.
With walls c. 7–10-mm thick, the internal diameter is c. 15 mm;
however, this diameter is reduced at the tip, which is domed
and bears a small perforation of only 4–5mm(Fig.5). They are
made of white-firing fabrics and show evidence of vitrification
at the tip, although not to catastrophic extents. Examination of
the impressions left on the inner surface of these objects indi-
cates that they would have formed with fresh clay applied over
one end of a hollow reed tied with ropes. The tip would then be
perforated from the outside, leaving diagnostic burrs on the
inside. Such manufacture is consistent with recent experimental
studies, which also concluded that fresh-clay nozzles had a
more efficient and longer use, than those fired prior to use
(Obón Zúñiga and Berdejo Arcéiz 2013).
No ceramic tuyérès or blowpipe nozzles had been docu-
mented up to now in the archaeometallurgical record for pre-
historic Southeast Iberia, even at sites where large amounts of
metallurgical debris such as crucibles and slag have been
found, such as Los Millares or Almizaraque. This has led to
the suggestion that the main air supply systems would have
been either natural draft, or reeds used as blowpipes without
any ceramic nozzle (e.g. Gómez Ramos 1996). The only
Chalcolithic nozzles hitherto recovered in Iberia come from
the northeast, Central Portugal and the southwest. Two frag-
ments of cylindrical tuyérès with an internal diameter of 7–
8 mm were recovered in La Bauma del Serrat del Pont
(Tortellà, Girona) (Alcaide et al. 1998,p.91).Incentral
Portugal, one conical nozzle with a maximum internal diam-
eter of c. 20 mm was recovered in Pedra do Ouro (Hunt 2003,
Archaeol Anthropol Sci

Tab l e 3 Mineral phases identified in ore samples and technique of identification. Due to the complexity of the ores analysed, only the main phases were identified by XRD, while minor phases and
inclusions were detected by SEM-EDS
ID 5-6491 5-6597 5-6726 5-6746 5-6826 5-6902 5-6972 5-7422 5-7706 5-8479 Tech.
Azurite Cu
3
(CO
3
)
2
(OH)
2
XXX XXX X SEM-EDS,XRD
Malachite Cu
2
(CO
3
)(OH)
2
XXXX XXXXXSEM-EDS,XRD
Zn-Malachite XXX X SEM-EDS
Conichalcite CaCu(AsO
4
)(OH) XX SEM-EDS
Olivenite Cu
2
(AsO
4
)(OH) XSEM-EDS
Ni-Olivenite XSEM-EDS
Zincolivenite Cu-Zn(AsO
4
)(OH) X X X X SEM-EDS
Cuprian Adamite (Zn,Cu)
2
AsO
4
OH X X X X SEM-EDS, XRD
Calcio-Duftite (Pb,Ca)CuAsO
4
(OH) XSEM-EDS
As-Claraite (Cu, Zn)
3
(CO
3
)(OH)
4
•4H
2
OX SEM-EDS
Chrysocolla (Cu,Al)
2
H
2
Si
2
O
5
(OH)
4
•nH
2
O X XXX SEM-EDS
Zn-Chrysocolla X X SEM-EDS
Willemite Zn
2
SiO
4
XX SEM-EDS
Smithsonite ZnCO
3
XSEM-EDS
Zn-Mimetite Pb
5
(AsO
4
)
3
Cl X SEM-EDS
Perroudite (Inclusions) Hg
5
Ag
4
S
5
(I,Br)
2
Cl
2
XX XSEM-EDS
Galena (Inclusions) PbS X X SEM-EDS
Cerusite (Inclusions) PbCO
3
XSEM-EDS
Mawbyite (Inclusions) PbFe
3+2
(AsO
4
)
2
(OH)
2
XSEM-EDS
Muscovite KAl
2
(AlSi
3
O
10
)(OH)
2
XXSEM-EDS
Dolomite CaMg(CO
3
)
2
XXXXSEM-EDS,XRD
Limonite FeO(OH) nH
2
OXXXXSEM-EDS,XRD
Archaeol Anthropol Sci

Tab l e 4 Q-ICP-MS results of ore samples
Na Mg Al P K Ca Sc Ti V Cr Fe Mn Co Ni Cu Zn As Rb Sr Y Ag
ID ppm ppm ppm ppm ppm ppm ppm ppm ppm ppm ppm ppm ppm ppm % ppm ppm ppm ppm ppm ppm
5-6726 60.58 7644 122.7 355.3 344 21950 0.143 2.379 74.51 2.183 4708 20.16 32.69 182.3 5.824 56070 41220 0.673 25890 2.31 5.451
5-6491 65.26 4681 493.7 806.8 374.6 2634 0.4881 6.66 100.5 3.778 13070 17.95 62.24 391.2 25.54 60520 70610 1.269 319.6 10.04 46.66
5-6740 65.82 3640 5001 333 2257 1443 4.163 79.14 23.22 5.879 7157 18.78 19.25 15.72 22.96 448.8 385.8 9.641 99.09 2.147 0.2895
5-7706 54.42 46780 569.6 567.7 377 11860 0.895 9.234 55.12 2.906 2965 44.14 63.3 206.9 13.13 37820 43830 1.429 462.8 7.233 33.89
5-6972 153.8 1984 1319 3461 668.6 5541 0.467 4.574 95.3 2.575 9409 183.6 4546 9731 23.84 4599 68290 1.509 498.2 2.331 310.3
5-8479 39.39 1182 592.7 635.7 248.4 3506 0.4735 4.19 106 2.524 77310 109.4 408.3 760.5 28.26 912.7 4998 0.5987 107 3.373 273.2
Sb Ba Pb Bi La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu Th U
ID ppm ppm ppm ppm ppm ppm ppm ppm ppm ppm ppm ppm ppm ppm ppm ppm ppm ppm ppm ppm
5-6726 27.76 541.9 29100 25.76 0.49 0.38 0.24 86.11
5-6491 151.3 58.72 38290 90.37 2.80 2.01 0.50 1.81 0.43 0.16 0.45 0.12 0.46 0.15 0.38 0.10 0.36 0.11 0.19 103.90
5-6740 1.823 14.8 342.80 0.85 2.96 4.65 0.85 3.31 0.64 0.21 0.54 0.45 0.09 0.28 0.26 0.94 20.22
5-7706 21.21 1426 15040 699.10 2.66 3.36 0.61 2.61 0.61 0.29 0.70 0.11 0.73 0.15 0.45 0.38 0.37 437.50
5-6972 60.77 117.2 538.90 783.10 1.69 3.01 0.34 1.30 0.33 0.10 0.49 0.43 0.22 0.22 0.39 127.00
5-8479 441.1 7.513 209.70 121.00 1.64 3.27 0.37 1.47 0.36 0.08 0.44 0.40 0.24 0.21 0.15 10.02
Archaeol Anthropol Sci

p. 303) and one cylindrical nozzle with a rounded end and a
smaller perforation, very similar to those found in Las Pilas,
was recovered in Vila Nova de Sao Pedro (Jalhay and Paço
1945: Lam. XXI). However, this nozzle would also constitute
an exception in Portugal, as no other nozzles have been
recovered in sites where the complete metallurgical process
has been recovered in situ such as the rounded house V of
Zambujal (Müller et al. 2007;Gauss2015). In Extremadura,
one possible tuyérè was recovered from the site of La
Sierrecilla (Santa Amalia, Badajoz) (Cruz Berrocal et al.
Fig. 4 Some of the mineral
phases identified in ore samples.
Ca-Duf calcio-Duftite; Chr
chrysocolla; Co conichalcite; Cu-
Ad cuprian adamite; Dol
dolomite; Gal galena; Mal
3malachite; Mus muscovite; Mw
mawbyite; Ni-Ol nickel-olivine;
Pe perroutite; Wi willemite; Zn-
Mal Zn-rich malachite; Zn-Ol
zincolivenite
Archaeol Anthropol Sci

2006, p. 65) while the best known examples are those from
Valencina de la Concepción (Sevilla) in the southwest (Nocete
et al. 2008, p. 728) although their exceptionally large internal
diameters (between 20 and 60 mm) would have prevented
their use as blowpipes.
2
Experimental studies have concluded
that the internal diameters of blowpipes usually fall in the
range of 5–10 mm, while tuyérès attached to bellows are about
15–35 mm and tuyérès or openings for natural draft furnaces
from 50 to over 100 mm (Rehder 1994, p. 348). Against this
background, all Copper Age tuyérès documented in Iberia up
to now would have been attached to bellows, except for the
examples ofLas Pilas and possibly, Vila Nova de Sao Pedro—
hence, our choice of the term ‘blowpipe nozzles’. The capa-
bilities of blowpipes are constrained by the power of the hu-
man breath, while air supplied by bellows can develop higher
temperatures and generate heat at about 70 times the rate de-
veloped by blowpipes (Rehder 1994, p. 349); therefore, the
use of blowpipes instead ofbellows generally indicates small-
er scale metallurgy. Bellows would be essential for larger and
more productive furnaces. In any case, as larger drafts are
usually linked to hollowed furnaces, and up to now we do
not have evidence for any such furnace in Chalcolithic
Iberia, the overall scarcity of blowpipes and the presence of
crucible sherds in most of the metallurgical sites would sug-
gest the extended use of reeds without any ceramic nozzle
over small crucibles.
Turning to the crucible sherds, these are much more com-
mon in the archaeological record. At Las Pilas, they usually
consist of heavily slagged amorphous fragments with slag
layers up to 3 mm thick. The rim of the vessel is occasionally
conserved, in some cases, with the slag layer overflowing it.
Although they are usually too fragmented to reconstruct the
typology of the crucibles, at least two types have been identi-
fied. Due to the lack of curvature of the rim and their flatter,
shallow shape, some of them seem to correspond to the so-
called crucible-furnace (horno vasija),i.e.flatopenvesselsof
1–2 cm in thickness and up to 50 cm in diameter (Rovira
2002, p. 89) (Fig. 6). In one sample, organic imprints were
documented on the outside surface of the sherd, reflecting a
plain weave basketry (Alfaro Giner 1984,p.113)thatseems
to correspond to an esparto basket used to mould the vessel.
Fresh clay would have been applied inside the esparto basket
and pressed against it to form the crucible; the mould would
have been either removed prior to use, lost during firing, or
2
We are not including here the only tuyérè recovered at Cabezo Juré (Alosno,
Huelva) due to its imprecise context and especially its morphology. Regarding
its context, a journal article (Nocete 2006, p. 651) claims that it was recovered
in the Southern Slope, where the so-called furnacesare described; however, in
the Spanish monograph on the site (Nocete 2004), this purported tuyérè is
drawn together with the domestic pottery and ascribed to US14 (Nocete
et al. 2004, p. 141, Fig. 8.11)—a stratigraphic unit described as sealing a
storage structure on the Upper Platform of the site and therefore not related
to the ‘furnaces’in the Southern Slope. Furthermore, in the micro-spatial
analysis of the site, this artefact is drawn on Northern Slope (Nocete 2004,
p. 349). This micro-spatial analysis also presents significant discrepancies with
the spatial description of the site published in the journal paper: while the site is
described as markedly functionally divided in the English version (Bactivities
were rigorously demarcated by function: processing of ore took place on the
south slope and copper casting in the residential area to the north. The fortress
[on the Upper Platform], by contrast, featured no metal-working^[Nocete
2006, p. 647]), remains of ores, crucibles and slags are drawn on both slopes
in the Spanish monograph (Nocete 2004, p. 369). The morphology of the
tuyérè also raises some doubtson its functionality, as it has a bi-conical profile
with a maximum diameter of c. 60 mm. This shape, which would seem inef-
ficient for a tuyérè, is reminescent of a domestic pottery type known in Iberia
as a ‘support’(see,e.g.Hunt2003). The same morphology and hence dubious
ascription applies to as at least two of the items described as tuyérès in
Valencina de la Concepción (Nocete et al. 2008,p.728).
Fig. 5 Drawings of some of the
blowpipe nozzles. Note the
internal imprints of the nozzles
Archaeol Anthropol Sci

