Archaeol Anthropol Sci

DOI 10.1007/s12520-015-0255-2

ORIGINAL PAPER

On the making of copper oxhide ingots:

evidence from metallography and casting experiments
Andreas Hauptmann 1 & Ralf Laschimke 2 & Maria Burger 2

Received: 5 May 2015 / Accepted: 9 June 2015

# Springer-Verlag Berlin Heidelberg 2015

Abstract We report on the texture, phase content and on the Their shape is unique: They are rectangular plates of copper
making of Late Bronze Age copper oxhide ingots as excavat- showing extremities at each corner and, hence, imitating the
ed from the shipwreck of Uluburun. This type of ingot repre- shape of a stretched skin of an ox, or, better to say, of a goat.
sented the Bworld market^ bulk metal in the Mediterranean Many ingots have a weight of ca. 24 kg, but some may reach
and beyond. The ingots are characterized by a high porosity, even 30 kg. It is generally assumed that the shapes of the ingots
by inclusions of slag fragments and cuprite which result from were made to disburden their transport as shown in the Egyptian
casting raw copper smelted before from ores. The ingots were wall painting in Fig. 1.
cast from several batches of copper. The shape of the oxhide The discovery of a shipwreck excavated off the coast at Cape
ingots was never fully understood, and the way of casting was Gelidonya, Turkey (Bass 1967), revealed nearly three dozens of
uncertain due to the absence of casting moulds. It is the aim of twelfth-century BC complete copper oxhide ingots and many
this paper to discuss the making of these ingots. We will pres- fragments of such ingots. The most spectacular find was the
ent an alternative model on how to cast such ingots using four discovery of the thirteenth-century BC shipwreck of Uluburun,
crucibles. The experiments will show that the texture of the again off the coast of southern Turkey (see overview in Pulak
ingots is created by gases such as hydrogen, carbon monoxide 2000). Hundreds of copper oxhide ingots and of bun-shaped
and water vapour. The high porosity is shown to be advanta- ingots made of copper and tin oxhide ingots and many fragments
geous for breaking the ingots into fragments. (ca. 10 t of copper and 1 t of tin) directed much attention to this
form of ingot. This was the largest find of copper oxhide ingots
Keywords Copper oxhide ingots . Experimental ever made. It demonstrated that copper oxhide ingots, and, sub-
archaeology . Sand casting . Crucibles . Texture . Porosity ordinated, bun-shaped ingots made up the overwhelming bulk of
metal that was traded Bworldwide^ during the Late Bronze Age
(LBA), both by sea and partly over land routes. It is a well-
Introduction: oxhide ingots in context known fact that copper oxhide ingots were widely distributed
from the eastern Mediterranean to Sardinia in the west, and these
Late Bronze Age copper oxhide ingots are well known in archae- ingots were found in Anatolia, in Bulgaria (Gale 1991, 201), and
ology since the middle of the nineteenth century (Spano 1858). they were even traded as far as southern Germany (Primas and
Pernicka 1998).
Copper oxhide ingots were repeatedly investigated in the past.
* Andreas Hauptmann Buchholz (1966, 1967) prepared a detailed classification of cop-
andreas.hauptmann@bergbaumuseum.de
per oxhide ingots based upon stylistic features. Maddin and
Ralf Laschimke Muhly (1974) reported on the metallography and chemical
laschimke@t-online.de
composition of some ingots from the Cape Gelidonya
Maria Burger shipwreck to obtain clues to the origin of the ingots and to
maria.burger@zollern.de
shed light on the origin of the ship or the origin of the copper.
1
Deutsches Bergbau-Museum Bochum, 44787 Bochum, Germany The study of these particular ingots answered only few of these
2
Zollern GmbH and Co KG, 72479 Sigmaringendorf, Germany questions but did stimulate further investigations of this class of
 Archaeol Anthropol Sci

