Content uploaded by Laura Chiarantini
Author content
All content in this area was uploaded by Laura Chiarantini on Dec 13, 2017
Content may be subject to copyright.
THE EVOLUTION OF PRE-ISLAMIC SOUTH ARABIAN
COINAGE: A METALLURGICAL ANALYSIS OF COINS
EXCAVATED IN SUMHURAM (KHOR-RORI,
SULTANATE OF OMAN)*
L. CHIARANTINI1,2† and M. BENVENUTI1
1Department of Earth Sciences, University of Florence, Via G. La Pira 4, 50121 Florence, Italy
2CNR-Istituto di Geoscienze e Georisorse, Via Moruzzi 1, 56124, Pisa, Italy
The present paper deals with compositional and microstructural features of 26 pre-Islamic,
South Arabian coins recently unearthed during archaeological excavations. Most of the
investigated coins come from Sumhuram (Khor Rori, southern Oman), and were minted by the
Hadramawt kingdom (fourth century BC to third century AD); only a few of them belong to
the Himyarite kingdom’s coinage (first to fourth centuries AD). In addition, some coins of both
the Hadramawt and the Himyarite kingdoms found at Qani’ (B’ir ‘Ali, Republic of Yemen)
have been analysed for comparison. Our main focus was to provide new hints towards the
comprehension of the chronological evolution in South Arabian coinage in terms of both metal
composition and minting techniques. In addition, some melting crucibles found at Sumhuram
have been examined in an attempt to make a comparison with the coins’ composition and to
test the hypothesis that they were used for minting operations.
KEYWORDS: SOUTH ARABIAN COINS, HADRAMAWT KINGDOM, HIMYARITE KINGDOM,
SUMHURAM, COMPOSITION, MICROSTRUCTURES
INTRODUCTION
Relatively few metallurgical studies have been specifically devoted to ancient (pre-Islamic) South
Arabian coinage. Among them, we can mention Grave et al.’s (1996) PIXE/PIGME study of
pre-Islamic silver and billon coins, and two papers dedicated to the analysis of the numismatic
collection of the Staatliches Museum für Völkerkunde, Munich (Germany)—one by Munro-Hay
(2003), who catalogued coins of ‘Arabia Felix’, mostly on a stylistic basis; and the other by
Kirfel et al. (2011), who carried out non-destructive, neutron diffraction analyses of about 100
old South Arabian coins.
Unlike these studies, which were mainly focused on the analysis of coins from museum
collections (sometimes even of uncertain authenticity), the present paper deals with composi-
tional and microstructural features of 26 pieces of South Arabian coinage recently unearthed
during archaeological excavations. All our coins are surely original and (perhaps with one
exception) have surely circulated in ancient Arabia. Following Sedov’s (2005, 2011) classifica-
tion, most of the investigated coins are Hadramawt and only a few of them belong to Himyarite
series. Seventeen coins come from excavations in Sumhuram, while nine samples were found at
B’ir ‘Ali settlement (ancient Qani’).
In particular, Sumhuram is one of the few sites in south-western Arabia where archaeological
excavations have unearthed numerous pre-Islamic South Arabian coins (about 950 in number),
*Received 23 October 2012; accepted 15 March 2013
†Corresponding author: email laura.chiarantini@unifi.it
bs_bs_banner
Archaeometry 56, 4 (2014) 625–650 doi: 10.1111/arcm.12036
© 2013 University of Oxford
mainly belonging to the Hadramawt and Himyarite series. Moreover, in the north-western sector
of the site, Albright (1982) reported the discovery of 1259 ‘coin blanks’, in what he interpreted
as a ‘mint-room’. Unfortunately, all these pieces have become lost and are not available for a
more detailed analysis (Sedov 2002).
The ancient city of Sumhuram was the most important pre-Islamic settlement of the Dhofar
region (southern Oman). The city, founded as an outpost of the Kingdom of Hadramawt, was an
important seaport within a vast international trading network, linking the Mediterranean area with
the Arabian Gulf and the Indian Ocean (Fig. 1). Its foundation dates back to the third century bc,
and it flourished until at least the late fourth to early fifth century ad (Avanzini 2002).
The American Foundation for the Study of Man (AFSM) team excavated the site in early 1950s
and 1960s. Since 1996, the Italian Mission to Oman (IMTO), a team from the University of Pisa
(Italy) headed by A. Avanzini, has been carrying out archaeological excavations at Sumhuram.
Many bronze and iron artefacts have been recovered from excavations, including tools for
everyday use (nails, chisels, hooks, needles, razors, various blades and knives, weights and bells),
Figure 1 A map of South Arabia, showing the sites featured in this paper.
626 L. Chiarantini and M. Benvenuti
© 2013 University of Oxford, Archaeometry 56, 4 (2014) 625–650
personal ornaments (bracelets, necklaces and pendants, finger-rings, ear-rings and hair-pins),
objects for cosmetics (mirrors, spoons and small rods for applying kohl), ritual items, and
decorated items such as vessels, incense burners and plates with votive inscriptions and plaques
with human and animal representations (Albright 1982; Avanzini and Sedov 2005; Chiavari et al.
2011).
At Sumhuram, only scanty traces of in-situ metallurgical activity have been found so far
(Chiarantini et al. 2007). They consist of about 50 kg of slag and a small smithing hearth
referable to ironworking activity, whereas local copper/bronze production may be indicated by
the recovery of numerous melting crucibles. The lack of any clear evidence of primary smelting
operations would thus suggest that the Sumhuram inhabitants imported raw metals (copper/iron)
for local production from elsewhere, through commercial exchanges (Chiarantini et al. 2007).
The ancient Qani’ was the main Hadramawt kingdom seaport, located on the southern coast of
modern Yemen. The kingdom, which spread along the south coast of Arabian Peninsula, through
the modern countries of Yemen and Oman, was independent until the third century ad, when it
was conquered by the neighbouring Himyarite kingdom (De Maigret 1996).
We have analysed the evolution of SouthArabian coinage over time, especially the Hadramawt
series, in terms of both metal composition and minting techniques. In addition, the textural and
compositional features of some melting crucibles found at Sumhuram have been examined in an
attempt to make a comparison with the composition of the coins and to test the hypothesis that
they were used for minting operations.
MATERIALS AND METHODOLOGIES
The main features of the 26 coins that we have analysed, listed in a chronological order according
to Sedov’s (2005) classification, are reported in Figure 2 and in Tables 1–3. A total of 21 coins
belong to Hadramawt stylistic series (types 1.1, 1.2, 3.0, 4.0, 5.2, 5.3 and 10.0, following Sedov
2005, 2011), and are dated from the fourth to second centuries bc to the first to third centuries ad,
while five other coins are Himyarite (series with two heads, and a small crude series with a
‘bucranium’). Their chronology may possibly fall in the time interval between the first and
the fourth century ad (Robin and Ba¯faqı¯h 1981; Munro-Hay 1991; Robin 1991; Davidde 1992;
Sedov 2011).
All of the coins were preliminarily analysed for their macroscopic features (colour, alteration
patinas, porosity, size, weight etc.). Then they were cut into two halves, one preserved for future
reference and the other sent to the laboratory for analysis. The mineralogy was determined by
optical and electronic (SEM) microscopy; etching by means of alcoholic ferric chloride was
employed for microstructural analysis.
The bulk composition of the coins (Table 2) was determined as the mean value of two or three
semi-quantitative SEM/EDS raster analyses, each covering an area of approximate 6–7 mm2
(corresponding to the entire cross-section of each coin). In Table 2, we have reported the
calculated (arithmetic) mean and standard deviation of the analysed elements. It can be seen that,
notwithstanding the variation in chemical composition of many coins, the (relative) precision of
the method seems to be adequate. If we calculate the coefficient of variation (CV) of the coins’
average bulk lead contents, it ranges between 0.02 and 0.48 (i.e., constantly below 0.5), with a
mean value of 0.21. In addition, if we consider coins with mean Pb contents higher than 20 wt%,
their mean CV is significantly lower (0.11). The semi-quantitative composition of individual
phases was determined by SEM/EDS (Table 3). In particular, the composition of the main
metallic phases (i.e., copper or the a-phase of the Cu–Sn system) has been determined from five
Metallurgical analysis of coins excavated in Sumhuram (Khor-Rori, Oman) 627
© 2013 University of Oxford, Archaeometry 56, 4 (2014) 625–650
Figure 2 The macroscopic features of the analysed coins.