decayed subsequently (Fig. 7). This ceramic technique has
been documented in domestic pottery at several sites in
Iberia including the nearby sites of Campos and Zájara, as it
is a common technique for the manufacture of open vessels
such as large dishes or platters (e.g. Camalich Massieu and
Martín Socas 1999; Delibes de Castro et al. 1998, pp. 166–
167; Valiente Cánovas et al. 2003). However, it has not been
documented in crucible manufacture up to now, which sug-
gests that pottery making technology did not require any sig-
nificant adaptation for metallurgy. The second crucible type
corresponds to vessels of rectangular shape. One of them was
recovered intact, and its dimensions are 27 cm × 6 cm in plan,
with a height of 4 cm and walls of up to c. 2 cm in thickness.
These crucibles display a much thinner slag layer and could
have been related to melting rather than smelting (see discus-
sion below).
The fabrics of rectangular crucibles are usually greyish
while ‘crucible-furnaces’are either greyish or orange, al-
though both of them typically show a gradient from a darker
inner surface to a lighter outside, consistent with exposure to
high temperatures and heating from the top—as typical for
prehistoric crucibles. They both are rather coarse in texture,
with abundant inclusions of quartz and potassium feldspar
reaching grain sizes of up to 1 and 3 mm, respectively. For
comparison, six samples of domestic pottery were selected.
Three of them are similarly coarse, with rough surfaces, while
the other three samples have finer textures and burnished
surfaces.
The compositions of the ceramic matrices of crucible and
blowpipe samples, obtained by SEM-EDS area scans that
avoided large inclusions, show exceptionally high alumina
levels (averages of 33.6 and 32.8 %, respectively). This fea-
ture is typical of highly refractory ceramics that proliferate
since the late Middle Ages (Martinón-Torres and Rehren
2009,2014). However, they also display remarkably high
concentrations of alkali and alkali earth oxides (averages of
9.4 and 8. 8%), which would have diminished their thermal
stability (Tables 5and 6). When compared to common pottery,
they display broadly similar compositions (Fig. 8). Bulk com-
positions are enriched in SiO
2
because of the abundant silicate
minerals, and they fall in the area defined by Freestone and
Tite (1986) as typical of ancient technical ceramics and build-
ing bricks. The compositionsof the ceramic matrices avoiding
quartz grains fall outside this area due to the higher alumina
Fig. 7 To p l e f t one crucible sherd
of a ‘crucible furnace’with
basketry imprints of its
manufacture in its outer surface.
Bottom left example of a Neolithic
basket from Cueva de los
Murciélagos (Albuñol, Granada)
at the Spanish National
Archaeological Museum. Right
example of a pottery vessel with
basketry imprints from the nearby
site of Campos (Camalich
Massieu and Martín Socas 1999;
Fig. 12-8)
Fig. 6 Sherds of ‘vase-furnace’crucibles. Note the slag flowing over the
rimofonesherd(top right) and through a crack of one broken crucible
(bottom right)
Archaeol Anthropol Sci

Tabl e 5 ‘Bulk’ceramic compositions obtained by SEM-EDS analyses
of large areas. Averages of up to six analyses per sample are reported.
Data are in wt% and normalised, with oxygen added by stoichiometry.
The low analytical totals in the technical ceramics are due to higher
vitrification and porosity. Tr = Traces (below 0.5 %)
ID Na
2
OSD MgO SD Al
2
O
3
SD SiO
2
SD K
2
OSD CaO SD TiO
2
SD FeO SD Total
Crucibles 5-6974 0.8 0.1 1.1 0.1 20.6 0.6 63.6 1.4 3.8 0.1 1.1 0.0 0.9 0.1 8.1 0.3 59.8
5-6618 0.8 0.3 1.4 0.1 22.1 1.6 60.6 2.2 3.9 0.4 0.9 0.4 1.0 0.1 9.3 0.2 65.0
5-7479 0.7 0.0 1.6 0.0 16.5 0.3 70.3 0.0 2.6 0.1 1.1 0.1 0.7 0.1 6.5 0.3 53.1
5-6608 0.8 0.1 1.5 0.1 24.0 0.3 57.5 1.4 4.8 0.6 1.3 0.5 tr 9.8 0.3 74.0
5-6735 0.9 0.2 1.7 0.4 23.0 2.9 55.9 6.0 4.3 1.0 5.4 3.4 0.8 0.3 7.9 1.8 78.3
5-6727 0.7 0.1 1.3 0.1 24.0 0.6 57.4 1.8 3.4 0.3 2.1 0.4 0.8 0.1 10.3 1.0 59.4
5-6616 0.7 0.1 1.3 0.2 24.5 0.7 56.5 1.1 3.4 0.1 1.3 0.3 1.2 0.3 11.1 0.3 52.3
5-6815 0.8 0.1 1.2 0.1 22.7 1.4 58.8 3.4 3.9 0.3 1.3 0.2 0.9 0.1 10.5 1.6 71.9
Average 0.8 0.1 1.4 0.2 22.2 2.5 60.1 4.5 3.8 0.6 1.8 1.4 0.8 0.2 9.2 1.5 64.2
Blowpipe 5-7250 0.9 0.3 0.9 0.0 21.0 1.8 62.0 2.5 4.6 0.5 1.1 0.4 0.9 0.1 8.7 0.4 59.3
Domestic pottery Coarse 5-5007-2 2.5 0.3 1.3 0.0 21.4 1.2 61.8 2.1 4.1 0.2 0.9 0.1 0.7 0.1 7.3 0.3 112.3
5-4087-2 1.1 0.2 1.4 0.0 21.3 0.4 60.2 0.5 5.0 0.1 0.7 0.1 1.0 0.0 9.4 0.1 111.9
5–1158 0.8 0.2 1.4 0.1 25.6 0.3 55.8 0.7 4.4 0.1 tr 0.9 0.1 10.7 0.8 90.6
Average 1.5 0.7 1.4 0.0 22.8 2.0 59.3 2.5 4.5 0.4 0.7 0.2 0.9 0.1 9.1 1.4 104.9
Thin 5-4145-19 0.7 0.1 2.1 0.0 18.4 0.0 62.7 0.6 3.5 0.1 5.0 0.8 0.7 0.0 7.0 0.2 105.5
5-2312-1 1.1 0.0 2.4 0.0 22.2 1.2 59.7 1.5 4.6 0.4 1.3 0.1 0.9 0.0 7.9 0.1 112.5
5-4128 0.9 0.0 1.3 0.1 21.0 0.8 60.3 0.6 5.3 0.2 1.2 0.3 0.7 0.1 9.3 0.4 105.6
Average 0.9 0.2 1.9 0.5 20.5 1.6 60.9 1.3 1.5 0.7 2.5 1.8 0.8 0.1 8.1 0.9 107.9
Tabl e 6 Matrix ceramics compositions obtained by SEM-EDS. Average of up to 6 analyses per sample are reported. Areas analysed were selected
avoiding any inclusions. Data are normalised in wt% with oxygen added by stoichiometry. Tr = Traces (below 0.5%)
ID Na
2
OSD MgO SD Al
2
O
3
SD SiO
2
SD K
2
OSD CaO SD TiO
2
SD FeO SD Total
Crucibles 5-6727 1.0 0.1 1.6 0.1 33.2 2.1 47.8 2.7 4.2 0.4 1.2 0.2 tr 10.6 0.1 95.4
5-6974 1.2 0.2 1.0 0.2 35.3 3.6 47.9 0.6 5.7 0.1 0.8 0.2 0.6 0.6 7.5 2.0 101.9
5-6618 1.2 0.1 1.2 0.2 35.9 0.8 45.7 2.8 5.9 0.3 1.5 0.3 tr 8.0 1.6 97.0
5-6608 1.0 0.4 1.7 0.1 33.5 4.0 47.2 4.7 5.4 1.3 0.7 0.2 0.5 0.1 9.9 1.2 103.5
5-6735 1.2 0.4 2.5 0.2 32.5 0.5 49.1 3.3 6.6 1.2 5.5 1.8 0.6 0.1 11.2 1.3 109.1
5-7479 1.0 0.1 1.9 0.7 33.5 4.9 50.5 1.4 4.7 1.1 1.2 0.5 tr 6.8 2.3 107.1
5-6616 0.9 0.1 1.3 0.3 34.8 0.6 47.7 0.6 4.3 0.1 1.3 0.1 0.5 0.1 9.2 0.9 94.7
5-6815 1.0 0.1 1.5 0.2 32.8 0.8 45.8 1.5 4.9 0.2 0.9 0.2 tr 12.5 1.7 101.3
Average 1.1 0.1 1.6 0.4 33.6 1.8 47.2 1.6 5.5 0.7 1.6 1.3 0.4 0.1 9.4 1.7 101.2
Blowpipe 7250 1.6 0.2 1.4 0.2 33.0 0.6 45.5 1.5 5.4 0.3 0.9 1.4 0.6 0.2 11.5 1.2 96.5
Domestic vessels Coarse 5-5007-2 1.1 0.6 1.7 0.5 28.8 2.2 52.3 1.9 7.5 1.4 0.8 0.4 tr 7.7 1.5 88.4
5-4087-2 1.7 1.2 1.3 0.2 28.4 2.9 51.9 0.8 5.3 0.7 0.7 0.1 tr 10.3 2.3 88.6
5-1158 0.9 0.2 1.3 0.3 29.5 3.5 52.5 4.2 5.2 1.4 tr tr 9.9 2.7 106.2
Average 1.2 0.3 1.4 0.2 28.9 0.4 52.2 0.2 6.0 1.0 0.6 0.1 tr 9.3 1.1 94.4
Thin 5-4145-19 1.0 0.2 1.2 0.0 34.9 0.2 50.3 0.2 8.5 0.4 tr tr 3.9 0.3 99.7
5-2312-1 0.7 0.0 2.2 0.8 30.6 4.0 50.4 1.1 8.4 1.0 0.9 0.6 tr 6.5 1.3 115.3
5-4128 0.8 0.5 2.3 1.9 29.8 4.5 50.6 3.2 7.9 1.6 tr tr 8.3 4.6 119.5
Average 0.8 0.1 1.9 0.5 31.7 2.2 50.4 0.1 8.2 0.2 0.4 0.3 tr 6.2 1.8 111.5
Archaeol Anthropol Sci

concentrations, although still plotting far from modern refrac-
tories because of their high alkali and alkaline-earth levels.
The chemical composition of the technical ceramic matrices
is remarkably similar to those of domestic pottery, especially
the fine wares, most notably in the high alumina, even if there
are slight differences such as the higher lime and lower potash
of the former. This similarity indicates that clay refractoriness
was not a major concern for ancient metallurgists—or, at least,
that specific clays were not reserved for the manufacture of
technical ceramics. However, a more specific study including
petrographic analyses of domestic pottery as well as technical
ceramics (crucibles and blow pipe nozzles but also loom
weights and adobes) is currently ongoing, which will shed
more light on the technological choices on pottery making.
Goingbacktothecrucibles,thecoarsegrainsizeandabun-
dance of temper would have contributed to improve the perfor-
mance of the vessels—although it should again be noted that
domestic pottery showed similar fabrics, hence arguing against
the hypothesis of a specialised metallurgical ceramic recipe.
Quartz temper is known to improve the performance of techni-
cal ceramics not only by virtue of its own refractoriness, but also
because of its expansion during heating and subsequent contrac-
tion upon cooling; this process creates a network of voids and
microcracks that improve toughness and thermal shock resis-
tance (Martinón-Torres and Rehren 2009, p. 56; cf. Tite et al.
2001; Kilikoglou et al. 1998). Feldspars behave differently and
have lower melting points; while in the right proportion and
grain size, they may act as fluxes and promote the crystallisation
of highly refractory mullite (Martinón-Torres et al. 2006,2008),
the melting during use of large feldspar grains such as those in
the Las Pilas crucibles could potentially create weak points and
lead to catastrophic failure, especially if high temperatures were
sustained for long periods. The same applies to some iron oxide
minerals also documented in the fabrics, and which often appear
clearly melted. Organic temper is not documented in any cruci-
ble in Las Pilas, contrary to many prehistoric crucibles in the
Old World, commonly tempered with organics (Bayley and
Rehren 2007,p.47).
All in all, these technical ceramics do not reveal the selection
of specific clays or tempering materials, and they are unlikely to
have been exceptionally refractory. This is consistent with the
heavy vitrification and bloating displayed towards the inner
surfaces, and the significant interaction between the slag and
the molten ceramic. In at least one case, the crucible clearly
failed during use, and slag can be seen flowing through the
crack to the bottom of the vessel (Fig. 9). Having said this,
the material properties of the crucibles were sufficient to smelt
copper, even if they were exposed to the very limit of their
thermal ability. In this, they are comparable to most Old
World prehistoric crucibles (Bayley and Rehren 2007;
Martinón-Torres and Rehren 2014).
Slag and slagged ceramics
Smelting is attested at the site both by small slag lumps and
thick slag layers on crucible-furnace fragments. Most slag frag-
ments are of small size (2–3 cm except for three larger pieces of
5 cm) and have irregular shape and an average mass of 4 g, with
Fig. 8 Bulk and matrix
compositions of technical and
domestic ceramics compared to
compositions of actual
refractories shown in the ternary
diagram FeO + alkali and alkaline
earths SiO2 Al2O
Archaeol Anthropol Sci