In spite of the previous studies, however, it was not possi-

ble to find out how and where the copper oxhide ingots were
cast. In the archaeological record, ceramic casting moulds, as
they are so often found for making Bronze Age tools, weapons
or jewellery, were missing not only in Cyprus but also in the
entire Mediterranean and beyond. Just a single so-called cast-
ing mould made of limestone suitable to cast an oxhide ingot
was found at Ras Ibn Hani in Syria (Lagarce 1986). A second
ceramic fragment of an eleventh-century BC questionable
mould was found at Timna (Ben-Yosef 2012).
Where are the moulds for casting the ingots? Were they cast in
sand, as previously supposed by Lagarce et al. (1983) and
Craddock et al. (1997)? As mentioned already by Hauptmann
et al. (2002), a number of other questions remained open about
the making and consistency of these ingots: What was the quality
of the copper of the oxhide ingots? Were oxhide ingots produced
by tapping metal from smelting furnaces or from a crucible? Do
the copper ingots consist of raw copper resulting from a
Bprimary^ smelting process or were they produced by remelting
or refining copper or even (in some cases) recycling bronze
objects? How was a Bductile^ metal such as copper broken up
Fig. 1 Workers carry so-called oxhide-shaped copper ingots and baskets
into small pieces? The metal is extremely malleable, that is, to
suggested to be filled with small rounded ingots of lead and tin. The shape
of the ingots was thought to be shaped to disburden the transport of the reduce it into smaller pieces, it would, under normal circum-
ingots which often reach a weight of 24 kg and more. Wall painting in the stances, be extremely difficult since it, as for any ductile metal,
tomb of Rekhimre, Thebes. ca. 1500 BC. After Davies (1943, plate 53) would simply deform when subjected to an external stress.
The present paper is a roundup of the investigations pub-
ancient copper. LoSchiavo et al. (1990) published an investiga- lished by Hauptmann et al. (2002) and of the experimental
tion of copper ingots of the bun-shaped and of the oxhide type work performed by Laschimke and Burger (2011). We will
from Sardinia. Part of this study was an investigation on the discuss in this paper an explanation on the material utilized for
microstructure, chemical composition and statistical evaluations making of the oxhide ingots and of the casting techniques
by Maddin and Merkel (1990). In this way, the authors tried to which were used in the Late Bronze Age. These are different
decipher the making of the ingots and to follow the path from from the general view. We suggest that the shape of the ingots
smelting to this product. These questions were deepened by was also a result from the technology of casting applied and
Hauptmann et al. (2002) and by Hauptmann and Maddin that it was not only chosen for practical reasons.
(2005). They were the first who were permitted to drill a set of
4–5-cm long-core samples with an inner diameter of 6 mm from
several copper oxhide ingots. Investigating these samples, the The metallography of the copper oxhide ingots
authors could demonstrate that the copper would be peppered
with slag inclusions and batches of primary copper oxide. In There are two persistent metallographic characteristics in just
addition, anomalous porous volumes inside the metals were ob- about all of the samples of both the oxhide and bun copper ingots
served, and cooling rims indicated that the ingots were obviously investigated so far from many localities in the Mediterranean.
cast in several batches. One is the constant and extremely high porosity (Begemann
The question of the provenance of copper (and tin) was et al. 2001; Hauptmann et al. 2002). The other is the ubiquitous
intensively studied during the last two decades by Gale and presence of slag inclusions inside the copper matrix.
Stos-Gale (see, e. g. Stos-Gale et al. (1998) and Stos-Gale
(2000)) and by Begemann et al. (2001) using measurements Porous volume of the copper ingots
of lead isotope abundance ratios. The authors meticulously
traced the origin of copper oxhide and bun-shaped ingots in Among the samples, we observed an anomalous high porous
the Mediterranean back to Cyprus. Part of these investigations volume which shows throughout the copper oxhide and bun-
was the aim to specify the implications connected with the shaped ingots to be around 20 % and more. The ingots contain
observation that many of the oxhide ingots (and of the bun- numerous cavities that can be classified (also by macroscopic
shaped ingots as well) are not homogeneously composed but view) in sizes of a few micrometres (micropores) to those up
were built up of several batches of copper. to the centimetre scale (macropores). For the first time, it
Archaeol Anthropol Sci

could be demonstrated by the plugs drilled from the ingots that Their texture shows evidence of reheating. Occasionally, we
the copper oxhide and bun-shaped ingots of the LBA have an identified copper sulphides (chalcocite, Cu2S) or mixed
enormous porous volume which exceeded our previous ideas copper-iron sulphides (nearly chalcopyrite, CuFeS2) in these
on the inner texture of the ingots by far (Fig. 2). slag inclusions. There are striking similarities to slag inclu-
In general, cavities like that are caused by the effervescence of sions described by Zwicker et al. (1980) in copper oxhide
gases such as oxygen, carbon monoxide and dioxide, by water ingots from Sardinia. Probably, the copper used to cast these
from burning charcoal as well as sulphur dioxide from oxidation ingots was a raw copper from a primary smelting process
of sulphide inclusions in the copper in the liquid state. The gases which was not thoroughly separated from adhering slag.
dissolve violently during solidification, a phenomenon known in This suggestion is supported by both the archaeological and
metallurgy as BSpratzen^. In addition, porous volume is caused the experimental evidence which indicates that a complete
all over the ingots by interdendritic shrinkage. The porosity is separation between slag and copper to remove a piece of clean
high in the upper part of the ingots where centimetre-sized, flat metal was hardly managed in ancient metallurgy. Examples
gas bubbles are formed which bulge the surface. At the bottom, are given by Bamberger and Wincierz (1990, 134, Fig. 132).
the porous volume is low, and the texture is denser. It is perfo- The metal produced by experimental smelting is covered by a
rated by a network of tubular pinholes. Also the oxygen content thin layer of slag. The same is true for a piece of copper from
of the copper is abnormally high. It is shown by the occurrence of Beer Ora, Timna (Tylecote and Bachmann 1990, 76, Fig. 105)
cuprite (Cu2O). Zwicker et al. (1980) suggested a surplus of that is covered by slag also. At Faynan, lumps of copper found
oxygen in the liquid which exsolved during solidification. The at smelting sites usually show adhering remains of slag (un-
equilibrium of the binary system Cu-O (Hansen and Anderko published results Bochum). As observed by Bisson (2000,
1958) is not in agreement with this hypothesis. 141 ff. from an ethno-archaeological context in Africa, such
As shown by previous studies (see above), the high poros- adhering slag crusts have to be removed by hammering the
ity seems to be a characteristic feature of LBA copper ingots metal lumps produced by smelting and by repeated melting
with a provenance from Cyprus. It was not observed, e.g. in processes in order to obtain high-quality metal. The author
bun-shaped ingots from the third millennium BC Oman points out that these were regular steps in copper production
(Hauptmann 1987, 210). This was the traditional bulk metal he observed at many metallurgical sites all over the middle
traded in the Bronze Age in the Persian Gulf area. Copper and southern part of Africa. Obviously, the Uluburun oxhide
ingots produced at the Early Bronze Age (ca. 2300 BC) metal and bun-shaped ingots were not processed in such a way.
production workshop at Khirbet Hamra Ifdan, Faynan, consist
of perfectly dense metal (Levy et al. 2002), though they have a
Cuprite inclusions
weight of only ca. 250–300 g.
Almost all copper samples investigated contain cuprite
Slag inclusions (Cu2O). Of subordinated interest for our investigation is the
formation of cuprite by corrosion processes that just mirrors
All ingots contain amounts of angular-shaped inclusions of the weathering of the ingots in the sea since the ship sunk in
slags up to 5 mm in diameter (Fig. 2). They consist mainly the LBA. More important are Cu-Cu2O intergrowths that re-
of a Fe-rich olivine (fayalite, Fe2SiO4) and magnetite (Fe3O4). sult from the nearly complete immiscibility of oxygen in