628 L. Chiarantini and M. Benvenuti
© 2013 University of Oxford, Archaeometry 56, 4 (2014) 625–650
Figure 2—continued.
Metallurgical analysis of coins excavated in Sumhuram (Khor-Rori, Oman) 629
© 2013 University of Oxford, Archaeometry 56, 4 (2014) 625–650
Table 1 A summary of the mineralogical and textural features of the analysed coins (S, samples from Sumhuram; Q, samples from Qani’)
Provenance Sample Type of alloy Additional phases Weathering phases Microstructure
Hadramawt stylistic series
Type 1.1 SCO103 Cu–Sn–Pb Cuprite, malachite, cotunnite Annealed twinned grains
Type 1.2 SCO330 Cu–Pb Cuprite, malachite, cotunnite From dendritic to equiaxed
Type 3.0 SCO336 Cu Cu–Fe sulphides; metallic Pb, Ag Cuprite, malachite Distorted twinned grains
Type 3.0 SCO106 Cu Cu–Fe sulphides Cuprite, malachite Distorted twinned grains
Type 3.0 SCO107 Cu Cu–Fe sulphides Cuprite, malachite Distorted twinned grains
Type 3.0 QT3QCu Cu–Fe sulphides; metallic Fe Cuprite, malachite Distorted twinned grains
Type 4.0 SCO354 Cu–Sn–Pb Traces of SnO2Cuprite, malachite, cotunnite From dendritic to pseudo-dendritic
Type 4.0 SCO92 Cu–Sn–Pb Cuprite, malachite, cotunnite From dendritic to pseudo-dendritic
Type 4.0 SCO82 Cu–Sn–Pb Cuprite, malachite, cotunnite From dendritic to pseudo-dendritic
Type 4.0 QT4QCu–Sn–Pb Cuprite, malachite, cotunnite From dendritic to pseudo-dendritic
Type 5.2 Q T5.2 Qa Cu Cu–Fe sulphides; metallic Pb, Ag Cuprite, cotunnite Distorted twinned grains
Type 5.2 Q T5.2 Qb Cu Cu–Fe sulphides; metallic Pb, Ag Cuprite, cotunnite Distorted twinned grains
Type 5.3 SCO332 Cu–Sn–Pb Cuprite, malachite, cotunnite, anglesite Cored dendritic
Type 5.3 SCO122 Cu–Sn–Pb Cuprite, malachite, cotunnite, anglesite Pseudo-dendritic with strain lines
Type 5.3 SCO76 Cu–Sn–Pb Cu crystals Cuprite, malachite, cotunnite, anglesite Pseudo-dendritic distorted
Type 5.3 Q T5.3 Q Cu–Sn–Pb Cuprite, cotunnite Pseudo-dendritic with strain lines
Type 10.0 SCO100 Cu Cu–Fe sulphides, metallic Ag Cuprite, malachite Distorted twinned grains
Type 10.0 SCO90 Cu–Sn–Pb Cu sulphides, metallic Ag Cuprite, malachite, cotunnite Cored dendritic
Type 10.0 SCO348 Cu–Sn–Pb Cuprite, malachite, cotunnite, litharge Pseudo-dendritic distorted
Type 10.0 Q T10 Q1 Cu–Sn–Pb Cu crystals Cuprite, malachite, cotunnite, litharge Pseudo-dendritic distorted
Type 10.0 Q T10 Q3 Cu–Sn–Pb Cuprite, malachite, cotunnite, litharge Pseudo-dendritic distorted
Himyarite stylistic series
two heads S CO355 Cu–Sn–Ag Cuprite, malachite Distorted twinned grains
‘bucranium’ S CO147 Cu–Sn–Pb Cu–Fe sulphides Cuprite, malachite, cotunnite Distorted twinned grains
‘bucranium’ S CO217 Cu–Sn–Pb Cu–Fe sulphides Cuprite, malachite, cotunnite Distorted twinned grains
‘bucranium’ Q Bu Q1 Cu–Sn–Pb Cu–Fe sulphides Cuprite, malachite, cotunnite Distorted twinned grains
‘bucranium’ Q Bu Q2 Cu–Sn–Pb Cu–Fe sulphides Cuprite, malachite, cotunnite Distorted twinned grains
630 L. Chiarantini and M. Benvenuti
© 2013 University of Oxford, Archaeometry 56, 4 (2014) 625–650
Table 2 The bulk mean composition of the analysed coins: data are determined as mean values of semi-quantitatively SEM/EDS raster analyses (two or three large
rasters of about 6–7 mm2have been performed on each analysed fragment) (S, samples from Sumhuram; Q, samples from Qani’)
Provenance Sample Bulk mean composition wt % (SEM/EDS raster analyses)
N/sCu Sn Pb Ni Fe Co Ag As S Cl Ca Mg Si
Hadramawt series
Type 1.1 S CO103 2 63.4 7.4 24.0 5.2
s1.9 2.0 2.3 0.1
Type 1.2 S CO330 2 78.0 16.3 1.8 2.6 0.3 1.0
s3.6 3.1 0.04 0.4 0.2 0.04
Type 3.0 S CO336 297.0 2.2 0.8
s0.2 0.1 0.1
Type 3.0 S CO106 2 98.3 1.7
s0.1 0.1
Type 3.0 S CO107 2 92.5 2.8 0.9 0.6 3.2
s0.7 0.4 0.1 0.2 1.1
Type 3.0 QT3Q3 92.9 1.1 4.4 1.4 0.2
s2.4 0.2 1.9 0.2 0.3
Type 4.0 S CO354 2 83.3 12.0 3.1 1.1 0.5
s1.8 0.5 1.0 0.1 0.1
Type 4.0 S CO92 4 80.7 8.3 7.9 1.1 2.0
s2.2 0.4 1.5 0.1 0.4
Type 4.0 S CO82 3 52.2 5.5 34.0 0.7 0.4 0.2 7.0
s8.9 0.7 8.5 0.2 0.4 0.2 1.6
Type 4.0 QT4Q4 73.9 18.7 2.9 0.7 0.3 1.7 0.1 1.7
s3.1 1.8 1.4 0.5 0.3 1.7 0.2 1.2
Type 5.2 Q T5.2 Qa 2 94.4 0.5 2.1 1.5 1.1 0.4
s1.1 0.7 0.7 0.04 1.0 0.5
Type 5.2 Q T5.2 Qb 2 99.3 0.5 0.1 0.1
s0.1 0.2 0.1 0.1
Type 5.3 S CO332 2 56.7 6.7 30.3 5.9 0.4
s2.6 0.04 2.0 0.5 0.04
Metallurgical analysis of coins excavated in Sumhuram (Khor-Rori, Oman) 631
© 2013 University of Oxford, Archaeometry 56, 4 (2014) 625–650
Table 2 Continued
Provenance Sample Bulk mean composition wt % (SEM/EDS raster analyses)
N/sCu Sn Pb Ni Fe Co Ag As S Cl Ca Mg Si
Type 5.3 S CO122 2 46.5 2.0 41.9 0.5 9.1
s1.1 0.1 0.9 0.1 0.1
Type 5.3 S CO76 350.1 8.6 35.1 6.2
s8.1 1.5 9.2 0.5
Type 5.3 Q T5.3 Q 2 68.7 12.8 18.3 0.2
s2.8 0.1 1.9 0.3
Type 10.0 S CO100 3 96.0 1.0 1.0 1.0 1.0
s0.3 0.5 0.1 0.2 0.3
Type 10.0 S CO90 2 53.8 2.7 39.7 3.0 0.8
s2.7 0.1 1.9 0.5 0.2
Type 10.0 S CO348 3 72.1 8.0 17.5 2.4
s4.7 0.6 3.2 2.1
Type 10.0 Q T10 Q1 2 73.3 13.3 12.5 0.4 0.3 0.2
s3.8 0.3 3.6 0.6 0.4 0.3
Type 10.0 Q T10 Q3 2 62.6 9.2 26.3 1.1 0.8
s0.9 0.7 1.7 0.1 0.04
Himyarite series
two heads S CO355 287.3 10.5 1.3 0.3 0.6
s1.0 0.7 0.5 0.2 0.1
‘bucranium’ S CO147 366.1 8.7 20.5 1.4 3.3
s2.9 1.3 1.7 0.1 0.5
‘bucranium’ S CO217 279.0 9.0 10.3 1.0 0.7
s2.2 0.5 2.8 0.1 0.1
‘bucranium’ Q Bu Q1 2 93.0 3.5 2.4 1.1
s0.9 0.2 0.9 0.2
‘bucranium’ Q Bu Q2 2 80.6 7.2 11.1 0.8 0.1 0.2
s4.9 0.3 5.0 0.1 0.2 0.3
632 L. Chiarantini and M. Benvenuti
© 2013 University of Oxford, Archaeometry 56, 4 (2014) 625–650
Table 3 The mean composition of the main metallic phases identified in coins: data are generally reported as mean
values of five semi-quantitative (SEM/EDS) point analyses—data for starred samples were obtained by EMPA
quantitative analyses (S, samples from Sumhuram; Q, samples from Qani’)
Provenance Sample N/sComposition of main metallic phase wt% (SEM/EDS-EMPA*)
Cu Sn Ni Fe Co Ag As S Sb
Hadramawt series
Type 1.1 S CO103 5 90.5 9.5