the largest fragment reaching 109 g (Fig. 10). No slag cakes or
larger lumps have been recovered. This level of fragmentation
would seem consistent with the crushing of slags by metal-
workers in order to recover copper prills trapped within (e.g.
Bachmann 1982;Rovira2002). Fragments of embedded char-
coal are often visible, as well as mineral relicts. In fact, some of
these lumps would best be described as only partially reacted
minerals which could derive from failed or incomplete smelting
operations, or at least from areas of the crucible where the
reaction was far from complete. Interestingly, however, they
appear not to have been regarded as worthy of further process-
ing or re-smelting. Slag layers in smelting crucible fragments
are usually thick (some exceeding 2 mm) and in some cases,
they overflow the rim of the smelting vessel (Fig. 6). They
display a greenish surface colour, denoting their copper-rich
nature; small charcoal remains are also macroscopically identi-
fiable in some samples.
The smelting process is affected primarily by four factors:
the composition of the charge and technical ceramics, the
firing temperature, the redox conditions and the length of the
process (Hauptmann 2007, p. 20). All of these parameters can
be inferred from the compositional and mineralogical analyses
of slag fragments.
Fig. 9 a Common quartz and b
feldspar inclusions in fabric
ceramics. Both pictures taken
frombottomareasofcrucibles
with less vitrification. cSlag
flowing through a cracked
crucible exposed to the very limit
of its refractoriness. d–fHigh
vitrification of ceramic fabrics
and high interaction between slag
and ceramic
Fig. 10 Some of slag samples
Archaeol Anthropol Sci

Tab l e 7 ‘Bulk’slag compositions obtained by SEM-EDS. Average of up to 6 analyses per sample are reported. Areas analysed were selected trying to include all representative features and avoiding big
inclusions and corroded zones. Data are normalised in wt% with oxygen added by stoichiometry. Low analytical totals are due to the presence of carbonates and porosity. Tr = Traces (below 0.5%)
ID MgO SD Al
2
O
3
SD SiO
2
SD SSD Cl SD K
2
OSD CaO SD TiO
2
SD FeO SD CoO SD
Slag fragments 5–8815 4.4 1.0 1.0 0.2 1.0 0.3 47.7 5.3
5–6855 0.6 0.6 tr 12.5 2.0 1.8 0.5 tr 2.8 0.5 20.0 7.7
5–6916 8.6 6.2 1.0 0.8 2.6 1.8 2.4 0.4 14.1 9.9
5–6825 5.7 1.7 6.6 4.9 22.5 1.1 tr 0.7 0.1 0.5 0.4 7.7 2.9 0.5 0.3 12.3 4.3
5–6933 tr 0.9 0.6 9.8 3.1 1.6 0.7 2.3 0.8 3.4 0.8 11.0 3.1 0.8 0.8
5–6606 3.2 0.8 1.5 0.3 15.6 1.9 0.8 0.2 0.1 0.1 5.6 0.9 8.7 1.4
5–6816 5.9 3.6 tr 12.6 1.8 tr 0.8 0.6 0.1 0.1 10.6 2.3 tr 5.0 1.2
5–6912 9.3 2.7 8.5 2.0 0.5 0.7 1.4 0.4 11.3 2.9 3.1 1.1
5–6918 2.0 0.1 7.9 6.2 3.5 2.1 0.9 0.3 8.7 1.1 tr 10.2 1.8
Average 3.0 3.1 1.1 2.0 11.4 5.0 1.0 1.1 1.2 0.7 5.9 3.6 13.6 13.3
Slag layers 5–6608 2.4 1.0 2.6 1.4 16.0 1.7 0.7 0.1 1.1 0.5 0.5 0.1 2.8 1.0 22.2 5.8
5–6616 2.3 0.1 3.8 0.2 25.7 3.4 0.6 0.4 1.4 0.5 6.2 0.9 0.5 0.1 21.8 5.6
5–6974 4.3 0.5 6.0 1.1 18.9 4.3 1.1 0.3 7.2 0.1 tr 13.7 2.7
5–6815 2.8 0.4 5.0 1.3 36.2 1.8 tr 0.9 0.2 12.0 1.2 tr 13.7 1.5
5–6735 4.2 0.7 4.2 0.5 26.3 3.4 0.5 0.2 10.6 1.4 tr 9.2 0.9
5–7479 2.8 3.0 3.8 0.8 25.0 2.9 tr tr 0.6 0.2 7.1 2.1 6.0 0.8
5–6618
a
1.8 16.9 0.9 0.9 9.4 4.0
Average 2.7 1.3 3.9 1.3 23.6 6.5 0.3 0.4 0.4 0.7 0.4 7.9 2.9 12.9 6.6
ID Ni
2
O
3
SD CuO SD ZnO SD As
2
O
3
SD BaO SD Ag
2
OSD PbO SD Total
Slag fragments 5–8815 41.7 3.2 0.5 0.4 3.2 0.1 73.4
5–6855 30.7 5.4 15.6 5.7 11.5 2.5 3.8 2.1 73.3
5–6916 37.5 9.0 10.6 2.1 11.9 2.0 0.7 0.6 10.5 1.5 82.6
5–6825 0.8 0.1 27.7 4.1 5.6 0.3 7.8 1.8 1.3 0.4 85.4
5–6933 1.9 1.5 34.2 0.9 13.1 6.7 14.4 2.2 6.0 2.1 73.7
5–6606 6.0 1.4 46.2 1.4 5.3 0.6 2.2 0.4 4.3 0.2 102.9
5–6816 12.1 3.4 26.4 3.0 12.3 1.5 13.8 1.5 89.3
5–6912 22.6 3.3 18.1 3.0 21.4 2.5 3.7 0.5 84.9
5–6918 3.0 0.5 30.2 0.2 3.6 0.1 29.4 2.7 88.8
Average 27.0 11.0 15.5 13.2 13.0 7.6 4.8 4.4
Slag layers 5–6608 16.8 5.5 19.5 5.4 6.2 1.4 9.2 1.0 85.6
5–6616 0.5 0.4 7.6 3.0 16.9 2.8 5.6 2.0 7.5 0.7 97.6
5–6974 13.2 1.0 16.9 1.5 13.0 0.7 tr 5.0 0.5 95.0
5–6815 tr 5.3 1.1 18.0 5.7 2.8 0.4 0.5 0.6 1.9 0.2 87.6
5–6735 5.5 3.0 25.5 3.7 4.6 0.3 9.2 0.7 102.2
5–7479 5.3 1.7 36.0 3.2 4.9 1.1 7.8 0.5 92.7
5–6618
a
6.1 35.4 8.4 16.3 80.2
Average 8.6 4.3 24.0 7.9 6.5 3.1 8.1 4.1
a
Only one bulk analysis was performed on this sample due to its heavy corrosion
Archaeol Anthropol Sci

Chemical composition
An approximation to the bulk composition of slags (Table 7)was
obtained by averaging SEM-EDS area analyses of the slag lumps
or crucible slag layers. The results show typically high and var-
iable levels of copper, zinc arsenic and lead, as well as low alkali
contents (max. 1.4 %). The great variability in concentrations of
the main oxides is notable, with wide ranges such as c. 4–36 %
SiO
2
,1–12 % CaO, 0.2–48 % FeO, 5–42 % CuO, 0.5–46 %
ZnO, 3–29 % As
2
O
3
and 0–16 % PbO. This heterogeneity in
bulk compositions both within and between samples must be
related to the complex geology of the region and consequently
the variable crucible charge composition, as well as to the fact
that most of them did not reach a completely liquefied state. In
fact, some of them cannot even be described as a silicate; the
combined weight percent of CuO, ZnO, As
2
O
3
and PbO is
higher than 50 % in most samples. As a general trend, levels of
iron and calcium are inversely correlated, reflecting variable con-
tributions of the either dolomitic or ferrous gangue components.
When compared to the slag layers still adhering to the crucible
sherds, discrete slag lumps are generally richer in copper and
arsenic oxide, probably reflecting a larger abundance of relatively
large unreacted minerals. Slag layers on crucibles are, in turn,
typically richer in silica, alumina and the oxides of zinc and lead.
The enrichment in silica and alumina probably reflects a higher
degree of interaction between the ceramic and the forming slag at
the interface; the relatively constant SiO
2
/Al
2
O
3
ratio in these slag
layers supports the suggestion of a common origin for both
oxides, though it should be noted that this ratio is typically higher
in the slag than in the associated ceramic, and hence, it is likely
that there was an additional contribution of silica in the charge,
probably from quartz gangue or other silicates in the charge. The
often higher levels of zinc and lead in these layers may also result
from the high reactivity between these oxides and the ceramic at
this interface, where they would form stable silicates (cf. Kearns
et al. 2010). Thus, overall, and in spite of these notable compo-
sitional differences, we believe that the slag lumps would have
formed inside crucibles, and that both types of residues derive
from the same processes (see also section ‘Copper prills’).
Bulk copper levels are remarkably high (mean 27 % CuO
in slag lumps) implying high copper losses in the slag
(Fig. 11), typically seen either as abundant metallic copper
prills or as remnants of unreacted copper minerals; dissolved
copper in the slag matrices is much lower (typically 2–3%
CuO; Table 8), reflecting the relatively low reactivity between
copper oxide and silica even at relatively high temperatures
(cf. Kearns et al. 2010). Copper losses in the slag are affected
primarily by both oxidation and viscosity. Too much oxidising
conditions will prevent the reduction of copper to metal, while
facilitating the formation of magnetite or other spinels. Spinels
will in turn increase the slag viscosity, making it difficult for
metallic prills to coalesce together into a larger metal bath. As
shown in the microstructural data presented below, both fac-
tors are at stake in this assemblage.
Fig. 11 Bulk chemical
composition of the slag fragments
(black) and slag layers (red)
shown in the ternary diagram
Cu
2
O FeO+MgO+CaO
SiO
2
+Al
2
O
3
Archaeol Anthropol Sci