Fig. 2 Copper oxhide ingot, four

handle, Uluburun (KW 3695).
Two of four cores taken near the
edges of the rectangular ingot.
Note the extremely high macro-
and microporosity of the ingot
(black holes) and the slag and
primary/secondary cuprite
inclusions (grey inclusions). Core
drilling from surface (above) to
bottom (below). Lengths of the
cores 3.3 cm (above) and 2.7 cm
(below). Photo from Hauptmann
et al. (2002), Fig. 5d
 Archaeol Anthropol Sci

copper both in the liquid and in the solid state. The texture from pouring several batches of metal into the mould to com-
formed by the intergrowth of Cu-Cu2O gives valuable infor- plete an ingot.
mation on the oxygen concentration in the original melt. The
eutectic intergrowth (Fig. 3a), e.g. indicates an oxygen con- Copper sulphide inclusions
centration of 0.39 wt%. The copper ingots from Uluburun
often contained lumps and larger equiaxed crystals of cuprite The third type of inclusion is made up of nearly pure copper
indicating a hypereutectic composition (Fig. 3b). This is sulphides as observed already in the slag inclusions. These are
caused by a considerable surplus of oxygen in the gaseous relics of the sulphidic ore used for smelting. Sulphide smelting
atmosphere of the reaction vessel or during casting. These seems to have been a well-developed technology at LBA
oxide particles affect the malleability of the copper and make Cyprus. The sulphides probably have been roasted, either be-
it brittle. Also hydrogen or carbon monoxide rapidly improves fore the first smelt, or, if matte was produced, after it, because
the embrittlement (Tyler and Black 1990). most of the ore available on Cyprus consists of chalcopyrite.
Under the microscope, one can see thin rims of cuprite This sulphide does not occur in the copper.
along overlapping, pillow-like lines (Fig. 4) in the copper
matrix. Hauptmann et al. (2002) interpreted these lines as
cooling rims formed by intermediate casting batches of Casting experiments
copper. This parallels what is already visible in the
macroscale. Pulak (2000, 142, Fig. 6) shows oxhide ingots Even if there is abundant evidence for a unique mass produc-
composed of two centimetre-sized layers which also result tion of copper in Cyprus, it still remains an open question how
the copper ores were smelted to metal. Remains of smelting
furnaces, as they were excavated by Fasnacht (1989, 62 ff. are
extremely rare. It also had to be investigated how the copper
oxhide ingots (and the bun-shaped ingots) were cast. The pur-
pose of the famous 17-l crucible or smelting furnace from
LBA Enkomi (Dikaios 1969, vol. IIIa, plate 159, 20) is
debated, and neither investigations of samples by Zwicker
(1986) nor by Tylecote (1982, 92) could solve the question
whether ore was smelted or metal lumps were melted in this
Breaction vessel^.
Therefore, two of the authors (R.L. and M.B.) decided to
perform some experimental work to clarify details of the cast-
ing operations to produce the two sorts of ingots. The exper-
imental work was performed at the Zollern GmbH laboratories
at Sigmaringendorf. Copper wirebar with an oxygen concen-
tration of 0.3 % was used for the melting simulations. During
our experiments, the oxygen concentrations increased to
0.5 % by remelting the wirebar copper. In this study, we will
focus on the making of copper oxhide ingots.
Considering the archaeological material, the oxygen con-
centration in the ingots was much higher. The idea of the
casting experiments, therefore, was to create a development
of gases by a reaction with the mould face in the casting
mould. This is in striking contrast to the basics of modern
casting technology where such reactions have to be avoided
because they lead to a gassing of the liquid metal. Reversing
this perception, we prepared a reactive mould consisting of a
Fig. 3 a (above) Copper oxhide ingot, Uluburun (KW 1396). Eutectic
gas-expelling material. We took care to use only such mate-
intergrowth of copper and cuprite (Cu2O) between equiaxed copper rials for the experiments which could have been available in
grains. Voids are black. Oil immersion, magnification ×200. Photo from ancient times. Agricola, in his famous opus BDe re metallica^
A. Hauptmann, Deutsches Bergbau-Museum Bochum. b (below) Copper (Agricola 1556), published a suitable recipe to prepare the
oxhide ingot, Uluburun (KW 3474). Agglomeration of cuprite formed by
a surplus of oxygen in the liquid copper during casting. Note the
lining of a smelting furnace. He recommended to mix one part
interdendritic shrinkage of copper. Oil immersion, magnification ×100. of clay and two parts of powdered charcoal in dry state and
Photo from Hauptmann et al. (2002), Figs. 10 and 11 add some water to shape it. Agricola termed this mixture
Archaeol Anthropol Sci