s0.3 0.3
Type 1.2 S CO330 5 96.8 0.4 2.8
s0.5 0.5 0.3
Type 3.0 S CO336* 10 96.92 1.22 0.11 0.21 0.02 0.51 0.09
s0.94 0.33 0.04 0.07 0.04 0.07 0.22
Type 3.0 S CO106 5 98.5 1.5
s0.1 0.1
Type 3.0 S CO107 5 96.4 2.4 0.5 0.8
s0.4 0.2 0.1 0.1
Type 3.0 QT3Q5 96.9 0.9 2.2
s2.3 0.6 1.7
Type 4.0 S CO354* 10 87.6 8.18 1.1 0.5 0.2 0.95
s2.71 2.69 0.14 0.12 0.04 0.36
Type 4.0 S CO92 5 94.2 4.6 1.2
s0.4 0.3 0.2
Type 4.0 S CO82 5 91.8 6.5 1.1 0.7
s6.1 6.4 0.0 0.4
Type 4.0 QT4Q5 90.2 9.1 0.5 0.1
s2.4 2.4 0.3 0.2
Type 5.2 Q T5.2 Qa 5 96.5 1.3 2.1 0.1
s0.6 0.3 0.3 0.2
Type 5.2 Q T5.2 Qb 5 99.2 0.7 0.2
s0.3 0.2 0.3
Type 5.3 S CO332 5 90.9 9.1
s0.9 0.6
Type 5.3 S CO122 5 97.9 1.5 0.3 0.4
s1.4 1.6 0.2 0.3
Type 5.3 S CO76 5 95.4 4.6
s0.7 0.7
Type 5.3 Q T5.3 Q 5 88.2 11.8
s1.0 1.0
Type 10.0 S CO100 5.0 96.3 1.8 1.0 1.0
s0.3 0.2 0.1 0.1
Type 10.0 S CO90 5 95.7 3.7 0.7
s1.7 2.1 0.4
Type 10.0 S CO348* 10 90.57 6.47 0.63 0.27 0.16 0.38 0.01
s1.48 1.58 0.04 0.05 0.03 0.07 0.02
Type 10.0 Q T10 Q1 5 88.6 11.1 0.3
s1.6 1.7 0.4
Type 10.0 Q T10 Q3 5 91.2 6.9 1.0 0.9
s1.7 2 0.1 0.4
Himyarite series
two heads S CO355* 10 88.09 9.04 0.12 1.37
s1.52 0.86 0.02 0.2
‘bucranium’ S CO147 5 91.5 7.2 1.2
s0.4 0.3 0.2
‘bucranium’ S CO217 5 91.1 7.4 0.9 0.6 0.1
s0.5 0.3 0.2 0.3 0.2
‘bucranium’ Q Bu Q1 5 95.5 3.3 1.1 0.1
s0.5 0.2 0.2 0.2
‘bucranium’ Q Bu Q2 5 92.4 5.8 1.0 0.3 0.2 0.2
s1.3 1.7 0.1 0.4 0.3 0.4
Metallurgical analysis of coins excavated in Sumhuram (Khor-Rori, Oman) 633
© 2013 University of Oxford, Archaeometry 56, 4 (2014) 625–650
spot analyses. The SEM/EDS analyses were performed at the Interdepartmental Centre for
Electron Microscopy and Microanalyses (MEMA), University of Florence, by means of a Zeiss
EVO MA15 scanning electron microprobe, equipped with an RDS detector and the Oxford INCA
250 microanalysis program. On a selected number of coins (indicated by asterisks in Table 3), the
quantitative analyses (10 spot analyses) were performed by EMPA at the CNR Institute for
Geosciences and Earth Resources of Padua (Italy). The instrument was a Cameca SX 50,
equipped with four WDS spectrometers (operating at 15 kV accelerating potential and 15 nA
sample current). Standards used for quantitative microanalyses of alloys included the following:
metallic iron (for Fe), cobalt (Co), nickel (Ni), copper (Cu), silver (Ag), sphalerite (for S and Zn),
AsGa (for As), cassiterite (Sn) and galena (Pb).
The mineralogy of the crucible fragments was determined by means of both X-ray powder
diffraction (using a Philips PW 3710 diffractometer, equipped with a Cu X-ray tube) and
SEM/EDS.
RESULTS
The macroscopic, mineralogical microstructural and compositional features of all of the analysed
coins are described in detail hereafter.
Hadramawt type 1.1 (Athena’s head/owl series)
Coin CO103 (Sumhuram) is made of a Cu–Sn–Pb alloy that is extremely rich in lead (7.4 wt%
Sn and about 24 wt% Pb: Table 2). Twinned and recrystallized grains of Cu–Sn a-phase (9.5 wt%
Sn) show a maximum size of about 20 mm; lead is mostly segregated at a-phase grain boundaries,
particularly towards the edges of the coin (Fig. 3 (a)). Strain lines (evidenced by corrosion) and
deformed grains are particularly abundant beneath the design of the coin (Fig. 3 (b)). The coin’s
microstructural features clearly indicate that this coin was annealed and struck. Coin CO103
shows some large fractures and strong intergranular corrosion, with development of Pb chlorides
(cotunnite) interstitial to the matrix grains. The surface is partially covered by patinas of cuprite
and malachite.
Figure 3 Coin CO103, type 1.1 (Athena’s head/owl series) from Sumhuram. (a) A cross-section of the coin
(backscattered electrons, BSE). Cu–Sn a-grains (grey) with abundant segregation of lead (white) along the coin edges.
(b) The microstructural features after etching with FeCl3(reflected light, RL). Recrystallized, twinned grains of Cu–Sn
a-phase; strain lines are particularly evident along the coin edges.
634 L. Chiarantini and M. Benvenuti
© 2013 University of Oxford, Archaeometry 56, 4 (2014) 625–650
Hadramawt type 1.2 (head/owl series)
Coin CO330 is made of leaded copper (16.3 wt% Pb in the bulk: Table 2 and Fig. 4). The coin
microstructure ranges from dendritic distorted to equiaxed (annealed) (Fig. 4 (b)). Copper grains
(max. width 100 mm) contain significant amounts of Fe and minor Ni (Table 3) and may be
twinned. Lead is segregated at copper grain boundaries. Calcite and quartz crystals have been
found attached to coin surfaces. In agreement with macroscopic observations, the structural
features indicate that this copper–lead coin was produced by striking.
Hadramawt type 3.0 (radiated head/winged caduceus series)
Four coins of this type have been analysed in detail, three from Sumhuram (CO336, CO106 and
CO107) and one from Qani’ (T3 Q). All pieces have small and regular flans (diameters between
1 and 1.4 cm) and may be rather thick (0.3–0.5 cm) (Figs 2 and 5 (a)). All of the coins show quite
homogeneous compositional and textural features (Tables 1 and 2): they are made of copper and
show an annealed, twinned-grain structure produced by striking (Fig. 5 (b)). Moving from the
core towards the surface, the grains become progressively more and more distorted. Microprobe
analyses of copper grains reveal the presence of Co, Fe and relatively abundant Ni (in the range
of 0.9–2.4 wt%: Table 3). Tiny (about 10 mm in size) flattened inclusions of Cu–Fe sulphides and
globules of eutectic copper–cuprite are common in all samples. Afew inclusions of metallic lead
and silver particles (about 10 mm in size) occur within CO336, while metallic droplets of the
iron-rich phase of the Cu–Fe system are scattered throughout the copper matrix of coin T3 Q. All
samples are rather well preserved, except for limited intergranular corrosion at the coin surfaces.