Tab l e 8 Matrix slag compositions obtained by SEM-EDS. Average of between 3 and 12 analyses per sample are reported. Areas analysed were selected avoiding any inclusions, crystals or prills. Data are
normalised in wt% with oxygen added by stoichiometry. Tr = Traces (below 0.5%)
ID Na
2
OMgOStD Al
2
O
3
StD SiO
2
StD P
2
O
5
StD SStD K
2
OStD CaO StD TiO
2
StD MnO FeO StD
Slag fragments 5-6606 3.5 1.7 2.3 1.5 25.7 4.1 0.5 0.4 1.0 0.5 16.6 3.7 tr 6.4 2.8
5-6816 16.7 1.5 1.3 0.1 tr 12.8 4.7 tr 4.7 2.1
5-6825 tr 8.7 0.6 8.1 4.2 26.5 7.0 1.7 0.2 24.1 8.0 1.2 0.1 12.9 8.8
Average 4.1 4.4 3.4 4.2 23.0 5.4 1.3 17.8 5.7 8.0 4.4
Slag layers 5-6974 4.5 0.4 3.8 1.5 24.9 2.5 1.3 0.3 10.0 0.8 0.5 0.2 5.9 1.5
5-6618 tr 5.9 0.9 28.5 2.3 0.9 0.6 1.7 0.3 8.8 2.1 tr 5.2 0.5
5-6815 3.0 1.7 6.9 0.7 35.0 2.0 0.7 0.8 tr 3.1 1.6 11.1 1.6 0.8 0.4 17.5 2.1
5-7479 1.6 1.4 6.8 3.4 32.4 5.9 tr 1.4 0.5 11.5 4.6 tr 7.5 0.7
5-6608 tr 2.0 1.0 4.3 1.8 30.5 7.1 tr 2.1 0.8 9.7 3.8 tr tr 5.2 2.7
5-6735 2.3 0.3 3.5 1.0 25.2 4.9 tr 1.4 0.3 9.1 1.3 tr 5.5 1.7
5–6616 tr 2.1 0.5 4.9 1.9 29.7 5.8 tr tr 1.7 0.7 9.6 2.8 7.5 2.3
Average 2.2 1.3 5.1 1.3 29.5 3.4 1.8 0.6 10.0 0.9 0.6 7.8 4.1
ID Ni
2
O
3
StD CuO StD ZnO StD As
2
O
3
StD BaO StD PbO StD Total
Slag fragments 5-6606 1.0 0.6 17.7 5.3 9.9 1.9 6.2 6.3 9.2 3.2 101.0
5-6816 5.3 2.2 16.1 3.1 17.0 1.9 0.8 0.3 24.9 2.4 80.2
5-6825 0.5 0.7 3.1 0.7 4.3 1.0 8.1 2.4 tr 101.1
Average 3.1 2.1 12.7 7.3 11.7 4.7 2.3 3.4 11.5 12.5
Slag layers 5-6974 11.5 3.1 15.1 0.1 16.5 2.9 6.0 0.9 97.5
5-6618 tr 16.9 2.2 7.1 2.2 24.3 2.6 92.3
5-6815 14.7 1.8 2.4 2.2 1.4 1.9 3.1 1.4 98.7
5-7479 4.1 2.3 18.5 4.1 5.6 3.2 9.9 2.4 95.7
5-6608 1.9 0.9 14.6 6.6 6.4 3.7 0.7 1.0 21.9 9.3 101.3
5-6735 5.5 3.8 14.4 2.9 9.7 1.7 1.4 1.4 21.5 6.6 101.3
5–6616 5.3 4.9 15.3 4.4 7.9 1.7 0.5 0.7 14.4 4.8 101.4
Average 4.1 3.7 15.6 1.4 8.0 4.1 0.6 0.6 14.4 7.8
Archaeol Anthropol Sci

The slag matrices are composed mainly by silica-rich
glasses bearing high levels of iron, zinc, arsenic, and lead
oxides. As noted above, copper, zinc, arsenic, and lead oxides
must come from the ore. The greater interaction of the slag
layers with the smelting vessels is evidenced by their higher
amount of silica, alumina and potash in their matrix composi-
tion. On the other hand, higher levels of oxides of calcium,
magnesium and arsenic are documented in the slag fragments.
The latter chemical signatures are mostly related to a higher
ore (and gangue) contribution, although charcoal ashes and a
post-depositional alteration may have been additional sources
of calcium for the slag (see below, section ‘Copper prills’).
Mineralogical examination
The heterogeneity of the slag already indicated by the bulk
chemical analyses becomes even clearer in microstructural
examination. The complete liquefaction of the slag or phase
equilibrium, which would have required higher temperatures
and longer reaction times, only occurred sporadically;
unreacted mineral relicts are identifiable in most specimens.
Slag microstructures contain crystallised and dendritic cuprite,
spinel, Mg-rich willemite and melilite crystals, as well as sul-
phide inclusions, with sporadic occurrence of pyroxene, olivine
or olivenite (Table 9. See Supplementary Material 2for formulas
of mineralogical phases mentioned). Interstitial glass only occurs
in three out of nine slag fragments while it is clearly recognisable
in all crucible slag layers, where the silica absorbed from the
ceramic vessel may have contributed to the glass development.
The absence of glassy matrix and the frequency of mineral relicts
in most of the slag fragments indicate non-equilibrium reactions
in variable redox conditions, with insufficient reaction times.
Sample 5-8815 is the only slag lump dating to the earliest
metallurgical phase at the site, and its microstructure is slightly
different to the rest: it consists predominantly of a fine mixture
of delafossite and magnetite (Fig. 12). Although peculiar in
the absence of silicates, this mineralogy might result from the
smelting of iron-rich copper ores, although it would denote
poor control of atmospheres (Hauptmann 2007, pp. 171–172).
The remaining slag fragments without glassy matrices
(samples 5-6918, 5-6855, 5-6912, 5-6916 and 5-6933), all
dated to the later phase 9, are more clearly connected to
smelting activities, given the presence of residual minerals.
They consist mainly of complex copper ores only partially
smelted, where large willemite inclusions, as well as copper
ore relicts bearing high levels of arsenic and some zinc and
cobalt are clearly visible (Fig. 13a, b, Table 10). Thermal
decay of the dolomitic gangue, which starts at 550 °C, could
have been the cause for the significant remains of Mg-rich
calcite identified in sample 5-6912 (Fig. 13c). Calcite itself
decomposes under oxygen influx at 600 or at 800 °C in a
reducing gas atmosphere (Hauptmann 2007, p. 176). Its pres-
ence is therefore suggesting either low temperatures or short
smelting processes. This dolomitic decomposition contributed
to the formation of tabular crystals of calcium arsenates and
euhedral magnesia silicates embedded in a CuAsZnPb oxide
compound (Fig. 14, Table 11).
Chalcocite inclusions, some of them bearing silver, are fre-
quently documented (Fig. 13d) indicating the presence of a
minor amount of sulphides in the charge, probably as
Tabl e 9 Mineralogical phases identified in slag samples by SEM-EDS. Complex CuAsZnPb, CuAsZn or CuAsZnCo oxides are present in samples
with asterisk
ID CuO ZnO PbO Delafossite Spinel Mg/Fe-rich
Willemite
Pyroxene Melilite Olivine Ca-Arsenates Sulphide Glass
Slag fragments 5-8815 X X X
5-6918* X X
5-6855* X X X X
5-6912* X X X X X
5-6916* X X
5-6933* X X X
5-6606 X X X X X
5-6825 X X X
5-6816 X X X X X X X
Slag layers 5-6608 X X X X X
5-6735 X X X X X
5-7479 X X X X X
5-6815 X X X X X
5-6618 X X X X X
5-6974 X X X X
5-6616 X X X X X X
Archaeol Anthropol Sci

impurities in the predominantly oxidic ores. It is worth noting
that no metallic prills were found in these less reacted samples.
However, the presence of rounded phases dominated by ox-
ides of copper and arsenic, and surrounded by a lead halo,
suggest that metal may have been present but subsequently
corroded (Fig. 15a). In any case, the excess of oxygen in the
smelting gas atmosphere is suggested by the presence of free
cuprite, spinels and other heavy metal oxides, compared to the
near absence of pyroxenes or olivines (Fig. 15, Table 12).
The rest of slags analysed, including all of the slag layers
adhering to crucibles, are more similar to each other and seem
to have been further melted than the ones described above. They
were formed by the crystallisation of almost completely lique-
fied melts under moderately (sometimes weakly) reducing gas
atmospheres. Cuprite is one of the most abundant phases.
Besides its precipitation from copper prills, cuprite also
crystallised from the melt either as a finely dispersed exsolution
in the glassy matrix, which is responsible for the reddish appear-
ance of the slag under the optical microscope, or as a dendritic
intergrowth (Fig. 16a, b). The presence of dendritic growths of
cuprite would suggest that temperatures of c. 1200 °C were
reached (Rovira 2005, p. 91). Likewise, the co-existence of
dendritic cuprite with skeletal magnetite, rhombohedric melilite
and metallic copper is testament to the variable oxidising-
reducing gas atmosphere during the smelting process.
Besides cuprite, spinels and Mg-rich willemite are the most
common oxide phases documented in the better reacted slag
samples. Depending on the iron, zinc, aluminium and magne-
sium content of the ore used, the composition of the spinel
group minerals formed ranges from magnetite to franklinite
,
hercynite or spinels proper (Fig. 16c).
The common presence of rhombohedric crystals of Mg-
rich willemite can likewise be explained by the decomposition
of zincolivenite or zinc silicates—both documented in the ores
(Table 4). These would have decomposed and recrystallised
from the melt-forming rhomboedric siliceous crystals
Fig. 12 SEM-BSE image of slag fragments 5-8815 showing the
formation of delafossite and magnetite under oxidising conditions
Fig. 13 a Complex copper ore
relict in sample 5-6918; b
Willemite relicts in sample 5-
6606; cInclusion of calcite in
sample 5-6912; dExample of
chalcocite relict in sample 5-
6912. See Table 10 for SEM-EDS
analyses
Archaeol Anthropol Sci

enriched on magnesia as well as globular zinc oxides
(Fig. 16d). The clustered occurrence of these Mg-rich willemite
crystals in some samples (Fig. 16e) also suggest that they are
the result of the decomposition of primary zinc silicate relicts.
Melilite and pyroxene are sporadically documented; olivine
occurs only occasionally, as a solid solution between monticellite
and kirschsteinite. The dolomitic gangue documented could have
contributed to the development of tetragonal and thin tabular
melilite crystals and pyroxene, whose compositions range from
hardysonite to Fe-rich diopsides with up to 12–13 % Fe depend-
ing on the Ca/Mg/Zn ratio in the charge. Zinc partially substitutes
magnesia also in the formation of Zn-rich rhombohedric pyrox-
enes with different Zn/Mg ratios (Fig. 16).
This co-existence of metal oxides with silicates is consis-
tent with an excess of heavy metals and variable pO2 atmo-
spheres with oxidising conditions in which magnetite,
delafossite or cuprite could grow but reaching reducing
enough gas atmospheres for silicates to develop and for me-
tallic copper to retain high levels of arsenic, zinc or iron.
Copper prills
The high amount of copper trapped in the slag is present not
only as unreacted minerals and newly formed oxides, but also
as metallic particles; these are usually rounded prills of only a
few micrometers in diameter, but occasionally they reach up
to c. 3 mm (Fig. 17). Due to the heterogeneity of the copper
ores used and variable conditions within the crucibles, prill
compositions are highly variable too, both within and between
samples (Table 13). The only recurring element (except for the
prills in slag sample 5-6912) is arsenic. Other elements such as
S, Fe, Co, Ni, Zn, As, Ag, Sb and Pb range from below de-
tection limits to quite considerable concentrations. Sample 5-
6815 stands out particularly, as it bears copper prills with high
levels of cobalt, nickel, antimony and lead as well as arsenic,
zinc and iron. Due to the large amount of impurities in the
copper prills, some of them exhibit dendritic microst ructures.
This can be clearly seen in Fig. 17b, showing a metal prills
where αgrains (orange) richer in copper (with 2.5 % As)
would have started to crystallise first at c. 1060 °C while the
inter-dendritic compound with up to 25 % As and up to 1.6 %
Pb would have remained molten until temperatures had
lowered down to c. 650–750 °C. Silver and lead segregates
have also been identified in some copper prills.
Within the overall spectrum of variability, there is a notable
trend that differentiates the metal prills in the slag fragments
from those in the slag layers still adhering to crucibles. The
latter tend to be richer in arsenic, iron and zinc, whereas the
former tend to be richer in silver. Rather than interpreting this
as indicative of two different processes, we take it as further
evidence to support the idea proposed above: namely, that the
slag fragments would have formed in the same crucible fur-
naces, but further from the ceramic interface and thus in atmo-
spheres that would have probably been more oxidising. This
Table 10 Composition of mineral relicts in slag samples shown in
Fig. 13 obtained by SEM-EDS. Area analyses as large as possible
trying to avoid large voids. Oxygen added by stoichiometry in 13a–c;
sulphur and chlorine reported as elements. Data in wt%. Analytical totals
of 7a and 7c are very low due to carbon not being measured and high
porosity, these results are therefore not normalised to 100%
ID Fig. MgO Al
2
O
3
SiO
2
SClK
2
O CaO MnO FeO CoO Ni
2
O
3
CuO ZnO As
2
O
3
Total
5-6918 13a 2.9 3.1 23.7 3.4 0.8 0.7 20.8 0.5 5.3 1.6 1.2 22.2 1.1 12.6 66.6
5-6606 13b 26.4 73.6 96.0
5-6912 13c 4.5 0.6 2.5 0.9 45.6 1.1 55.4
ID Fig. S CuO Ag Total
5-6916 13d 33.8 62.7 1.5 102.5
Fig. 14 SEM-BSE image of
sample 5-6912. Note how the
dolomitic decomposition
contributed to the formation of
tabular crystals of calcium
arsenates and euhedral magnesia
silicates embedded in a mixture of
CuAsZnPb oxides. See Table 11
for SEM-EDS results
Archaeol Anthropol Sci