Fig. 4 Copper oxhide ingot,

Uluburun (KW 3421). Porous
volume of a bun-shaped ingot
caused by interdendritic
shrinkage and gas bubbles. Note
several interfaces of dense metal
which point to different batches
during casting. Oil immersion,
magnification ×100. Photo from
Hauptmann et al. (2002), Fig. 6

BGestübbe^. It turned out that the charcoal powder added to was thought to be necessary to create gas-producing reactions
the mould material was a useful temper because it minimized with the liquid copper. To reach oxygen concentration
the decrease of its volume during drying considerably. as high as in the archaeological ingots, we melted the
wirebar copper for our casting experiments in a chamber
furnace by air access in cylindric crucibles made of a
Casting oxhide ingots in Bsand^
mixture of clay and graphite.
In several respects, we were extremely cautious during the
In Fig. 5 the experimental arrangement is shown. The material
casting processes. We were afraid that, due to the low stability
was set into a foundry flask. The moulding sand manufactured
of the air-dried moulds, a vertical casting of liquid copper
following Agricola’s recipe was air-dried for 6 weeks. After
would destroy the oxhide shape. We therefore decided to cast
that period, the moulding material had a water content of ca.
the liquid copper simultaneously out of four small crucibles
15 wt%. This was acceptable as water in the mould material
placed at each corner almost horizontally into the open sand
casting (Fig. 6). This method has a convincing modern equiv-
alent: cast iron stove plates were cast in the same way until the
recent past. It also solves another problem: it is difficult to
understand how a ceramic crucible, as it was used in ancient
times, could keep 25–30 kg of liquid copper. After our expe-
rience, such a crucible would hardly be stable enough to han-
dle it for a casting process. Hence, it seemed more realistic to
cast the liquid copper in several batches as proved by the
archaeological evidence. Merkel (1986, 1990) demonstrated,
by his smelting experiments, that according to Bronze Age
methods, masses of copper up to ca. 10 kg could be produced
in a smelting furnace. He pointed out, however, that the pro-
duction of 30 kg of copper in a smelting furnace to form an
oxhide or a bun-shaped ingot would be extremely difficult.
Following this model, we suggest that remelting of batches
Fig. 5 Experimental arrangement to replicate casting a Late Bronze Age
and lumps of solid copper to larger units must have been a
copper oxhide ingot. The picture shows the situation 10 min after casting. common practice during this period and was practised in
At this moment, the copper had a temperature of ca. 750–800 °C. The crucibles, as pointed out by Cooke and Nielsen (1978) and
liquid metal was cast simultaneously from four crucibles into a mould Zwicker et al. (1985). It will be the focus of future experimen-
made of clay and charcoal powder. The casting mould is embedded in a
modern foundry flask with a width of ca. 1 m. We suggest the Late
tal work to perform casting of ingots using flat ceramic cruci-
Bronze Age moulds to be deepened in the soil. Photo from R. bles as depicted in the wall painting of the tomb of the Vizier
Laschimke and M. Burger, Sigmaringendorf Rekhmire at Thebes, ca. 1500 BC (Fig. 7).
 Archaeol Anthropol Sci

Fig. 8 Gas producing reactions during casting an oxygen-containing

copper melt into an almost rectangular-shaped mould made of clay and
charcoal cause an inflammation of H2 and CO. Water steam evaporates
also. The gases are mainly formed in the mould material. They lead to the
Fig. 6 Horizontal casting of liquid copper from four crucibles into a formation of centimetre-sized gas bubbles. Micropores are formed by
mould made of clay and crushed charcoal . The liquids will not be water through secondary reactions. The casting mould again is embedded
evenly mixed. Gases produced by reactions at the moulding face are in a modern foundry flask with a width of ca. 1 m. Photo from R.
inflamed in greenish-yellowish colour at the surface of the liquid. The Laschimke and M. Burger, Sigmaringendorf
casting mould again is embedded in a modern foundry flask with a width
of ca. 1 m. Photo from Laschimke and Burger (2011), Fig. 9 At the freezing point, a cooling rim is formed at the surface of
the melt. Rising gas bubbles with a size of up to 3 cm are now
Macroporosity and breakable copper prevented to escape and are captured in the setting melt and
are frozen (Fig. 9). They lead to a humpy and blistered surface
It was found that when the melt gets in contact with the mould as it is also characteristic for the archaeological ingots
face, water abruptly steams from its upper layer. This layer (Hauptmann et al. 2002).
will develop a porosity and enable a suitable permeability for Similar to the formation of the surface, a cooling rim is
gases formed in deeper layers of the mould to escape with formed subsequently at the bottom of the ingot. Here, the
retardation. Hence, no splitting effect will be developed like solidification prevents a gas intrusion from the mould material
it will be the case in a stone mould. At the same time, reactive into the metal. On the other hand, gases captured in the metal
gases are formed along with the evaporation of water. They matrix finally break through the set bottom, where a large
will be inflamed at the surface of the melt, as shown in Fig. 8. number of vertical tubular pinholes will be formed (Fig. 10).