Hadramawt type 4.0 (head/eagle series)
The four coins analysed (three from Sumhuram and one from Qani’) are all composed of
a Cu–Sn–Pb alloy of variable composition (5.5–18.7 wt% Sn; 2.9–34 wt% Pb: Table 2).
These coins are large, microstructurally homogeneous (Fig. 2 and Table 1) and were clearly
produced by casting. They are generally porous, with cored dendritic or pseudo-dendritic micro-
Figure 4 Coin CO330, type 1.2 (head/owl series) from Sumhuram. (a) A section of the coin (BSE). Copper grains
(medium grey) with segregations of metallic lead (white). (b) Dendrites (to the left) and equiaxial grains of almost pure
copper (to the right) (etched with FeCl3, RL).
Metallurgical analysis of coins excavated in Sumhuram (Khor-Rori, Oman) 635
© 2013 University of Oxford, Archaeometry 56, 4 (2014) 625–650
structures (Figs 6 (a) and 6 (b)): these are particularly evident in coin CO82, which also displays
the highest Pb content. Dendrites of the a-phase of the Cu–Sn system are generally strongly
cored, with average Sn contents (obtained from 5–8 EMPA spot analyses) ranging from 4.6 to 9.1
wt% Sn, with significant amounts of Ni and minor Fe (Table 3). Lead, mainly altered to lead
chlorides, is preferentially segregated towards the edges of the coins and interstitial to a-phase
dendrites, but it also occurs as large, subcircular plagues (up to 200 mm in size), or in cracks
parallel to coin sides (Fig. 6 (c)).
Small plagues of a SnO2phase interstitial to a-dendrites in coin CO354 indicate partially
oxidizing conditions during the alloying process (cf., Northover and Rehren 1992; Klein and
Hauptmann 1999). Extensive surface corrosion is evidenced by thick patinas of copper carbon-
ates and lead chlorides (mainly cotunnite).
Hadramawt type 5.2 (radiated head/bull series)
The two coins of this series (T5.2 Qa and T5.2 Qb) come from Qani’ and show very similar
compositional and textural features (Tables 1 and 2). Both are characterized by irregular and
rather thick (0.3 cm) flans (Fig. 2). The coins are mostly made of twinned copper grains,
containing variable amounts of Ni, Fe, Co and Sb (Table 3), heavily deformed, particularly in the
superficial layers. A few plagues of metallic lead or silver are dispersed throughout the copper
matrix of coin T5.2 Qb. Cu–Fe sulphides occur as elongated inclusions parallel to coin sides.
At low magnification, the T5.2 Qb sample shows a dendritic structure, formed by large
dendrites (up to 400 mm in length) of cuprite within the copper matrix (Fig. 7 (a)). Etching with
ferric chloride revealed the presence of annealed and twinned copper grains (up to 50 mm) in the
interdendritic areas (Fig. 7 (b)). One possible interpretation of these microstructural features is
that this coin was struck a long while after the blank casting, in which the large dendritic
microstructure originated, was produced. As a consequence, the piece would have undergone
extensive oxidation (with the formation of copper oxides, mainly cuprite) before the coin was
finally struck. The latter process would have caused recrystallization and twinning of the uncor-
roded copper grains and the distortion of the large dendrites of cuprite. Alternatively (and more
likely), an inhomogeneous annealing of the dendritic structure during striking, with subsequent
development of twinned crystals in some parts of the coin and partial distortion of residual
Figure 5 Coin T3Q, type 3.0 (radiated head/winged caduceus series) from Qani’. (a) A cross-section of the coin (BSE),
showing a homogeneous matrix of lead-free copper (grey). (b) Twinned grains of copper and tiny inclusions of Cu–Fe
sulphides (etched with FeCl3, RL).
636 L. Chiarantini and M. Benvenuti
© 2013 University of Oxford, Archaeometry 56, 4 (2014) 625–650
Figure 6 (a) Coin CO354, type 4.0 (head/eagle series) from Sumhuram; a cross-section of the coin (BSE), showing the
outer thick corrosion patinas of copper carbonates (medium grey) and lead chlorides (white). The sample is composed
of Cu–Sn alloy (light grey), with variable amounts of lead (white) segregated towards the edges. (b) Coin CO354; large
pores (black) occupy interdendritic areas. Cu–Sn a-phase dendrites are distinctly cored (etched with FeCl3, RL). (c) Coin
CO82, type 4.0 (head/eagle series) from Sumhuram; a cross-section of the coin (BSE). Abundant lead chlorides (white
and light grey) occur as large, subcircular plagues (up to 200 mm in size) amidst the Cu–Sn a-phase (dark grey), or in
cracks parallel to the coin sides.
Figure 7 Coin T5.2 Qb, type 5.2 (radiated head/bull series) from Qani’. (a) A cross-section of the coin (BSE). Large
dendrites of cuprite (black) are interspersed with a matrix of almost pure copper (light grey). (b) Twinned copper grains
occur interstitial to the large dendrites of cuprite (etched with FeCl3, RL).
Metallurgical analysis of coins excavated in Sumhuram (Khor-Rori, Oman) 637
© 2013 University of Oxford, Archaeometry 56, 4 (2014) 625–650
dendrites in others, could be invoked (D.A. Scott pers. comm.). Subsequent corrosion of the coin
may have led to the formation of cuprite and of the observed high porosity.
Hadramawt type 5.3 (radiated head/bull series)
From this series, we have analysed four coins (three from Sumhuram and one from Qani’). They
all show small central cavities on one or both sides (Fig. 2: the cavity is evident in the right-hand
picture of CO332). The four coins display some similarities with regard to bulk chemical
composition and microstructural features. They are made of a Cu–Sn–Pb alloy that is extremely
rich in lead (2–12.8 wt% Sn; 18.3–41.9 wt% Pb: Table 2). The fully dendritic (CO332) or
pseudo-dendritic microstructures seem to suggest that the coins were produced by simple casting
(Fig. 8 (b)). Dendrites are generally slightly cored and show variable tin concentrations in
different samples (from 1.5 to 11.8 wt% Sn) and trace contents of Fe and Co (Table 3). Large,
subcircular islands of metallic lead or lead chlorides (cotunnite: Fig. 8 (a) to Fig. 8 (c)) are
scattered though the Cu–Sn matrix. Some distorted dendrites (just beneath the coin surface) and
slightly deformed plagues of lead have been observed in proximity to coin surfaces. Large cracks
develop parallel to coin sides.
Figure 8 Coin CO76, type 5.3 (radiated head/bull series) from Sumhuram. (a) A cross-section of the coin (BSE). Cu–Sn
a-phase (medium grey) is associated with lead (white to light grey), which may occur either in large subspherical bubbles
or within cracks subparallel to the coin sides. Note the extreme segregation of lead in the whitish plague to the left
(arrow). (b) Partially distorted cored dendrites of the Cu–Sn a-phase with large subspherical bubbles (black) (etched
with FeCl3, RL). (c) Lead chlorides (white) segregated in the interdendritic spaces or as flattened, large plagues; they also
infill part of the large crack in the centre of the picture (BSE).
638 L. Chiarantini and M. Benvenuti
© 2013 University of Oxford, Archaeometry 56, 4 (2014) 625–650
All of the coins are deeply corroded. Alteration products are represented by abundant copper
carbonates, lead chlorides and, occasionally, sulphates (gypsum), which occur both as surface
patinas or as fracture infillings. Loss of tin and redeposition of twinned copper crystals along
cracks is fairly common (CO76: Table 1). In particular, CO76 is a very heterogeneous, heavily
corroded coin, with large cracks parallel to the coin sides (Fig. 8 (a)). Two compositionally
and structurally different areas can be recognized therein. The larger one (medium grey, on
the right-hand side of Fig. 8 (a)) is characterized by partially distorted Cu–Sn dendrites, with
abundant metallic lead and lead chlorides either segregated in the interdendritic spaces or as
flattened, large plagues; lead chlorides commonly infill some of the large cracks developed
parallel to the coin sides (Fig. 8 (c)). The smaller area (the white spot to the left of Fig. 8 (a)) is
characterized by a segregation of lead oxides with disseminated small Cu–Sn droplets and
dendrites.