would explain the loss of arsenic, zinc and iron from the metal
in the more oxidised area of the reaction,with the correspond-
ing increase inthe concentration of the more noble copper and
silver. In this sense, it is also significant that while the prills in
the slag fragments are poorer in arsenic, the bulk composition
of these samples shows higher arsenic oxide levels (Table 7),
again showing that arsenic was present in these outer layers,
but in oxide rather than metallic form.
All in all, the high levels of metals with high affinity for
oxygen such as iron, zinc or arsenic indicate that, even if
fluctuating, sufficiently strongly reducing atmospheres were
reached in the crucibles. As for the overall variability in metal
compositions, we need to acknowledge that these cannot be
taken as direct indications of the composition of the metal
being produced (Dungworth 2000); indeed, arsenic levels re-
corded in the crucibles are much higher than in any artefacts
known from the region and period (Rovira et al. 1997).
Arsenic levels would decrease by oxidation and evaporation
upon melting (McKerrell and Tylecote 1972) and much of the
compositional diversity would be erased when the metal was
melted and homogenised. As such, we can propose that the
main product of the site would be arsenical copper, probably
with relatively high silver at the trace level, but it would be
risky to hypothesise further a ‘typical’impurity composition
of the metals produced at Las Pilas.
Melting crucibles
The second type of crucibles is those with a quadrangular
section. One complete specimen and half of a second one were
recovered in addition to several fragments. As described
before, these crucibles have a max. dimension of c.
27 × 11 cm, a height of 4 cm and walls up to c. 2 cm.
Contrary to the thick slag layers described for the smelting
vessels, these crucibles exhibit a thin whitish-yellowish layer
of calcareous appearance that covers their interior, but occasion-
ally appears as exterior patches as well (Fig. 18). In some areas
on the inner part of the complete crucible, a thin greenish slag
layer was also visible; however, its appearance was more rem-
iniscent of corroded metal than of vitreous and viscous slag.
The intentional lining of moulds and crucibles with a variety
of materials can prevent chemical interactions between the ce-
ramic and the liquid metal and facilitate the removal of the
metal from the mould (Zori et al. 2012, cf. Karageorghis and
Kassianidou 1999). Intentional parting layers in crucibles have
been documented in regions such as Egypt, Chile or Argentina.
Glass-making crucibles in Late Bronze Age Egypt were inter-
nally coated with a lime-rich parting layer which formed a
physical barrier and allowed the easy separation of the glass
ingot from the ceramic (Smirniou and Rehren 2016). The lining
of crucibles and casting moulds with bone ash has been docu-
mented at several sites in Chile (Niemeyer et al. 1993;Zorietal.
2012; Plaza and Martinón-Torres 2015) and Argentina
(González 2010; Raffino et al. 1996). However, such crucible
parting layers have not been reported for prehistoric European
metallurgy (but see a possible coated mould in Soares et al.
2008). Thus it was particularly important to determine whether
the calcareous layer in the Las Pilas melting crucibles was an
intentional attribute or the result of post-depositional alteration.
Two samples of these rectangular crucibles were examined
under the SEM-EDS. These internal layers are 0.3–0.5 mm in
thickness (Fig. 19a), very porous, and currently dominated by
Tabl e 1 1 Composition of phases indicated in Fig. 14 obtained by SEM-EDS. Oxygen has been added by stoichiometry. Data are normalised and
analytical totals given. Tr Traces (below 0.5 %)
Spectrum ID Phase MgO SiO
2
K
2
OCaOFeOCuOZnOAs
2
O
3
PbO Total
9c 1 5-6912 Mg silicates 42.5 30.6 1.9 1.1 2.6 12.5 8.9 104.2
9c 2 5-6912 Calcium arsenates 3.4 1.2 37.2 4.4 1.9 47.5 1.5 97.4
9c 3 5-6912 Complex oxide 1.0 tr 4.2 1.7 34.2 12.3 20.4 25.3 97.12
Fig. 15 SEM-BSE image of
samples a5-6933 and b5-6588.
Spinels indicate a rather oxidising
atmosphere. Globular copper and
arsenical copper oxides as well as
copper chloride possibly as a
result of post-depositional
oxidation. Note white halos
surrounding globular copper
oxide in sample 5-6933 aas a
consequence of lead segregation
Archaeol Anthropol Sci

lime, silica, copper oxide and arsenic oxide. These oxides
occur as a fine mixture of two phases—one rich in calcium
and arsenic, the other one in copper and silicon (Fig. 19b). In
sample 5-7166, a small area of melted ceramic was document-
ed under this layer, containing a few corroded prills of arsen-
ical copper entrapped—hence, suggesting that the calcareous
layer post-dates the last high-temperature utilisation
(Fig. 19c). In spite of the enrichment of the calcareous layer
in arsenic and copper, there are no obvious microstructural
features such as neoformed crystals that would confirm
high-temperature reaction between the heavy metals and this
material. As such, while weacknowledge that further analyses
may be needed, we are currently more inclined to consider this
layer as a natural deposition from the calcareous burial envi-
ronment. Arsenic is known to form stable calcium arsenates at
room temperature in oxidising environments (Navarro et al.
2004), and this may well be post-depositional phenomenon
explainingthe current composition of this layer. It might seem
surprising that this layer is not currently present on domestic
pottery from the site, but its absence might simply be a result
of more thorough post-excavation cleaning. Similar layers are
also macroscopically visible in crucibles of the nearby site of
Santa Bárbara as well as in Los Millares; further analyses of
these residues are currently under development.
Turning to the diagnostic metallurgical residues in these
technical ceramics, as noted above these are much less sub-
stantial than in the smelting vessels. Their composition is
much richer in ceramic oxides, with some alkali enrichment
(especially potash) probably derived from fuel ash. The small
but significant presence of oxides of copper, lead and arsenic
is consistent with the melting of the metals smelted on site,
while the absence of zinc and other oxides abundant in ores
and smelting slag argues against the use of these ceramics for
smelting.
While it might be argued that these constitute casting moulds
rather than melting crucibles, their relatively large volume would
argue against this. The volume of the void in the best preserved
one is estimated at c. 380 cm
3
, corresponding with over 3.5 kg of
copper, which is far higher than the weight of the heaviest objects
known in the period. Axes from the contemporary site of Los
Millares weigh 250 g on average and even some of the heaviest
ones, those found in Valencina de la Concepción (Seville) weight
up to 1600g(Lopez Aldana and Pajuelo 2013). Furthermore, the
relatively high fabric vitrification and the enrichment in fuel ash
oxides seem more consistent with their use as melting crucibles,
where lumps of metal would have been mixed with charcoal and
thus necessitating a relatively larger volume.
Metal artefacts
Copper artefacts are remarkably scarce at the site. Besides
some amorphous pieces of corroded metal, only two awls c.
7 and 2.5-cm long, respectively, and one fragment of a possi-
ble burin of quadrangular section were recovered. One of the
awls had some imprints of wood on its surface but it was too
corroded to conduct further analyses, so only one awl and the
burin were sampled.
Bulk composition analyses by SEM-EDS did not detected
any impurity documented in the slag prills, except for arsenic,
quantified as up to 2.5 % (Table 14). However, small inclu-
sions rich in silver as well as antimony and bismuth were
identified, allowing a connection between these artefacts and
the other metallurgical waste.
The high levels of zinc detected in the ores used were almost
completely lost during smelting, with very little ending up in the
metal. Part of the zinc remained in the slag, either unreduced or
reoxidised following reduction, and much was certainly evapo-
rated as it was reduced (boiling point of zinc is 907 °C). Most of
the lead partitioned into the slag compounds, although still pres-
ent in several slag prills. Its absence in the final objects could
derive from its oxidation during melting.
The iron content in copper artefacts is another important tech-
nological parameter, with higher levels expected in copper
smelted in slagging furnaces (Craddock and Meeks 1987).
Although some of the slag-trapped prills in Las Pilas are ferrugi-
nous, iron would be oxidised during re-melting, something also
documented in the nearby site of Almizaraque, with iron levels
up to 5.7 % in some slag prills (Müller et al. 2004, p. 44). The
low levels of iron in the artefacts from Las Pilas are consistent
with the pattern described for the whole of Iberia, with Fe usually
Tabl e 1 2 Composition of phases indicated in Fig. 15 obtained by SEM-EDS with oxygen added by stoichiometry. Data are normalised. Tr Traces
(below 0.5 %)
Spectrum ID Phase MgO SiO
2
P
2
O
5
K
2
O CaO FeO Ni
2
O
3
CuO ZnO As
2
O
3
AgO PbO Total
15a 5-6933 FeZn spinel 2.3 59.4 0.8 1.6 34.3 1.4 2.1 93.1
15b 1 5-6588 Fe-rich willemite 1.8 27.0 5.3 0.5 64.4 99.4
15b 2 5-6588 1.4 0.6 2.5 1.8 16.2 31.3 38.7 1.6 5.9 92.3
15b 3 5-6588 1.3 tr 2.6 3.8 11.8 20.5 25.0 40.0 90.3
15b 4 5-6588 Cuprian adamite 1 3.1 59.5 36.4 89.1
15b 5 5-6588 1.3 tr 0.5 1.2 30.8 30.2 33.8 1.9 88.6
Archaeol Anthropol Sci

less than 0.05 % and only exceptionally reaching up to 0.7 %
(Junghans et al. 1960,1968; Hook et al. 1987; Rovira et al.
1997). Recently, it has been shown that iron levels increase in
copper alloys during the Early Iron Age, which may be related to
technical improvements of higher reducing conditions associated
to furnace smelting, in contrast to previous crucible smelting
(Valério et al. 2015) as already proposed by Craddock and
Meeks (1987).
As a noble metal, silver is more difficult to remove from
copper, and its lower levels in the artefacts could be a conse-
quence of the homogenisation of the metal during melting.
Cobalt, nickel and antimony are only present in one slag
Fig. 16 SEM-EDS images of
most significant phases
documented in slag samples. a
Unstable reducing conditions
evidenced by the precipitation of
metallic copper prills and the
growth of dendritic cuprite in
sample 5-6735. bFinely disperse
exsolution of cuprite and Zn-rich
spinels embedded in the glassy
matrix in sample 5-6974. Note a
silver inclusion in the large
arsenical copper prill. cArsenical
copper prills bearing iron and Zn-
rich spinels as results of redox
conditions in sample 5-6608. d
Globular zinc oxides, Zn-rich
spinels and Mg-rich willemite in a
glassy matrix in sample 5-6606. e
Cluster occurrence of these Mg-
rich willemite crystals in sample
5-6816. fPyroxene and zinc ox-
ide in a glassy matrix in sample
6816. Note clustered dendritic
growth of calcium arsenates. g
Zn-rich rhombohedric pyroxene
and Mg-rich Willemite in glassy
matrix in sample 5-6735. h
Acicular delafossite and crystals
of Zn-rich delafossite in sample 5-
6616
Archaeol Anthropol Sci