Fig. 7 Wall painting in the tomb

of Rekhmire showing melting and
casting of copper or bronze in an
open hearth surrounded by stones.
The flat ceramic crucible was
handled by two persons with two
(wooden) sticks used as pliers.
Probably the crucible was set
upon a (nonvisible) stone block.
In the background, perhaps a heap
of charcoal ash and a vessel for
charcoal. Ca. 1500 BC. From
Davies (1943), plate 52
Archaeol Anthropol Sci

and forms H2O. During cooling, Cu2O will be precipitated,

and the water vapour forms small bubbles which keep
adhering to the Cu2O crystallites and will be frozen as
micropores (Fig. 10). The masses of micropores in the
texture of the archaeological as well as in the experimentally
produced ingots are an indication that the Bronze Age ingots
were not cast in (lime-) stone moulds like that found in Ras
Ibn Hani (Fig. 11).

The case of Ras Ibn Hani

Following the archaeological evidence published by Lagarce

et al. (1983), Craddock et al. (1997) experimentally cast a
Fig. 9 The surface of the ingot 8 min after the liquid metal was cast into
the mould. A solidified surface forms an impermeable skin and prevents copper oxhide ingot in a reduced scale in a limestone replica
gases from escaping. This causes large gas bubbles to swell under the mould. We followed their simulations and we cast an oxygen-
surface and to arch upwards at many points. Width of the picture ca. containing copper melt in a limestone mould with an original
15 cm. Photo from Laschimke and Burger (2011), Fig. 5 dimension of the Bronze Age copper oxhide ingots
(Laschimke and Burger 2011). This mould was preheated
The pinholes of our experimentally cast ingots are identical for 24 h at 300 °C to expel any moisture from the stone.
with the archaeological ingots. When the hot copper melt (1140 °C) contacted the mould,
the limestone immediately was thermally decomposed, and a
violent gas evaporation with carbon dioxide developed. This,
Microporosity
however, decreased shortly after. It was in contrast to the long-
lasting degassing in the previously used mould made of a
The microporosity of the ingots allows valuable conclusions
mixture of clay + charcoal.
about the Bronze Age casting technology. In Fig. 7, it is dem-
The explanation for this behaviour is the following: In a
onstrated that chemical reactions will take place not only in
mould made of clay + charcoal, water will evaporate also from
the liquid copper cast in the sand mould but also in the mould-
deeper layers as soon as temperatures of ca. 100 °C are
ing sand itself. When the water content in the moulding ma-
reached. It will react with high-temperatured carbon at the
terial evaporates, it will partly react with the carbon embedded
interface of the mould face and the melt and forms hydrogen
in the BGestübbe^ and forms a gas mixture of hydrogen +
+ carbon monoxide (see above). This process continues until
carbon monoxide (H2 + CO). Similar to the mechanism of
the water in the mould face is more or less completely
the hydrogen brittleness of solid copper (Dies 1967), hydro-
gen is diffusing into the liquid metal. Here, it causes a reduc-
tion of Cu2O which is solved in the copper in the liquid state

Fig. 10 Section of an experimentally produced copper oxhide ingot

(upside down), showing the dense pattern of pin holes at the bottom (in Fig. 11 Microtexture of an experimentally produced copper oxhide
the picture upside). At the surface of the ingot (in the picture below), flat ingot. It contains ca. 0.45 % O2. The eutectic composed matrix shows
tubular macropores are formed which caused a humpy skin. Width of the an intergrowth of Cu + Cu2O. Black pores are often associated with larger
oxhide ingot ca. 40 cm. Photo from R. Laschimke and M. Burger, crystals of (bluish-grey) cuprite. Photo from R. Laschimke and M.
Sigmaringendorf Burger, Sigmaringendorf
 Archaeol Anthropol Sci

evaporated. The thermal decomposition of limestone starts not The intention of breaking copper plates into frag-
before ca. 650 °C. This temperature penetrates only very slow- ments was probably to meet the individual demands of
ly into the stony mould face and leads to a quick stop of any traders and dealers. Smaller pieces also facilitated fur-
gas development. ther processing. It must have undergone additional local
Neither the copper oxhide ingots produced by our experi- refining to finally obtain a quality of metal suitable for
ments nor the one cast into a limestone replica mould by casting (alloyed) objects potentionally alloyed with tin.
Craddock et al. (1997) did show any micropores as they were Smaller portions improved the reactions in small-sized
observed as a rule in the many ingots from Uluburun or from melting furnaces or crucibles.
Sardinia (Hauptmann et al. 2002; Maddin and Merkel 1990). Breaking a ca. 3–5-cm thick copper plate like the
The reason for the absence of such micropores in these ingots oxhide ingots would be impossible if the ingot would
is that only CO2 was produced during casting by the decom- consist of massive copper. This, however, is easily pos-
position of the limestone at high temperatures. This gas, how- sible if the copper plates had a porosity as high as the
ever, does not react with Cu2O; no bubbles of water vapour ones and would contain inclusion of slag and cuprite.
are produced, and hence, no micropores could develop. We, Figure 12 demonstrates that one of our experimentally
therefore, suggest that the missing microporosity in the exper- produced copper oxhide could be broken by simply
imentally produced ingots is a proof that the so-called stone crashing it with a handle-sized hammer.
mould from Ras Ibn Hani could not (repeatedly) have been
used for casting oxhide ingots. Another argument against the
The crucibles
limestone casting moulds are tension cracks which will be
formed while pouring a >1000 °C melt of copper into it.
All copper oxhide (and bun-shaped) ingots shown in Fig. 13
Such cracks will disable their further use. Karageorghis and
were cast by the technology explained above. They all provide
Kassianidou (1999) argued reasonably that the stone of Rasi
the same characteristics of macro- and microporosity as found,
Ibn Hani perhaps had a cultic meaning.
e.g. in the ingots found at Uluburun (Hauptmann et al. 2002).
It was pointed out above that clearly identifiable finds of
Their oxygen concentration was at ca. 0.5 wt%. So far, a
casting moulds are totally missing in the archaeological re-
conclusive explanation for those high-oxygen concentrations
cord, not only in Cyprus but also in the entire Mediterranean
was missing.
and beyond.
Our explanation is the following: A basic aspect is the way
The archaeological situation is in striking disproportion to
how the liquid copper was poured from four edges into the
the large number of copper oxhide and bun-shaped ingots
mould. This is supported by the metallographic evidence that
because, presumably, all the fragile casting moulds were
the ingots were composed of non-homogenized batches of
destroyed. This also happened to our casting moulds.
metal which were not fully mixed in their liquid state.
Exposed to the climate in Middle Europe after 3 years, they
Figure 7 illustrates the shape and dimension of flat crucibles
were highly weathered. Their original shape was
used in Late Bronze Age Egypt for melting and casting activ-
irrecognizable. The same would be true if the mould face
ities. Comparable crucibles with a diameter of ca. 0.4 m were
would contain horse dung or chaff instead of charcoal as a
also used in the famous foundry of Phidias at Olympia
basis for the development of gases in the mould.
(Zwicker 1984). The metal was melted under a heap of char-
coal with artificial air supply from two opposite sides.
Breaking oxhide ingots