As mentioned before, despite a dendritic or pseudo-denditric microstructure just beneath
the central cavity, most coins of Hadramawt type 5.3 show the presence of either strain lines
(especially CO122 and T5.3 Q) and/or deformed dendrites (CO76). This evidence, together with
partial flattening of lead islands and the formation of large and evident cracks parallel to the
coins’ sides, seems to indicate cold-working. In more detail, the small conical cavities at the
centre of one or both coins’ sides were probably produced during cold-finishing of the coins after
casting (cf., Sedov and ‘Aydarus 1995; Sedov 2005).
Hadramawt type 10.0 (s2qr/bull’s head series)
Coins belonging to Hadramawt type 10.0 are characterized by thick flans, roughly square or
rectangular in shape (Fig. 2). They were previously classified—on the basis of morphological
features only—as struck coins. Actually, the analysed items (five in total) display very hetero-
geneous microstructures and bulk compositions (Tables 1 and 2).
Coin CO90 looks very different from all other coins of this series. The equidimensional,
squared shape of this coin, with convex sides and the absence of any design, is worthy of mention.
In addition, the coin has been clearly truncated on one side (Figs 2 and 9 (a)). CO90 is a low-tin,
highly leaded bronze piece (2.7 wt% Sn; about 40 wt% Pb: see Table 2). Lead mostly occurs
as lead oxides and chlorides in the interdendritic regions and as large, subspherical islands
(Fig. 9 (a)). Particles of metallic silver and a few inclusions of copper sulphides have also been
identified. The Cu–Sn a-phase contains trace contents of Ni (Table 3) and displays a pseudo-
dendritic, undeformed structure, suggesting that the coin was cast. The abrupt truncation of
dendrites along the broken side of the coin suggests that the coin blank was obtained by breaking
a longer stick after casting (cf., Finetti 1987).
Coin CO100 (Fig. 9 (b)) is made of copper, with minor amounts of Co, Ni and Fe (Tables 2 and
3), and it shows a distorted, twinned-grain structure (Fig. 9 (c)), produced by annealing and
striking. The deformation of the copper grains is higher near the coin sides and less pronounced
in the core, where straight twinned lines are observed. Silver plagues and abundant flattened
inclusions of Cu–Fe sulphides have been observed.
The last three coins belonging to type 10.0, one from Sumhuram (CO348) and two from Qani’
(T10 Q1 and T10 Q3), display quite similar compositions and microstructures (Table 2). They are
all made of a lead-rich Cu–Sn–Pb alloy (8–13.3 wt% Sn; 12.5–26.3 wt% Pb). A well-defined
dendritic pattern is formed by strongly cored Cu–Sn a-phase dendrites, with average Sn contents
varying from 6.5 to 11.1 wt% and minor amounts of Ni, Fe and Co (Table 3). Lead is distinctly
segregated among Cu–Sn a-phase dendrites, preferentially in proximity to the edges of the coins
Metallurgical analysis of coins excavated in Sumhuram (Khor-Rori, Oman) 639
© 2013 University of Oxford, Archaeometry 56, 4 (2014) 625–650
(Figs 9 (d)–(f)). A few grains of pure copper were observed in T10 Q1. Distortion of dendrites is
visible near the sides of the coins, suggesting a cold striking process which, notwithstanding the
relatively high lead contents, did not cause cracking, probably due to the small size of the coins
and their thick flans. The abrupt cut of the large Cu–Sn dendrites along the edges of the coins
Figure 9 Coins belonging to type 10.0 (s2qr/bull’s head series). (a) A cross-section of coin CO90 from Sumhuram
(BSE). Note the extreme enrichment in lead, mostly as lead oxides/chlorides (white) and the truncated side on the bottom.
(b) A cross-section of coin CO100 from Sumhuram (BSE). (c) The microstructure of coin CO100 (etched with FeCl3, RL):
strongly deformed copper grains with flattened inclusions of Cu–Fe sulphides. (d) A cross-section of coin T10 Q3 from
Qani’ (BSE). Lead (white) is preferentially segregated towards the edges of the coin. (e) The microstructure of coin T10
Q3 (etched with FeCl3, RL). Lead chlorides (grey) occur interstitial to large dendrites of the Cu–Sn a-phase, which are
truncated along the edge of the coin. (f) The microstructure of coin T10 Q3 (BSE), with abundant lead chlorides (white)
interstitial to Cu–Sn a-phase dendrites.
640 L. Chiarantini and M. Benvenuti
© 2013 University of Oxford, Archaeometry 56, 4 (2014) 625–650
(Fig. 9 (e)) suggests that the molten alloy was originally poured into a large mould and the
‘sticky’ ingot thus obtained was broken to produce several coin blanks.
Hadramawt type 10.0 coins generally show only minor corrosion, mostly represented by
patinas of lead oxides and chlorides associated with copper carbonates.
Himyarite, series with two heads
Coin CO355 (from Sumhuram) is a silver-bearing, struck coin with a 1.3 wt% as bulk Ag content
(Table 2). Grains of Cu–Sn a-phase (9 wt% Sn; 1.4 wt% Ag: Table 3) develop an annealed,
twinned microstructure, with disseminated Pb plagues and scattered flakes (max. 10 mm in size)
of a probable silver-rich Cu–Ag phase (due to the small size, analytical accuracy is very low). The
Cu–Sn a-phase grains are perfectly equiaxed in the core of the coin, but they become deeply
distorted, with evident strain lines close to the surface (Fig. 10). The state of preservation of coin
CO355 is very good, with very limited development of intergranular and surface corrosion
patinas made of cuprite, malachite and, occasionally, gypsum, probably due to the burial condi-
tion within the Sumhuram soils, which are extremely enriched in both calcium sulphates and
carbonates (Platel et al. 1998).
Himyarite, small crude series with a ‘bucranium’
The four coins of this series, two from Sumhuram (CO147 and CO217) and two from Qani’ (Bu
Q1 and Bu Q2), are quite similar. They are all made of a Cu–Sn–Pb alloy (bulk contents: 3.5–9
wt% Sn, 2.4–20.5 wt% Pb) with minor Ni contents (1 wt%: Table 2). Grains of the Cu–Sn
a-phase contain about 1 wt% Ni and variable (trace) amounts of Fe and Co (Table 3). Micro-
structural evidence of striking is quite clear within the recrystallized, twinned grains of the Cu–Sn
a-phase. Both lead and lead chlorides are interstitial to Cu–Sn grains (Fig. 11 (b)), and prefer-
entially segregated towards the margins of the flans. Inclusions of Cu–Fe sulphides, often
flattened and deeply distorted, may be observed dispersed among Cu–Sn grains.
Notwithstanding the extremely small dimensions (about 9 mm in diameter and 2 mm in
thickness), coin CO147 is distinctly zoned between an inner core and an outer rim (Fig. 11 (a)).
Figure 10 CO355, from the Himyarite series with two heads (from Sumhuram). (a) A cross-section of coin with an
homogeneous, annealed microstrusture (BSE). (b) Annealed, twinned grains of the Cu–Sn–Ag phase with intergranular
corrosion; some strain lines are evidenced by intergranular corrosion near the coin surface (etched with FeCl3, RL).
Metallurgical analysis of coins excavated in Sumhuram (Khor-Rori, Oman) 641
© 2013 University of Oxford, Archaeometry 56, 4 (2014) 625–650
The zoning is essentially due to different lead concentrations, which vary from 3 wt% (core) to
23 wt% (rim). This microstructure indicates that at least some lead was extruded along coin
fractures (Fig. 11 (c)). If we cannot exclude intentional lead addition during blank casting,
perhaps to adjust the size of the coin, reminting seems highly unlikely (cf., Finetti 1987),
considering both the very small size of the coin and the identical composition of the Cu–Sn alloy
within the two layers.