sample, implying that ores containing these elements were
only sporadically. They are detected in low quantities in the
final objects, although inclusions rich in antimony were iden-
tified under the SEM.
Arsenic losses will be affected primarily by the conditions of
the smelting and melting (the evaporation rate of arsenic being
higher under oxidising conditions), the length and intensity of
the working techniques when shaping the objects, and the pos-
sibility of recycling. In the case of Las Pilas, high levels of
arsenic are documented in the ores, with a Cu/As ratios up to
1.4:1 (Table 15). These levels of arsenic are still documented in
the copper prills of the smelting slags (Cu/As ratio up to 1.8:1).
Although the average amount of arsenic in copper prills is lower
than in ore samples (see Table 15), the evaporation rate of
arsenic during smelting seems to be low. The main losses of
arsenic are more likely to occur during melting and casting,
when metal is exposed to high temperatures and more oxidising
atmospheres (see McKerrell and Tylecote 1972 for
experimental rates of arsenic losses) leading to low arsenic
levels in the final objects. Although starting from high arsenic
ores, the two objects found in Las Pilas only have 1 % arsenic
on average, lower than the levels observed in other sites in the
area such as Almizaraque.
Discussion and conclusions
Recent studies of early metallurgy have focused mostly on broad
models for the emergence and spread of metallurgy in the Old
World, or on long distance contacts and exchange. These studies
rarely focus these questions on the local scale. However, detailed
contextual studies are essential to understand the development of
Fig. 17 a SEM-BSE image of a large copper prill in sample 5-6735. b
Detail of the same prill under the optical microscope. Unetched sample.
Note high porosity (black holes) and small round segregates of lead
(indicated with white arrows). The dendritic microstructure shows α
grains (orange) with 2.5 % As growing from the inter-dendritic
compound with up to 25 % As and up to 1.6 % Pb. General composition
results in area analysis are 8.9 % As and 0.6 % Pb
Tabl e 1 3 Compositions of copper prills in slag samples obtained by SEM-EDS. Average analyses of between 4 and 20 prills per sample are reported.
Data are normalised in wt% as elements. Analytical totals given. Tr = Traces (below 0.5%)
ID O SD SFe SD Co SD Ni SD Cu SD Zn SD As SD Ag SD Sb SD Pb SD Total
Slag fragments 5-6606 0.5 0.3 1.4 1.1 80.0 10.1 tr 17.4 9.5 tr tr 104.1
5-6816 tr tr 97.4 7.4 1.0 0.3 0.6 0.5 0.5 0.5 92.3
5-6912 0.6 0.1 98.2 1.2 1.2 0.2 106.3
5-6825 0.7 0.0 1.3 0.3 93.7 0.6 4.3 0.2 102.7
Average 0.5 0.1 0.7 0.8 93.5 9.0 5.7 8.0 0.5 0.5 0.2 0.3
Slag layers 5-6974 0.6 0.2 0.9 0.5 95.5 3.6 2.7 1.8 tr 97.6
5-6618 1.0 1.3 tr 69.7 1.2 0.8 1.1 28.3 5.5 96.1
5-7479 1.1 0.6 tr tr tr 81.7 6.0 tr 15.4 5.2 tr 1.1 1.2 97.4
5-6815 0.8 0.2 tr 1.7 1.0 4.3 3.0 6.2 4.9 52.9 21.2 0.5 0.8 28.2 7.7 0.7 0.0 3.9 2.8 99.6
5-6608 tr 1.4 1.1 tr 83.3 8.7 tr 14.8 7.0 105.6
5-6735 0.9 0.2 0.5 0.5 87.3 18.5 0.5 0.7 10.1 10.9 0.5 1.1 tr 96.4
5-6616 1.1 0.5 2.3 1.0 95.7 0.4 tr tr tr 102.3
Average 0.8 0.3 1.1 0.8 0.9 2.3 80.9 15.2 tr 14.2 11.1 tr 0.8 1.4
Archaeol Anthropol Sci

early metallurgy in its sociotechnical context, and to begin to
characterise traditions whose evolution and possible interaction
may be mapped in space and time.
The technology of arsenical copper smelting and melting
in Las Pilas.
Las Pilas provides evidence of in situ extractive metallurgy
spanning several centuries since the beginnings of the third
millennium BC. Complex oxide ores containing mineral mix-
tures of copper, arsenic, and other heavy elements were
smelted with charcoal in large, flat, circular crucible-furnaces,
with the air supplied by blowpipes made of reed tubes
protected with a clay nozzle. The resulting metal, typically
arsenical copper, was subsequently melted and refined in rect-
angular melting crucibles which, like the smelting crucibles,
were heated from above. The fabric of the ceramics employed
for technical purposes is virtually identical to coarse ceramics
employed locally for domestic use, and the technique of
moulding pottery by means of an esparto basket has been
documented for the first time in crucible manufacturing.
This use of large, flat crucible-furnaces and blowpipe nozzles
would differ from early near eastern and eastern European
metallurgical traditions, where no crucibles have been recov-
ered (e.g. in Serbia Radivojevićet al. 2010), smelting was
achieved in simple holes in the ground (e.g. in the Levant
Golden et al. 2001) or hearth installations are lined with bro-
ken pottery (e.g. in Bulgaria Rehren et al. 2016)
Metallurgy at Las Pilas is characterised by its small scale.
Productivity, in terms of optimisation of metal extraction, does
not seem to have been a major concern.
3
Ores were
transported to the site with the gangue and crushed in situ
(as indicated by the ore samples recovered and the two grind-
ing stones found at the site), and simple but less efficient
technologies were chosen: blowpipes instead of bellows and
open air structures instead of closed furnaces. Slags and par-
tially smelted ores were not re-smelted to recover the high
amounts of copper still trapped in them, nor even when the
smelting process failed before had it been finished. The abun-
dance of copper resources in the area may have been one of
the reasons why efficiency was not a constraint.
In terms of the technological proficiency, control over at-
mospheres and other smelting parameters were not mastered.
Most of the slags were not completely liquefied, and they are
relatively poor in silica and other light oxides that would have
facilitated the production of a homogeneous slag layer sepa-
rate from the metal. Non-equilibrium reactions were devel-
oped in variable redox conditions, sometimes with too
oxidising atmospheres and with insufficient reaction times.
The low refractoriness of the crucibles may have constrained
the smelting times. Notwithstanding its technical inefficiency,
this metallurgy was sufficiently cost-effective to meet the
needs of these societies, and metal was undoubtedly obtained.
In this aspect, the evidence from Las Pilas is similar to that
described for the nearby site of Almizaraque (Müller et al.
2004) although some of the metallurgical features of the work-
shop of Las Pilas (i.e. blowpipe nozzles or crucibles with
basketry imprints) have not been documented at Almizaraque.
There is a recurrent use of complex copper minerals bear-
ing high levels of arsenic,zinc and lead. While some of the ore
samples recovered are relatively pure malachite, most of the
minerals and slag samples bear high concentrations of other
elements. It has recently been proposed for the Chalcolithic
Balkans that early smelters carefully selected black-and-green,
manganese-rich malachite for smelting, while green pure mal-
achite was set aside for lapidary work (Radivojevićand
Rehren 2015). In the case of Las Pilas, however, there is no
indication that specific mineral types were selected for. While
green and blueare the predominant colours overall, it does not
seem likely that metallurgists would have had the ability (or,
indeed, the desire) to distinguish between pure copper carbon-
ates and those containing copper and cobalt arsenates, zinc
carbonates or lead and copper sulphides—as typical of the
local geology.
The complexity and variability of the ore charges and
smelting conditions was reflected in the composition of the
prills trapped in the slag. If we were to classify these prills
3
We acknowledge that one of the authors, FMG, isnot in agreement with the
interpretation of Southeastern Chalcolithic metallurgy as a low-efficiency tech-
nology with a low scale of production and a limited degree of specialisation.
Fig. 18 Sample of a melting crucible. Note the inner whitish-yellowish
layer
Archaeol Anthropol Sci

qualitatively based on the presence/of absence of Ni, As, Ag
and Sb, as proposed in recent approaches to legacy copper-
alloy data (e.g. Bray and Pollard 2012; Pollard et al. 2015), we
would find five different metal types at a single production
site. This variability would be further increased if we consid-
ered the presence/absence of additional elements such as S,
Fe, Co, Zn and Pb. This variability would be partly erased
during metal re-melting, so that the final objects would only
bear copper and arsenic in significant concentrations—but it is
still worth noting that the two objects analysed from the site
would still represent two different metal groups as per the
above classification, while four groups could be identified in
other Chalcolithic sites in the province such as El Malagón
and Los Millares (Hook et al. 1991) and three groups in the
slag prills of Almizaraque (Müller et al. 2004).
While the objects from Las Pilas are bothrelatively poor in
arsenic in spite of the arsenic-rich nature of the ores, the het-
erogeneity in the arsenic compositions becomes more promi-
nent if we consider the composition of other Chalcolithic ob-
jects recovered in the area. The average arsenic at El Malagón
was quantified as 1.67 % by AAS (with a maximum value of
3.31 %), and this value is 2.25 % (max. 6.04 %) at Los
Millares (Hook et al. 1989, p. 73). Comparable data was ob-
tained by XRF by the Project Archaeometallurgy of the
Iberian Peninsula for artefacts from Los Millares (mean
2.2 % As, max. 11.3 %) (Montero Ruiz 1994; Rovira et al.
1997). The same project reported a mean 3.6 % As (max.
11.2 %) at Almizaraque (max. 11.2 %) and La Encantada
(max. 7.0 %) down to 0.9 % at El Barranquete (max. 2.0 %)
(Montero Ruiz 1994;Roviraetal.1997). In Loma de
Belmonte, a necropolis close to the site of Las Pilas and prob-
ably associated to it, levels of arsenic are 2.3 % on average
(max. 3.5 %) (Montero Ruiz 1994;Roviraetal.1997). On
balance, the variability and the higher arsenic content in ob-
jects from Almería province when compared to other areas
seems to be a reflection of the local geology and its complex
mineralogy, much more than of a metallurgical awareness of
the properties of different alloys (Delibes de Castro et al.
1989;Rovira2004).
The organisation of copper production in its social context.
Beyond the more strictly technical aspects, we should address
how copper production was organised at the site and within its
specific social and cultural context. The earliest evidence of high-
temperature metallurgy at the site is dated to 2905-2743 2σcal.
Fig. 19 a SEM-BSE image of the inner layer of sample 5-6727. bDetail
of the slag layer with the calcium arsenate compound (whitish) and the
copper silicate (greyish). cMelted area of sample 5-7166. All copper
prills are heavily weathered
Tabl e 1 4 Composition of metal
objects obtained by SEM-EDS.
Average of between four and
seven area analyses per sample
are reported. Data are normalised
in wt% as elements
Sample ID Technique Cu
%
As
%
Total Sb
inclusions
Ag
inclusions
Bi
inclusions
Copper lump 5-8862 SEM-EDS 97.7 2.3 97.2 X X X
Copper burin 5-2997 SEM-EDS 98.9 1.1 101.4 X X X
Copper Awl 5-11454 SEM-EDS 98.9 1.0 103.1 X X
Copper Average 98.5 1.5
Archaeol Anthropol Sci