The extremely high porosity and the inclusions of slag, as

well, were probably not a severe quality shortage which ham-
pered the business with the Bworld bulk copper^ (Hauptmann
and Maddin 2005). Rather, it provided some important advan-
tages: in order to trade the copper to various customers, it had
to be broken often into fragments and pieces. Trading and
exchange of commodities like copper in various weights were
a commonplace in the Bronze Age. Forty-eight different-sized
fragments of copper oxhide ingots were found on the ship-
wreck of Uluburun (Pulak 2000). The cargo of the shipwreck
discovered at Gelidonya comprised of 34 complete copper
Fig. 12 The bottom of the ingot is laced by pinholes. The oxhide ingot
oxhide ingots, five-halves of ingots, 12 edges and 75 kg of shown in the picture was easily breakable into fragments, as it often
different-sized fragments (Bass 1967). Hundreds of fragments occurs in the archaeological record. Width of the oxhide ingot ca.
were excavated in Sardinia (LoSchiavo 1989). 40 cm. Photo from Laschimke and Burger (2011), Fig. 12
Archaeol Anthropol Sci

have shown that this shape may also result from a special
casting process. A decisive technological feature was a cast
from four crucibles at the edges of the oxhide-shaped mould in
a horizontal way (Fig. 6). Here, we followed previous obser-
vations that many of the ingots from archaeological contexts
are composed of several batches of metal. The batches of the
copper were not fully mixed in the liquid state (Fig. 4) but
formed layers and pillow-like cooling rims visible macroscop-
ically and under the microscope.
It was also generally accepted that oxhide ingots have been
produced in open moulds. In view of the disproportion be-
tween the large number of ingots and the virtually absence
of casting moulds in the archaeological record, casting in sand
was considered. We doubt that permanent moulds like the
Fig. 13 Results of our experimental work: six copper oxhide ingots and limestone mould discovered at Ras Ibn Hani have been used.
two bun-shaped ingots. The weight of the oxhide ingots is between 28 Our experimental studies have shown that such moulds rapid-
and 30 kg. Their length is 65 cm. Bun ingots could be remelted in flat ly cracked due to their thermal instability. Furthermore,
crucibles without any problems. Photo from Laschimke and Burger
oxhide ingots cast in limestone moulds do not exhibit the
(2011), Fig. 17
significant porosity as observed in the Bronze Age copper
oxhide ingots.
Although the metal was covered by glowing charcoal, ox-
These ingots, which were found in the eastern
idizing conditions would prevail at the surface. The chemo-
Mediterranean and beyond, exhibit both extremely high
physical explanation for this seemingly paradox is twofold. It
micro- and macroporosity. They also contain inclusions of
might be that directly above the crucible, the charcoal is more
slag and cuprite. Slag inclusions are interpreted as indication
or less completely burned and transformed to ash and slag.
for remelting raw copper that was produced in a smelting
There will be a surplus of oxygen which leads to the formation
furnace and was not completely separated from the slag. The
of cuprite at the surface of the melt and will oxidize iron and
high porosity was not a drawback but appeared to be perhaps
sulphur present in the copper. Furthermore, a complete reduc-
an intended feature. It facilitated the breaking of the large
tion of cuprite, once it is formed, is not possible because the
copper plates into fragments of individual size according to
liquid is not in direct contact with carbon. According to
the requests of traders. Smaller fragments with a high porosity
Boudouard’s equilibrium, at temperatures necessary to melt
will also improve reactions such as alloying copper with tin in
copper (≥1083 °C), CO formed in a distance in the charcoal
smaller furnaces.
heap will not reduce the cuprite and will not be oxidized itself
We suggest that the extraordinary porosity of the oxhide
to CO2.
ingots is caused by carbon monoxide and hydrogen. These
The flat crucibles illustrated in Fig. 7 had a capacity of ca. 7
gases develop when water vapour reacts at high temperature
to 8 kg of metal and could be handled with a pair of wooden
with solid carbon, such as charcoal.
sticks used as tongs. To cast an ingot of ca. 30 kg, liquid
In our experiments, we applied the technology of open
copper had to be prepared in four crucibles simultaneously.
sand casting. However, instead of normal moulding sand,
This obviously was not a problem at all. Excavations of a
we used a mixture of clay containing pulverized charcoal
foundry at Qantir/Pi-Ramesse (Egypt, capital of
which, air-dried, contained 15 % water. Oxygen-rich copper
Ramses the Great and his followers, ca. 1300–1000
was poured into these moulds.
BC) revealed an excellent insight into the industrial or-
The moulds reacted intensively with the liquid copper, and
ganization of a bronze-casting workshop, where dozens
water vapour, carbon monoxide and hydrogen developed.
of workers were melting metal at the same time to pro-
These gases led to a considerable macroporosity in the copper
duce larger objects in an almost industrial scale (Pusch
ingots. Furthermore, free hydrogen reacted with the copper
1990; Rehren 1999).
oxide dissolved in the copper melt, generating water vapour,
which led to the formation of countless micro bubbles within
the melt. They were frozen during the solidification of the
Summary copper. This produced an enormous microporosity in the
oxhide ingots. All in all, the experimentally produced oxhide
So far, it was widely accepted that the shape of Late Bronze ingots exhibited a type of macro- and micro porosity identical
Age oxhide ingots was deliberately chosen to facilitate their to the porosity of the oxhide ingots excavated from the ship-
transport. This is convincing. However, our investigations wreck of Uluburun.
 Archaeol Anthropol Sci