DISCUSSION
Evolution of Hadramawt coinage
The analytical results allow us to point out some differences in the Hadramawt kingdom’s
coinage between the fourth century bc (early Hadramawt series, type 1.1) and the third century
ad (late Hadramawt series, type 10.0)—although the scarcity of analytical data on South Arabian
coinage, together with uncertainties on absolute chronology, do not permit an exhaustive inter-
pretation of Hadramawt coinage techniques and their evolution, and only some striking aspects
can be pointed out. First, with the notable exception of Hadramawt type 10.0, coins belonging to
the same series, independently from the site of discovery (i.e., Sumhuram or Qani’) show similar
Figure 11 Coin CO147, from the Himyarite small crude series, with a ‘bucranium’ (from Sumhuram). (a) A
cross-section of the coin (BSE). A core made of a Cu–Sn–Pb alloy (grey), very low in lead, is surrounded by an outer shell
clearly enriched in segregated lead (white). (b) The microstructural features of the outer shell, with twinned grains of the
Cu–Sn phase and interstitial lead chlorides (grey) (etched with FeCl3, RL). (c) A close-up view of the outer lead-rich shell,
with segregation of lead along small fractures (white alignments on the left of the picture) (BSE).
642 L. Chiarantini and M. Benvenuti
© 2013 University of Oxford, Archaeometry 56, 4 (2014) 625–650
microstructural features (indicating that minting techniques were quite standardized throughout
the whole Hadramawt kingdom), although compositions may vary over a relatively wide range;
the latter feature is clearly evidenced by inspection of Figure 12, where the compositional fields
of the different coinage series are reported in the Cu–Sn–Pb ternary system.
Different hypotheses can be put forward to explain the significant variations of chemical
composition evidenced in some Hadramawt coin series, including: different and/or variable
provenance and supply of metals employed for coin productions through ages; a not well-
mastered practice of alloying techniques; production—within the same numismatic series—of
coins with different values; and, finally, different sites of production (mints).
Two samples of the earliest Hadramawt series (i.e., types 1.1 and 1.2, the so-called Hadramawt
owls series) have also been investigated by Degli Esposti (2009), among other metallic items
from Sumhuram. The most ancient Hadramawt coin series analysed in this work (Hadramawt
type 1.1) is an imitation of the Athenian tetradrachm. By comparing our results (CO103: see
Tables 2 and 3) with those obtained by Degli Esposti (2009) on sample CO163 from the same
series, we can suggest that both coins were struck, as evidenced by the fully recrystallized Cu–Sn
a-grains. The high silver contents (about 15 wt% Ag) found by Degli Esposti (2009) in sample
CO163, however, were not detected in our sample CO103, which is made of a ternary Cu–Sn–Pb
alloy with a high lead content (around 24 wt%).
Hadramawt type 1.2 coin CO330 was also struck. It is made of leaded copper (16.2 wt% Pb)
but does not show detectable tin, unlike sample CO78 from the same series (containing about 6.5
wt% Sn), analysed by Degli Esposti (2009). It is not clear whether the observed compositional
differences between the two sets of samples are statistically significant: in the event that they are,
Figure 12 A Cu–Sn–Pb ternary diagram, showing the compositional fields of each coin series analysed in this paper.
Metallurgical analysis of coins excavated in Sumhuram (Khor-Rori, Oman) 643
© 2013 University of Oxford, Archaeometry 56, 4 (2014) 625–650
this could confirm the circulation of different currencies within the same numismatic series, as
inferred by Sedov (2005), who maintains that the Hadramawt owls series were minted in both
silver and bronze alloys.
Interestingly, notwithstanding the fact that they were probably minted in different periods, the
Hadramawt type 3.0 and 5.2 coins show similar compositions. In fact, these coins are made of
copper with constantly high Ni(1Co,Fe) impurities (up to 2.4 and 1.3 wt% Ni in the 3.0 and 5.2
types, respectively: Table 3). These coins contain abundant inclusions of Cu–Fe sulphides and all
appear to have been struck.
The presence of detectable (sometimes relevant) impurities of Ni(1Co,Fe) especially in types
3.0 and 5.2, and less commonly in other series (types 4.0 and 10.0), compares well with the
chemical composition of many metallic items found at Sumhuram (Chiavari et al. 2011) and, in
general, that of metal artefacts produced in the Persian Gulf (Hauptmann 1995; Weeks 1997,
2003; Prange et al. 1999; Prange and Hauptmann 2001), although the latter usually show
high As contents as well. We suggest that both Ni–Co–Fe enrichment and the occurrence of
inclusions of Cu–Fe sulphides may indicate the smelting of copper–iron sulphidic ores hosted
within the ophiolitic suites of northern Oman and Masirah Island (Weisgerber 1978, 1980, 1981;
Peters 2000), particularly the copper deposits of the lower crustal and mantle sequences of the
Semail Ofiolite (Weeks 2003), where Ni–Co sulphides (such as pentlandite (Ni,Fe)9S8and
cobaltite CoAsS) are commonly associated with chalcopyrite, pyrite and pyrrhotite (Hauptmann
et al. 1988; Begemann et al. 2010). The complete absence of any tin and lead in the alloy
composition of the type 3.0 and 5.2 coins, is, however, unique compared to all other Hadramawt
coinage.
To our knowledge, this is the first time that ‘copper-only’ coins have been detected in the whole
Hadramawt coinage. Among possible explanations for this different alloy composition, one could
invoke reduced availability of alloying metals (particularly tin) and/or different minting site(s). It
is known that the addition of alloying elements such as Pb, Sn or Ni may produce significant
technical and aesthetic changes (e.g., colour variations) to the final metal product. According to
Lechtman (1988), the addition of 1–5 wt% Ni to copper could significantly modify the mechani-
cal properties and, possibly, the colour of the metal. However, the quantities of nickel detected in
Hadramawt coins are generally lower and of the order of typical concentrations of native copper
(up to some hundreds of ppm: cf., Rapp 1981; Pernicka 1999). However, we cannot exclude the
possibility that the casting of coins devoid of any lead or tin was intentionally practised in order
to produce coins that were distinctly different from other Hadramawt series. Further analytical
and numismatic investigations could hopefully clarify this point. According to Sedov’s (2005)
classification, the coin series types 4.0, 5.2, 5.3 and 10.0 circulated approximately between
the mid-second century bc and the third century ad. With the one exception of type 5.2, the
microstructural features of coins from all these series generally indicate low-quality manufac-
turing. From a compositional point of view, they are mostly made of a ternary Cu–Sn–Pb alloy,
which is often extremely rich in lead.
Type 4.0 coins are bigger and heavier with respect to previous coinage series. Three out of four
samples have lead contents below 8 wt%, while sample CO82 contains about 34 wt% Pb (Table
2). Microstructural analysis indicates that all the type 4.0 coins were produced by simple casting
in a mould.
It is not easy to compare our data with Kirfel et al.’s (2011) neutron diffraction analyses of two
coins belonging to Hadramawt type 4.0. In fact, this non-destructive technique does not allow
us to detect the presence (if any) of tin in the investigated samples—and this is not a minor
drawback, given the relevance of this element in copper alloys! However, Kirfel et al. (2011)
644 L. Chiarantini and M. Benvenuti
© 2013 University of Oxford, Archaeometry 56, 4 (2014) 625–650
calculated that about a 2.5% phase fraction of their coins were made of metallic lead, corre-
sponding to about 7.7 wt% Pb bulk contents, in agreement with values found in some of our
samples.
All type 5.3 coins are highly enriched in lead (18.3–41.9 wt%: Table 2 and Fig. 12). After
casting, blanks were more or less cold-worked: the lack of (or insufficient) annealing produced
a slight deformation of the dendritic microstructure and the local development of strain lines. The
lead content in Hadramawt type 5.3 coins is so high that it caused mechanical failure of the alloy
and deep fracturing after cold-working. Copper alloys bearing such high lead contents are neither
easy to cast nor to work. In fact, high lead contents may produce lead pooling during solidifi-
cation of the alloy, with the subsequent formation of large bubbles of molten lead (Scott 2010),
like those observed in some of the above coins, particularly types 4 and 5.3. Moreover, highly
leaded bronzes are difficult to anneal and shape, since lead creates planes of weakness in the
hammered metal, producing the delamination of the artefact (Scott 2010) and the formation of
deep fractures, as observed in sample CO76.