BC. Metallurgical activity continued at the site during subsequent
phases, although evidence is scarce (of course, based on the
relatively small site area excavated compared to the extended
surface estimated for the site). The main documented metallur-
gical evidence corresponds to the phase dated to 2578-2276 2σ
cal. BC. In this period, metallurgical activity is established in a
communal area where other daily activities related to a central
fire place took place. A semi-circular structure made of clay
bricks was erected next to the central fire place as a way to
demarcate the smelting area. The highly vitrified and burnt clay
of the structure where some slag, and charcoal fragments were
found, as well as the associated blowpipes and crucible found in
situ support the interpretation of this as the smelting location.
This productive setting, inside the settlement in a productive
area in which other activities were simultaneously carried out,
and the relatively inefficient technology used (blowpipes and
crucible smelting), together suggest a low degree of specialisa-
tion and scale of production. As also proposed by experimental
studies (e.g. Hanning et al. 2010), part-time metalworkers may
have been able to produce copper with this rather simple tech-
nology, similar to the one described in the nearby site of
Almizaraque (Delibes de Castro et al. 1989,1991;Mülleretal.
2004) or in the southwestern site of Zambujal (Müller et al. 2007;
Gauss 2015). In this aspect, the evidence found at Las Pilas
would be coherent with the ‘household production’model pro-
posedbyStrahmandHauptmann(2009)intheir‘Innovation
Phase’, technologically characterised by smelting in simple cru-
cibles which do not seem associated to social hierarchisation. It
should be noted, however, that, their scheme does not seem fully
applicable to the development of metallurgy in Iberia, as recently
discussed by Rovira and Montero Ruiz (2013), since the ‘Initial
Phase’of native copper exploitation has not been documented so
far, and the ‘Consolidation Phase’is not recognised until the Iron
Age.
The context of metal production shows that metal-
workers were integrated with the rest of the communal
activities and crafts. There is no indication that the com-
munication of metallurgical knowledge had to rely on
strong leadership and/or political control to be efficient
(see also Kienlin 2016); nor that metallurgical knowledge
was secret, as proposed for instance in the Southeastern Alps
where smelting and funerary practises took place under rock
shelters and in ritual contexts away from domestic villages
(Dolfini 2014, p. 483; cf. Budd and Taylor 1995). At Las
Pilas, metallurgy is developed inside the village in an area
destined to pyrotechnical activities. In this sector, two com-
bustion structures were documented adjoining the wall which
bounds the site, in the most external area in relation to the
dwellings.The bigger one dedicated mainly to cereals roasting
(mostly T. aestivum durum), and the second one, made with a
plinth of adobe on which the blowing pipes sit, exclusively
devoted to metallurgy. Nothing suggests that metalworkers
were in any way detached from the community: metallurgy
is developed in a productive area in close interaction with
other crafts.
Furthermore, this productive context together with the neces-
sity of ore procurement, convey a picture of communal and
collaborative work in which the boundaries between different
crafts are faint. In fact, craftspeople may have faced common
technical problems, such as the control of heat and social rela-
tionships may have allowed the transfer of skills and techniques
from one production system to another (e.g. Sofaer 2006). In the
case of Las Pilas, this is exemplified by the manufacture of some
crucibles which embody basketry, pottery and metallurgy—thus
materialising social communication of knowledge and skills in
contexts where cooperation is necessary.
Our work has provided an example of high-resolution char-
acterisation of the engineering parameters, scale, and efficiency
of early metallurgy in Southeast Iberia, with inferences about
context and craft organisation derived from the archaeological
excavation. It is hoped that a proliferation of studies of this kind
may facilitate more nuanced comparisons that may allow for a
better grounded discussion of the possible existence, nature and
direction of lines of knowledge transmission, both within specific
regions and across larger spaces.
Tabl e 1 5 Cu/As proportions in ores, prills and objects re-normalised to
100%
Sample ID Technique Cu% As % Ratio
Ore 5-6726 ICP-MS 58.55 41.44 1.4/1
Ore 5-6491 ICP-MS 78.34 21.65 3.6/1
Ore 5-6740 ICP-MS 99.87 0.16
Ore 5-7706 ICP-MS 74.97 25.02 3/1
Ore 5-6972 ICP-MS 77.73 22.26 3.5/1
Ore 5-8479 ICP-MS 98.29 1.73 56.8/1
Ore Average 81.3 18.7 4.3/1
Prills 5-6606 SEM-EDS 82.1 17.8 4.6/1
Prills 5-6816 SEM-EDS 99.0 1.0
Prills 5-6912 SEM-EDS 100 nd
Prills 5-6825 SEM-EDS 95.6 4.4 15/1
Prills 5-6974 SEM-EDS 97.2 2.7 36/1
Prills 5-6618 SEM-EDS 71.1 28.9 2.4/1
Prills 5-7479 SEM-EDS 84.1 15.8 5.3/1
Prills 5-6815 SEM-EDS 65.2 34.7 1.8/1
Prills 5-6608 SEM-EDS 84.9 15.0 5.6/1
Prills 5-6735 SEM-EDS 89.6 10.4 8.6/1
Prills 5-6616 SEM-EDS 100 tr
Prills average 88.0 11.9 7.3/1
Copper lump 5-8862 SEM-EDS 97.7 2.3 42/1
Copper awl 5-2997 SEM-EDS 98.9 1.1 90/1
Copper awl 5-11454 SEM-EDS 98.9 1.1 90/1
Copper Average 98.5 1.5 66/1
Archaeol Anthropol Sci

Acknowledgements This research was supported by a Marie Curie
Intra European Fellowship within the 7th European Community
Framework Programme (‘Society, Metallurgy and Innovation: The
Iberian Hypothesis’—SMITH project, PN623183); by the the R&D
Projects HAR2011-29068 and HAR2012-38857 funded by the Spanish
Ministry of Economy and Competitiveness as well as by the Culture
Office of the Government of Andalucía (Spain).
References
Alcaide G, Molist M, Montero I, Planagumà L, Saña M, Toledo A (1998)
Producciones Metalúrgicas en el Nordeste de la Península Ibérica
durante el III Milenio Cal. AC: El Taller de la Bauma del Serrat del
Pont (Tortellà, Girona). Trab Prehist 55:81–100
Alcaraz Hernández FM (1992) Excavación arqueológica de emergencia
en las Pilas-Huerta Seca (Mojácar, Almería). Anuario Arqueológico
de Andalucía 1990:18–24
Alfaro Giner C (1984) Tejido y cestería en la península Ibérica. Historia
de su técnica e industrias desde la prehistoria hasta la romanización.
Bibliotheca Praehistorica Hispana, XXI. Madrid: CSIC
Aranda G, Alarcón E, Murillo-Barroso M, Montero I, Jiménez-Brobeil S,
Sánchez Romero M, Rodríguez-Ariza MO (2012) El yacimiento
argárico del Cerro de San Cristóbal (Ogíjares, Granada). Menga 3:
141–164
Bachmann HG (1982) The identification of slags from archaeological
sites. Institute of Archaeology Occasional Publication No. 6,
London
Bartelheim M (2007) Die Rolle der Metallurgie in vorgeschichtlichen
Gesellschaften: Sozioökonomische und kulturhistorische Aspekte
der Ressourcennutzung. Ein Vergleich zwischen Andalusien,
Zypern und dem Nordalpenraum. Forschungen zur Archäometrie
und Altertumswissenschaft 2. Rahden
Bartelheim M, Contreras F, Moreno A, Murillo-Barroso M, Pernicka E
(2012) The silver of the south Iberian el Argar culture: a first look at
production and distribution. Trab Prehist 69(2):293–309
Bayley J, Rehren T (2007) Towards a functional classification of cruci-
bles. In: La Niece S, Hook D, Craddock P (eds) Metals and mines.
Studies in Archaeometallurgy. The British Museum, Archetype,
London, pp 46–55
Bertran-Oller O, Buireu-Cabello F, Febrer-Morlà M, Enrique-Gisbert P,
Melgarejo JC (2012) Mineralogía del Depósito de Cuesta Alta,
Cerro Minado, Huércal-Overa, Almería. Macla, Revista de la
Sociedad Española de Mineralogía 16:246–247
Bray PJ, Pollard AM (2012) A new interpretative approach to the chem-
istry of copper-alloy objects: source, recycling and technology.
Antiquity 86(333):853–867
Budd P, Taylor T (1995) The faerie smith meets the bronze industry:
magic versus science in the interpretation of prehistoric metal-mak-
ing. World Archaeol 27(1):133–143
Camalich Massieu MD, Martín Socas D. (eds.) (1999) El territorio
almeriense desde los inicios de la producción hasta fines de la
Antigüedad. Un modelo: la depresión de Vera y cuenca del río
Almanzora.Sevilla: Conserjería de Cultura, Junta de Andalucía
Castro P, Lull V, Micó R (1996) Cronología de la prehistoria reciente de la
Península Ibérica y Baleares (c. 2800 900 cal ANE). British
Archaeological Reports. International Series, 652. Tempus
Reparatum. Oxford: Archaeopress
Craddock PT, Meeks D (1987) Iron in ancient copper. Archaeometry
29(2):187–204
Cruz Berrocal M, Cerrillo Cuenca E, García Solano JA (2006) Nuevos
datos sobre el Calcolítico de Extremadura: El yacimiento de La
Sierrecilla (Santa Amalia, Badajoz). Spal 15:51–70
Delgado Raak S, Escanilla N, Risch R (2014) Mazas ocultas. Rastros de
minería prehistórica en el Cerro Minado de Huercal-Overa
(Almería). Cuadernos de Prehistoria y Arqueología de la
Universidad de Granada 24:13–44
Delibes de Castro G, Fernández-Miranda M, Fernández-Posse MD,
Martín C, Montero I, Rovira S (1991) Almizaraque (Almería,
Spain). Archaeometallurgy during the chalcolithic in the south-east
of the Iberian peninsula. In: Éluère C, Mohen JP (eds) Découverte
du Métal. Picard, Paris, pp 303–315
Delibes de Castro G, Viñé A, Salvador M (1998) Santioste, una factoría
salinera en el inicio de la Edad del Bronce en Otero de Sariegos
(Zamora). In: G. Delibes de Castro (ed.): Minerales y metales en la
prehistoria reciente. Algunos testimonios de su explotación y
laboreo: 155–197. Studia Archaeologica, 88. Valladolid:
Universidad de Valladolid
Delibes de Castro G, Fernández-Miranda M, Fernández-Posse MD,
Martin C, Rovira S, Sanz M (1989) Almizarque (Almería):
minería y metalurgia calcolíticas en el sureste de la Peninsula
Ibérica. In: Domergue C (ed) Minería y Metalurgia de las
Antiguas Civilizaciones Mediterráneas y Europeas. Ministerio de
Cultura, Madrid, pp 81–96
Dolfini A (2014) Early metallurgy in the Central Mediterranean. In:
Roberts BW, Thornton CP (eds) Archaeometallurgy in global per-
spective. Springer, New York, pp 473–506
Domergue C (1987) Catálogo de minas y fundiciones antiguas de la
Península Ibérica. Casa de Velázquez, Madrid
Dungworth D (2000) A note on the analysis of crucibles and moulds. Hist
Metall 34(2):83–86
Favreau G, Eytier C, Eytier JR, Escanilla N (2013) Les mines de Cerro
Minado, Huércal-Overa (Almería, Espagne). Le Cahier des
Micromonteurs 121 (3)
Gauss R (2015) Zambujal und die Anfänge der Metallurgie in der
Estremadura (Portugal). Technologie der Kupfergewinnung,
Herkunft des Metalls und soziokulturelle Bedeutung. Früher
Bergbau und Metallurgie auf der Iberischen Halbinsel Faszikel 1.
Iberia Archaeologica, 15. Tübingen: Wasmuth
Golden J, Levy T, Hauptmann A (2001) Recent discoveries concerning
ancient metallurgy at the chalcolithic (ca. 4000 BC) village of
Shiqmim, Israel. J Archaeol Sci 9:951–963
Gómez Ramos P (1996) ¿Dónde están las toberas de la Edad del Bronce
en la Península Ibérica? Acontia Revista de Arqueología 2:29–38
González L (2010) Fuegos sagrados. El taller metalúrgico del Sitio 15 de
Rincón Chico (Catamarca, Argentina). Boletín del Museo Chilenos
de Arte Precolombino 15(1):47–62
Hanning E, Gauß R, Goldenberg G (2010) Metal for Zambujal: experi-
mentally reconstructing a 5000-year-old technology. Trab Prehist
67(2):287–304
Hauptmann A (2007) The Archaeometallurgy of copper. Evidence from
Faynan, Jordan. Springer, Berlin
Hook DR, Arribas A, Craddock PT, Molina F, Rothenberg B (1987)
Copper and silver in bronze age Spain. In: Waldren WH, Kennard
RC (eds) Bell beakers of the Western Mediterranean. British
Archaeological Reports. International Series 331. Archaeopress,
Oxford, p 147–172
Hook, D.R; Freestone, I.C.; Meeks, N.D.; Craddock, P.T.; Moreno, A.
(1991): The early production of copper alloys in south East Spain.
In: E. Pernicka; G.A. Wagner (eds.): Archaeometry ‘90: 65 76.
Basel: Birkhauser Verlag.
Hunt MA (2003)Prehistoric mining and metallurgy in south west Iberian
peninsula. BAR, International Series 1188. Oxford: Archaeopress
Jalhay E, Paço A (1945) El Castro de Vilanova de San Pedro. Actas y
Memorias de la Sociedad Española de Antropología, Etnografía y
Prehistoria 20. Madrid
Junghans S, Sangmeister E, Schroder M (1960) Metallanalysen
Kupfer.zeitlicher und fruhbronzezeitlicher. Bodenfunde aus
Archaeol Anthropol Sci