References Laschimke R, Burger M (2011) Archäometallurgische Experimente zum

Giessen von bronzezeitlichen Ochsenhautbarren aus Kupfer. Metall
3:86–92
Agricola G (1556) De Re Metallica Libri XII. Translated and annotated Levy T, Adams RB, Hauptmann A, Prange M, Schmitt-Strecker S,
by G. Frauenstadt and H. Prescher. Deutscher Verlag der Najjar M (2002) Early Bronze Age metallurgy: a newly dis-
Wissenschaften Berlin 1974 covered copper manufactory in southern Jordan. Antiquity
Bamberger M, Wincierz P (1990) Ancient smelting of oxide copper ore. 76:425–437
In: Rothenberg B (ed) The ancient metallurgy of copper. Researches LoSchiavo F (1989) Early metallurgy in Sardinia. In: Hauptmann A,
in the Arabah 1959–1984, vol 2. Institute of Archaeo-Metallurgical Pernicka E, Wagner GA (eds), Old World Archaeometallurgy, Der
Studies, London, pp 123–157 Anschnitt, Beiheft 7: 33–38
Bass GF (ed) (1967) Cape Gelidonya: a Bronze Age shipwreck. Trans LoSchiavo F, Maddin R, Merkel JF, Muhly JD, Stech T (1990)
Am Philos Soc 57: 8 Metallographic and statistical analyses of copper ingots from
Begemann F, Schmitt-Strecker S, Pernicka E, Loschiavo F (2001) Sardinia. Quaderni 17:1–227
Chemical composition and lead isotopy of copper and bronze from Maddin R, Merkel JF (1990) Metallographic and statistical analyses. In:
Nuragic Sardinia. Eur J Archaeol 4(1):43–85 LoSchiavo F, Maddin R, Merkel JF, Muhly JD, Stech T (eds)
Ben-Yosef E (2012) A unique casting mould from the new excavations at Metallographic and Statistical Analyses of Copper Ingots from
Timna 30 (Israel): evidence from western influence. In: Kassianidou V, Sardinia. Quaderni 17: 43–199
Papasavvas G (eds) Eastern Mediterranean metallurgy and metalwork Maddin R, Muhly JD (1974) Some notes on the copper trade in the
in the second millennium BC. A conference in honour of James D. ancient Mid-East. J Met 5:1–7
Muhly, Nicosia, October 2009. Oxbow Books, Oxford, pp 188–196 Merkel JF (1986) Ancient smelting and casting of copper for oxhide
Bisson MS (2000) Precolonial copper metallurgy: sociopolitical context. ingots. In: Balmuth M (ed) Studies in Sardinian Archaeology 2:
In: Bisson MS, Childs ST, de Barros P, Holl AFC (eds) Ancient Sardinia in the Mediterranean. Ann Arbor, p 251–264
African metallurgy. Walnut Creek, Altamira-Press, pp 83–145 Merkel JF (1990) Experimental reconstruction of Bronze Age copper
Buchholz HG (1966) Talanta. Neues über Metallbarren der smelting based on archaeological evidence from Timna. In:
ostmediterranen Spätbronzezeit. Schweiz Münzblätter 62:58–72 Rothenberg B (ed) The ancient metallurgy of copper. Researches
Buchholz HG (1967) Analysen prähistorischer Metallfunde aus Zypern in the Arabah 1959–1984. Institute of Archaeo-Metallurgical
und den Nachbarländern. Berl Jahrb Frühgeschichte 7:189–256 Studies, London, pp 78–122
Cooke SRB, Nielsen BV (1978) Slags and other metallurgical products. In: Primas M, Pernicka E (1998) Der Depotfund von Oberwilfingen.
Rapp G, Aschenbrenner SE (eds) Excavations at Nichoria in Southwest Neue Erkenntnisse zur Zirkulation von Metallbarren. Germania 76:
Greece. University of Minnesota Press, Minneapolis, pp 182–224 25–65
Craddock PT, Freestone IC, Dawe CD (1997) Casting metal in limestone Pulak C (2000) The copper and tin ingots from the Late Bronze Age
moulds. J Hist Metall Soc 31:1–7 shipwreck at Uluburun. In Yalcin Ü (ed), Anatolian Metal I. Der
Davies NG (1943) The tomb of Rekh-mi-Re at Thebes. Egyptian expe- Anschnitt, Beiheft 13: 137–157
dition, vol II. Publ. Metropolitan Museum of Art, New York, pls. Pusch E (1990) Metallverarbeitende Werkstätten der frühen
XVIII-XXII Ramessidenzeit in Qantir-Piramesse/Nord. Ägypten und Levante I:
Dies K (1967) Kupfer und Kupferlegierungen in der Technik. Springer, 75–113
Berlin Heidelberg, New York Rehren T (1999) Antike Fayenceproduktion im Kontext früher