The intentional addition of lead to increase the weight of coins was a common practice in
antiquity (cf., Morrisson 1981; Calliari et al. 1999; Ingo et al. 2006), reasonably related to
inflation (Finetti 1987). Cold-working of highly leaded blanks such as those of type 5.3, with the
production of intense fracturing, also favoured strong corrosion of the items. In addition, ineffi-
cient control of redox conditions during the alloying process is suggested by the formation of
SnO2crystals in coin CO354.
The Hadramawt type 10.0 coins show variable bulk composition and production techniques
(Fig. 12). Coins from this series have been found in many Hadramawt sites; they appear to have
been issued (and in circulation) over a long time period (Sedov 2005). This could possibly
explain their great heterogeneity (A. V. Sedov pers. comm.). CO100 is a unique coin, made of
almost pure copper and showing a fully annealed microstructure, obtained by hot-working. All
other coins of this series are made of a Cu–Sn–Pb alloy. Three of them (CO348, T10 Q1 and T10
Q3) are squared pieces; their flans were apparently obtained by cutting and cold striking of cast
metal sticks or bars. Cold-working caused deep deformation of the dendritic structure, although
less pronounced than in type 5.3 coins. Sample CO90 is very similar to the last three coins, except
that it bears no evidence of striking. The lack of any design on either side may suggest that it is
a coin blank.
Notwithstanding the fact that we have analysed a comparatively small number of Himyarite
coins, microstructural analysis indicates that their production technique was much more efficient
and better mastered than Hadramawt coinages. The analysed Himyarite pieces are extremely
small, thin and in a good state of preservation. A generally high level of manufacturing (by
striking) is clearly evidenced by the good quality of the coin designs. The coin from the two heads
series is distinctly enriched in silver and contains almost no lead. Coins of the small crude series
with a ‘bucranium’ are made of a Cu–Sn–Pb alloy which, although relatively enriched in lead (up
to about 20 wt%: Tables 2 and 3), show a fully recrystallized microstructure obtained by careful
annealing and striking processes.
Although working of these relatively lead-rich coins did not produce intense deformation, such
as observed in other coins (e.g., in the Hadramawt type 5.3 series), a network of small fractures,
infilled by lead extruded from the inner core, may be observed along the edges of sample CO147
(Fig. 11 (c)).
It can be argued that lead remained in supersaturated solution when the cast blank was
produced and began to segregate to the grain boundaries during the cooling of the cast blank (cf.,
Ingo et al. 2004). It should be noted that the bulk composition of our ‘bucranium type’ coins is
Metallurgical analysis of coins excavated in Sumhuram (Khor-Rori, Oman) 645
© 2013 University of Oxford, Archaeometry 56, 4 (2014) 625–650
different from that reported by Kirfel et al. (2011). According to the latter authors, in fact, coins
from this numismatic series are made of an alloy that is extremely rich in silver (from 35 to 91
wt%). On the contrary, we could not find any trace of silver in our samples. If we exclude
analytical problems or wrong classifications—both bronze and silver coins are, in fact, mentioned
for the two heads Himyarite series (Hill 1922; Sedov 2005)—we could suggest that two distinct
types of currency (within the same series) were minted, one being made of the ‘traditional’
Cu–Sn–Pb alloy and the other of the more precious Cu–Ag alloy, as already seen for Hadramawt
types 1.1 and 1.2.
Sumhuram: a mint site?
Many small bronze items found during archaeological excavations at Sumhuram (such as rings,
needles, nails etc.: see Avanzini and Sedov 2005), as well as small bronze fragments and detached
letters from an inscribed bronze vessel found on the site (Chiavari et al. 2011), indicate the
extensive use of ternary (Cu–Sn–Pb) alloys for the production of artefacts found at Sumhuram.
Up to now, however, no evidence is available for a local production of at least some of these
items. As reported in the introduction, the only evidence of ‘in-situ’ metallurgical production at
Sumhuram is represented by some 50 kg of ironworking slag and numerous fragments of ceramic
crucibles related to metal manufacturing (Chiarantini et al. 2007). Analyses of the metal remains
of some of these crucibles indicated that the latter could have been employed for copper/bronze
production, and possibly even for production of coin blanks (Chiarantini and Benvenuti 2011).
All of the crucibles found at Sumhuram have a hemispherical shape and very small dimen-
sions; many of them have pointed bases. The preserved heights range from 2.3 to 4.5 cm and the
internal diameters are about 2–3 cm. We can estimate that they contained from 4 to 10 cm3of
melted metal (Table 4). The crucible fabric was highly refractory, as evidenced by the occurrence
in the ceramic body of abundant mullite and quartz. Straw was added as a tempering agent.
Although an exhaustive classification of crucibles’ shapes, ceramic fabrics and technical
functions is difficult (cf., Bayley and Rehren 2007) and many different types of pot have been
employed through the ages for smelting, cementation, assaying and melting (cf., Rehren 2003),
the small, narrow shape and the external vitrification of the Sumhuram vessels is consistent with
heating from below, their rounded or pointed bases ensuring good stability on a heap of hot
charcoal.
Table 4 The main features of the ceramic (melting) crucibles from Sumhuram
Sample Estimated internal
diameter (cm)
Thickness (cm) Preserved
height (cm)
Main alloy
components
G11 4 0.5–0.6 4.5 Cu–Sn
G13 Not determined 0.3–0.4 3 Cu–Sn–Pb
G23 2.5 0.3–0.5 4.5 Cu–Sn–Pb
G51 1.5 0.5–0.6 3 Cu–Sn–Pb (Ni, Co)
G60 4.5 0.3 3 Cu–Sn–Pb
G71 2.5 0.4–0.7 5
G80 Not determined 0.5–0.6 2.5 Cu–Sn (Ni, S, As)
G84 2 0.6–1.2 6
G87 Not determined 0.3–0.4 2.5 Cu–Sn
G90 Not determined 0.3–0.4 3
646 L. Chiarantini and M. Benvenuti
© 2013 University of Oxford, Archaeometry 56, 4 (2014) 625–650
Crucibles employed in Roman and medieval times for copper/brass melting with very similar
shapes, although larger in size, are reported in the literature (Tylecote 1987; Bayley and Rehren
2007).
In order to test the hypothesis as to whether the crucibles could have been employed for the
casting of coin blanks, we have analysed the residues of metals found attached to some of them
and compared the results with the composition of the coins.
A few well-preserved crucibles (G71, G84 and G90: Table 4) do not contain any metal residue
and were probably never used for metallurgical purposes. Other crucibles, however, clearly show
traces of metals (Table 4). The inner vitrified portion of such vessels commonly contains droplets
of Cu–Sn or Cu–Sn–Pb alloys. Small needle crystals or aggregates of SnO2, rare magnetite and
lathy silicate crystals (melilites) have also been observed. Lead may occur both as small globules,
segregated from Cu–Sn droplets, or as a major component (up to 20–30 wt% Pb) of the glassy
groundmass.
Therefore, the composition of the crucibles’ metal residues may be mostly referred to the
ternary Cu–Sn–Pb alloy, which characterizes all the Hadramawt coinage series except for types
3.0 and 5.2 (which are made from almost pure copper).
It is, however, relevant to recall that these small crucibles were possibly employed several
times, for the melting and casting of alloys with variable compositions, making it difficult to
connect the metal residues with a specific alloy composition.
CONCLUSIONS
The large number of coins recently found in the Southern Arabia peninsula, particularly in the
ancient site of Sumhuram (Oman) and (in minor amounts) at the settlement of B’ir ‘Ali (ancient
Qani’, modern Yemen), allowed us to perform semi-destructive analyses on a certain number of
pieces of different chronology, and to characterize in detail the compositional and microstructural
features of South Arabian coinage during the existence of the Hadramawt kingdom.
Notwithstanding the few data available nowadays on earlier Hadramawt coins (types 1.1 and
1.2, fourth–second centuries bc), these imitations of Athenian tetradrachm are small, well-
manufactured pieces of Cu–Pb and Cu–Sn–Pb alloys, produced by striking.
A significant difference in the metal composition of the coins is apparently indicated by
types 3.0 and 5.2, which are made of copper with significant Ni impurities, also obtained by
striking.
Starting from the mid-second century bc and up to third century ad, coin manufacturing in the
Hadramawt kingdom was characterized by a definitely lower quality of production. Coin types
4.0, 5.3 and 10.0 are highly leaded Cu–Sn–Pb alloys produced by casting (type 4), by casting plus
cold-working (type 5.3) or cast in bars and then cut and cold-struck (type 10.0).