Europe.Studienzu den Anfangeder Metallurgie 1. Berlin: Romisch-
Germanisches Zentralmuseum
Junghans S, Sangmeister E, Schroder M (1968) Kupfer u. Bronze in der
fruhen Metallzeit Europas. Studien zu den Anfange der Metallurgie.
2. Berlin: Romisch-Germanisches Zentralmuseum
Höppner B, Bartelheim M, Huijsmans M, Krauss R, Martinek K-P,
Pernicka E, Schwab R (2005) Prehistoric copper production in the
inn Val-ley (Austria) and the earliest copper in Central Europe.
Archaeometry 47:293–315
Karageorghis V, Kassianidou V (1999) Metalworking and recycling in
late bronze age Cyprus. The evidence from Kition. Oxf J Archaeol
18(2):171–188
Kearns T, Martinón-Torres M, Rehren T (2010) Metal to mould: alloy
identification in experimental casting moulds using XRF. Hist
Metall 44(1):48–58
Kienlin, T.L. (2016): Some thoughts on evolutionist notions in the study
of early metallurgy. In: M. Bartelheim; B. Horejs; R. Krauß (eds.):
Von Baden bis Troia. Ressourcennutzung, Metallurgie und
Wissenstransfer. Eine Jubiläumsschrift für Ernst Pernicka: 123–
137. Rahden/Westfalia: Verlag Marie Leidorf.
Kienlin TL (2010) Traditions and transformations: approaches to
Eneolithic (Copper Age) and Bronze Age metalworking and
Society in Eastern Central Europe and the Carpathian Basin. BAR
International Series 2184. Oxford: Archaeopress
Kilikoglou V, Vekkinis G, Maniatis Y, Day PM (1998) Mechanical per-
formance of quartz tempered ceramics: part 1, strength and tough-
ness. Archaeometry 40:261–279
Lopez Aldana P, Pajuelo A (2013) La secuencia ocupacional durante el III
Milenio a.n.e. en C/Trabajadores 14–18 (Valencina de la Concepción,
Sevilla). In: García Sanjuán L, Vargas M, Hurtado V, Ruiz T, Cruz-
Auñones R (eds) El asentamiento prehistórico de Valencina de la
Concepción (Sevilla). Universidad de Sevilla, Sevilla
Martínez Frías J, García Guinea J, López Ruiz J, López GarcíaJA, Benito
García R (1989) Las mineralizaciones epitermales de Sierra
Almagrera y de la Cuenca de Herrerías. Cordilleras Bética, Boletín
de la Sociedad Española de Mineralogía 12:261–271
Martinón-Torres M, Rehren T (2009) Post-medieval crucible production
and distribution: a study of materials and materialities.
Archaeometry 51(1):49–74
Martinón-Torres M, Rehren T (2014) Technical ceramics. In: Roberts
BW, Thornton CP (eds) Archaeometallurgy in global perspective.
Springer, New York, pp 107–131
Martinón-Torres M, Uribe-Villegas MA (2015) Technology and culture
in the invention of lost-wax casting in South America: an
archaeometric and ethnoarchaeological perspective. Camb
Archaeol J 25(1):377–390
Martinón-Torres M, Rehren T, Freestone IC (2006) Mullite and the mys-
tery of the Hessian wares. Nature 444:437–438
Martinón-Torres M, Freestone IC, Hunt A, Rehren T (2008) Mass-
produced mullite crucibles in medieval Europe: manufacture and
material properties. J Am Ceram Soc 91(6):2071–2074
McKerrell H, Tylecote RF (1972) The working of copper-arsenic alloys
in the early bronze age and the effect on the determination of prov-
enance. Proceedings of the Prehistoric Society 38:209–218
Molina F, Cámara JA, Capel J, Nájera T, Sáez L (2004) Los
Millares y la periodización de la Prehistoria Reciente del
Sudeste. II-III Simposio de Prehistoria Cueva de Nerja:142–
158. Nerja: Fundación Cueva de Nerja
Montero Ruiz I (1994) El origen de la metalurgia en el sudeste de la
Península Ibérica. Instituto de Estudios Almerienses, Almeria
Müller R, Rehren T, Rovira S (2004) Almizaraque and the early copper
metallurgy of Southeast Spain: new data. Madrider Mitteilungen 49:
94–114
Müller R, Goldenberg G, Bartelheim M, Kunst M, Pernicka E (2007)
Zambujal and the beginnings of metallurgy in southern Portugal.
In: La Niece S, Hook D, Craddock P (eds) Metals and mines.
Studies in Archaeometallurgy. Archetype; The British Museum,
London, pp 15–26
Murillo-Barroso M, Montero Ruiz I, Bartelheim M (2014) Native silver
resources in Iberia. In: Meller HH, Risch R, Pernicka E (eds) Metals
of power—early gold and silver. Landesmuseum für Vorgeschichte,
Halle, pp 257–267
Navarro P, Vargas C, Araya E, Martín I, Alguacil FJ (2004) Precipitación
de arsénico desde efluentes metalúrgicos. Rev Metal 40:409–412
Niemeyer FH, Castillo G, Cervellino M (1993) Estrategia del Dominio
Inca en el Valle de Copiapó. In: Actas del XII Congreso Nacional de
Arqueología Chilena, vol. 1, p 333–371
Nocete F. (ed.) (2004) ODIEL: Proyecto de Investigación Arqueológica
para el análisis de la desigualdad social en el Suroeste de la
Península Ibérica. Sevilla: Monografías de Arqueología, 19.
Consejería de Cultura, Junta de Andalucía
Nocete F (2006) The first specialised copper industry in the Iberian pen-
insula: Cabezo Juré (2900-2200 BC). Antiquity 80:646–657
Nocete F, Lizcano R, Nieto JM, Alex E, Inacio N, Bayona MR, Delgado
A, Orihuela A, Linares JM (2004) La ordenación espacio-temporal
de Cabezo Jure. In: Nocete, F. (ed.): ODIEL: Proyecto de
Investigación Arqueológica para el análisis de la desigualdad social
en el Suroeste de la Península Ibérica: 129–239. Monografías de
Arqueología, 19. Sevilla:. Consejería de Cultura, Junta de Andalucía
Nocete F, Queipo G, Sáez R, Nieto JM, Inácio N, Bayona MR,Peramo A,
Vargas JM, Cruz-Auñón R, Gil-Ibarguchi JI, Santos JF (2008) The
smelting quarter of Valencina de la Concepción (Seville, Spain): the
specialised copper industry in a political Centre of the Guadalquivir
Valley during the third millennium BC (2750-2500 BC). J Archaeol
Sci 35:717–732
Obón Zúñiga A, Berdejo Arcéiz A (2013) Estudio Experimental de los
Orígenes de la Metalurgia del Cobre en el Sur de Iberia: La Vasija
Horno. In: Palomo A, Piqué RTerradas X (eds)Experimentación en
Arqueología. Estudio y Difusión del Pasado. Serie Monográfica del
MAC-Girona 25(2):407–415
Plaza MT, Martinón-Torres M (2015) Metallurgical traditions under Inka
rule: a technological study of metals and technical ceramics from the
Aconcagua Valley, Central Chile. J Archaeol Sci 54:86–98
Pollard AM, Bray P, Gosden C, Wilson A, Hamerow H (2015)
Characterising copper-based metals in Britain in the first millennium
AD: a preliminary quantification of metal flow and recycling.
Antiquity 89(345):697–713
RadivojevićM, Rehren T, Pernicka E, Šljivar D, Brauns M, BorićD
(2010) On the origins of extractive metallurgy: new evidence from
Europe. J Archaeol Sci 37(11):2775–2787
RadivojevićM, Rehren T (2015) Paint it black: the rise of metallurgy in
the Balkans. J Archaeol Method Theory 23(1):200–237
Raffino R, Iturriza R, Iácona A, Capparelli A, Gobbo D, Montes V,
Vázquez R (1996) Quillay: centro metalúrgico Inka en el Noroeste
Argentino. Tawantinsuyu 2:59–68
Rehder JE (1994) Blowpipes versus bellows in ancient metallurgy.
Journal of Field Archaeology 21(3):345–350
Rehren T, Leshtakov P, Penkova P (2016) Reconstructing chalcolithic
coppe smelting at Akladi cheiri, Chernomorets, Bulgaria. In:
Nikolov V, Schier W (eds) Der Schwarzmeerraum vom
Neolithikum bis in die Früheisenzeit (6000–600 v.Chr.). Verlag
Marie Leidorf GmbH, Rahden/ Westfalia, pp 205–214
Renfrew C (1969) The autonomy of the south-east European copper age.
Proceedings of the Prehistoric Society 35:12–47
Rovira Buendía N (2007) Agricultura y gestión de los recursos vegetales
en el sureste de la península ibérica durante la prehistoria reciente.
Unpublished PhD. Barcelona: Pompeu Fabra University
Rovira S (2002) Early slags and smelting by-products of cop-per metal-
lurgy in Spain. In: Bartelheim M, Pernicka E, Krause R (eds) The
beginnings of metallurgy in the old world. Verlag Marie Leidorf
GmbH, Rahden/ Westfalia, pp 83–98
Archaeol Anthropol Sci

Rovira S (2004) Tecnología metalúrgica y cambio cultural en la Prehistoria
de la Península Ibérica. Norba. Revista de Historia 17:9–40
Rovira S (2005) Metalurgia de crisol: La obtención de cobre en la
Prehistoria de la Península Ibérica. De Re Metallica 5:87–94
Rovira S, Montero Ruiz I (2013) Iberia: technological development of
prehistoric metallurgy. In: Burmeister S, Hansen S, Kunst M,
Müller-Scheeßel M (eds) Metal matters. Innovative technologies
and social change in prehistory and antiquity. Verlag Marie
Leidorf GmbH, Rahden/Westfalia, pp 231–239
Rovira S, Montero Ruiz I, Consuegra S (1997) Las Primeras Etapas
Metalúrgicas en la Península Ibérica. I Análisis de Composición.
Instituto Universitario Ortega y Gasset, Ministerio de Cultura, Madrid
Ruiz-Taboada A, Montero-Ruiz I (1999) The oldest metallurgy in
Western Europe. Antiquity 73:897–903
Smirniou M, Rehren T (2016) The use of technical ceramics in early
Egyptian glass-making. J Archaeol Sci 67:52–63
Smith CS (1982) A search for structure: selected essays on science, art
and history. MIT Press, Cambridge (MA)
Soares AM, Valerio P, Frade JC, Oliveira MJ, Patoilo D, Ribeiro I, Arez
L, Santos FJC, Araujo MF (2008) A Late Bronze Age stone mould
for flat axes from Casarao da Mesquita 3 (Sao Mancos, Evora,
Portugal). In: Archaeometallurgy in Europe 2007, 2nd
International Conference, 145–157. Milan: Associazione Italiana
di Metallurgia
Sofaer J (2006) Pots, houses and metal: technological relations at the bronze
age tell at Százhalombatta, Hungary. Oxf J Archaeol 25(2):127–147
Strahm Ch, Hauptmann A (2009) The metallurgical developmental
phases in the Old World. In: Kienlin, T.; Roberts, B. (Eds.): Metals
and Society. Studies in honour of Barbara S. Ottaway, 116–127.
UPA 169. Bonn: Verlag Dr. Rudolf Habelt GmbH
Tite MS, Kilikoglou V, Veknis G (2001) Strength, toughness andthermal
shock resistance of ancient ceramics, and their influence on techno-
logical choice. Archaeometry 43:301–324
Valério P, Silva RJC, Soares AMM, Araújo MF, Gonçalves AP, Soares
RM (2015) Combining x-ray based methods to study the protohis-
toric bronze technology in western Iberia. Nucl Inst Methods Phys
Res B 358:117–123
Valiente Cánovas S, Gea García A, López Cidad JF, Ayarzagüena Sanz M
(2003) Algunos datos sobre cestería y fibras vegetales aplicadas a
vasijas en barro de la Edad del Bronce en las BSalinas de Espartinas
(Ciempozuelos, Madrid)^. Pátina 12:101–108
Zori C, Tropper P, Scott D (2012) Copper production in late prehispanic
northern Chile. J Archaeol Sci 40(2):1165–1175
Archaeol Anthropol Sci