Pyrotechnologien. In: Busz R, Gehrke P (eds) Türkis und Azur.
Dikaios P (1969) Enkomi. Excavations 1948–1958, vol IIIa. Philipp von
Minerva, Kassel, pp 122–125
Zabern, Mainz
Spano G (1858) Stele Ortuarie di Bronzo. Nuovo Bull Archaeol Sardo 4:
Fasnacht W (1989) Excavations at Ayia Varvara-Almyras. Report of the
11–15
Dept. Antiquities Cyprus, p 59–76
Stos-Gale ZA (2000) Trade in metals in the Bronze Age Mediterranean:
Gale NH (1991) Copper oxhide ingots: their origin and their place in the
an overview of lead isotope data for provenance studies. In: Pare
Bronze Age metals trade in the Mediterranean. In: Gale NH (ed)
CFE (ed) Metals make the world go round. Oxbow Books, Oxford,
Bronze Age Trade in the Mediterranean. Studies in Mediterranean pp 56–69
Archaeology XC. Jonsered, p 197–239
Stos-Gale ZA, Gale NH, Bass G, Pulak C, Galili E, Sharvit J (1998) The
Hansen M, Anderko K (1958) The constitution of binary alloys. McGraw copper and tin ingots of the Late Bronze Age Mediterranean: new
Hill, New York scientific evidence. In: Proceedings of the Fourth International
Hauptmann A (1987) Kupfer und Bronzen der südostarabischen Conference on the Beginning of the Use of Metals and Alloys,
Halbinsel. Der Anschnitt 5–6:209–218 May 25–27, 1998, Kunibiki Messe, Shimane, Japan. Sendai:
Hauptmann A, Maddin R (2005) Die Kupferbarren von Uluburun, Teil I: Japan Institute of Metals, p 115–126
Qualitätsmetall für den Weltmarkt? In: Yalcin Ü, Pulak C, Slotta R Tylecote RF (1982) The Late Bronze Age: copper and bronze
(eds) Das Schiff von Uluburun – Welthandel vor 3000 Jahren. metallurgy at Enkomi and Kition. In: Muhly JD, Maddin R,
Ausstellungskatalog (= Veröffentlichungen aus dem Deutschen Karageorghis V (eds) Acta of the International
Bergbau-Museum Bochum Nr. 138), p141-147 Archaeological Symposium BEarly Metallurgy in Cyprus,
Hauptmann A, Maddin R, Prange M (2002) On the texture and compo- 4000–500 BC. Nicosia^, p 81–100
sition of copper and tin ingots excavated from the shipwreck of Tylecote RF, Bachmann HG (1990) Metallographic notes on some
Uluburun. Bull Am Sch Orient Res 328:1–30 ingots from the Arabah and Sinai. In: Rothenberg B (ed) The
Karageorghis V, Kassianidou V (1999) Metalworking and recycling in ancient metallurgy of copper. Researches in the Arabah
late bronze age cyprus - the evidence from kition. Oxf J Archaeol 18: 1959–1984, vol 2. Institute of Archaeo-Metallurgical Studies,
171–188 London, pp 74–77
Lagarce J (1986) The production of copper ingots at Ras Ibn Hani [Syria]. Tyler DE, Black T (1990) Introduction to copper and copper alloys. In:
Acta Praehist Archaeol 18:85–90 Tyler L (ed) Metals handbook, vol 2. American Society for Metals
Lagarce J, Lagarce E, Bouni A, Salibi N (1983) Stone oxhide mould at International, Cleveland, pp 216–240
Ras Ibn Hani. Comptes-Rendu Academie des Inscriptions Belles- Zwicker U (1984) Metallographische und analytische Untersuchungen an
Lettres 127:249–280 Proben aus den Grabungen der Bronzegießerei in der Phidias-
Archaeol Anthropol Sci

Werkstatt von Olymp und Versuche zum Schmelzen von Bronze in Craddock PT (ed) Studies in Early Mining and Extractive
flachen Tiegeln. Berliner Beiträge zur Archäometrie 9:61–94 Metallurgy., British Museum Occasional Papers 20: 135–163.
Zwicker U (1986) Ancient metallurgical methods for copper production Zwicker U, Greiner H, Hofmann KH, Reithinger M (1985) Smelting,
in Cyprus, part I: oxide, sulphate and silicate ore. Bull Cyprus Assoc refining and alloying of copper and copper alloys in crucible fur-
Geol Min Eng 3:79–91 naces during prehistoric up to Roman times. In: Craddock PT,
Zwicker U, Viridis P, Ceruti ML (1980) Investigations on copper ore, Hughes MJ (eds) Furnaces and Smelting Technology in Antiquity.
prehistoric copper slag and copper ingots from Sardinia. In: British Museum Occasional Papers 48: 103–115