Unlike these later Hadramawt coinages, the coins produced by the Himyarite kingdom are small,
well-manufactured pieces, obtained by striking blanks mostly made of Cu–Sn–Ag–Pb alloys.
We tested Albright’s (1982) hypothesis—proposed after the discovery of many coin blanks
(unfortunately now lost: cf., Sedov 2002)—that Sumhuram was a mint site. In particular, we have
analysed metal residues still adhering to the inner surfaces of crucibles excavated at Sumhuram
by IMTO. The residues mostly belong to the Cu–Sn–Pb ternary system, which also characterizes
most Hadramawt coins (with the notable exceptions of types 3.0 and 5.2), as well as many small
items found at Sumhuram. Therefore, we cannot exclude the possibility that these crucibles were
at least partly used for coinage production, although we have no conclusive evidence. More
interesting is the microstructural analysis of coin CO90 (type 10.0), which could be interpreted
Metallurgical analysis of coins excavated in Sumhuram (Khor-Rori, Oman) 647
© 2013 University of Oxford, Archaeometry 56, 4 (2014) 625–650
as a coin blank, thus reinforcing the hypothesis that Sumhuram, perhaps during the late period of
the Hadramawt kingdom, could have been a mint site.
ACKNOWLEDGEMENTS
This study has been carried out in the framework of a collaboration project with the IMTO team
from the University of Pisa, headed by Professor Alessandra Avanzini. The authors are deeply
indebted to her and her colleagues, particularly Alexia Pavan, for inviting us to join the research
team at Sumhuram and for all the scientific and logistic support during the fieldwork in 2008 and
2009.
The selection of coins from both Sumhuram and Qani’ was made in co-operation with
Alexander V. Sedov, director of the Oriental National Museum of Moscow, who is also thanked
for profitable discussions of numismatic interpretations of our results.
We thank Mario Paolieri for assistance with the SEM/EDS analysis at the MEMA laboratories
in Florence.
Some of the analytical results presented in this paper were obtained by Eleonora Bertini,
Valentina Cilio, Elisabetta La Porta and Beatrice Tasselli, in the course of their thesis
dissertations at the University of Florence (Italy).
Finally, we thank David A. Scott for profitable discussions about microstructural interpreta-
tions, and two anonymous referees for their constructive reviews.
Financial support for this project was provided by an Italian Ministry for Instruction, Univer-
sity and Scientific Research (MIUR) grant (PRIN 2007).
REFERENCES
Albright, F. P., 1982, The American archaeological expedition in Dhofar, Oman, 1952–1953, American Foundation for
the Study of Man, Philadelphia, PA.
Avanzini, A., 2002, The history of the Khor Rori area. New perspectives, in Khor Rori report 1 (ed. A. Avanzini), 13–27,
Edizioni Plus, Pisa.
Avanzini, A., and Sedov, A. V., 2005, The stratigraphy of Sumhuram: new evidence, in Proceedings of the Seminar for
Arabian Studies, 35th meeting (ed. M. C. A. Macdonald), 11–17, London.
Bayley, J., and Rehren, Th., 2007, Towards a functional and typological classification of crucibles, in Metals and mines,
studies in archaeometallurgy: selected conference papers from the conference Metallurgy: a Touchstone for Cross-
cultural Interaction held at the British Museum 28–30 April 2005 to celebrate the career of Paul Craddock during
his 40 years at the British Museum (eds. S. La Niece, D. Hook and P. Craddock), 46–55, Archetype, London.
Begemann, F., Hauptmann, A., Schmitt-Strecker, S., and Weisgerber, G., 2010, Lead isotope and chemical signature of
copper from Oman and its occurrence in Mesopotamia and sites on the Arabian Gulf coast, Arabian Archaeology and
Epigraphy,21, 135–69.
Calliari, I., Magrini, M., Zambon, A., Guerriero, P., and Martini, R., 1999, Microstructural and compositional charac-
terisation of Roman coins, X-Ray Spectrometry,28, 86–90.
Chiarantini, L., and Benvenuti, M., 2011, Sumhuram as centre of production: the metal working, in Along the aroma and
spice routes. The harbor of Sumhuram, its territory and the trade between the Mediterranean, Arabia and India (ed.
A. Avanzini), 68–80, Bandecchi & Vivaldi, Pontedera.
Chiarantini, L., Benvenuti, M., Costagliola, P., and Avanzini, A., 2007, Preliminary investigation of pre-Islamic
metallurgical slags from Sumhuram, Dhofar region (Oman), in Proceedings of 2nd International Conference
Archaeometallurgy in Europe, CD-ROM file 173, Grado, Italy.
Chiavari, C., Degli Esposti, M., Garagnani, G. L., Martini, C., and Ospitali, F., 2011, Ancient metallurgy at Sumhuram
(Sultanate of Oman): technical aspects of raised inscriptions on South Arabian bronzes, Archaeometry,53, 528–46.
Davidde, B., 1992, Le monete di ‘Amdân Bayyin Yuhaqbid rinvenute nelle tombe di Haribat Al-Ahgar, presso Waragah
(Dama¯r), Yemen, in Studi archeologici, storici e filosofici sull’Arabia meridionale, vol. 1 (ed. A. de Maigret), 41–54,
IsMEO, Rome.
648 L. Chiarantini and M. Benvenuti
© 2013 University of Oxford, Archaeometry 56, 4 (2014) 625–650
De Maigret, A., 1996, Arabia Felix, Rusconi, Milan.
Degli Esposti, M., 2009, Antica metallurgia a Sumhuram (Sultanato dell’Oman). Indagini archeometriche per la
ricostruzione del quadro tecnologico, Unpublished thesis, University of Bologna.
Finetti, A., 1987, Numismatica e tecnologia, La Nuova Italia Scientifica, Rome.
Grave, P., Bird, R., and Potts, D. T., 1996, A trial PIXE/PIGME analysis of pre-Islamic Arabian coinage, Arabian
Archaeology and Epigraphy,7, 75–81.
Hauptmann, A., 1995, Chemische Zusammensetzung von Metallobjekten aus der Siedlung Umm an-Nar, in The island
of Umm an-Nar 2: the third millennium settlement (ed. K. Frifelt), 246–8, Jutland Archaeological Society Publications
26/2, Aarhus.
Hauptmann, A., Weisgerber, G., and Bachmann, H. G., 1988, Early copper metallurgy in Oman, in The beginning of the
use of metals and alloys (ed. R. Maddin ), 34–51, The MIT Press, Cambridge, MA.
Hill, G. F., 1922, Catalogue of the Greek coins of Arabia, Mesopotamia and Persia, Trustees of the British Museum,
London.
Ingo, G. M., De Caro, T., Padeletti, G., and Chiozzini, G., 2004, Microchemical investigation on Renaissance coins
minted at Gubbio (central Italy), Applied Physics A,79, 319–25.
Ingo, G. M., De Caro, T., Riccucci, C., and Khosroff, S., 2006, Uncommon corrosion phenomena of archaeological
bronze alloys, Applied Physics A,83, 581–8.
Kirfel, A., Kockelmann, W., and Yule, P., 2011, Non-destructive chemical analysis of old South Arabian coins, fourth
century bce to third century ce,Archaeometry,53, 930–49.
Klein, S., and Hauptmann, A., 1999, Iron Age leaded tin bronzes from Khirbet edh-Dharih, Jordan, Journal of Archaeo-
logical Science,26, 1075–82.
Lechtman, H., 1988, Traditions and styles in central Andean metalworking, in The beginning of the use of metals and
alloys (ed. R. Maddin), 344–78, The MIT Press, Cambridge, MA.
Morrisson, C., 1981, Monnaies en plomb Byzantines de la fin du VIeet du début VIIesiècle, Rivista Italiana di
Numismatica,83, 119–32.
Munro-Hay, S. C. H., 1991, The coinage of Shabwa (Hadhramawt), and other ancient South Arabian coinage of the
National Museum, Aden, Syria,68, 393–418.
Munro-Hay, S. C. H., 2003, Coinage of Arabia Felix: the Pre-Islamic coinage of the Yemen, Nomismata 5, Edizioni
ennerre, Milan.
Northover, J. P., and Rehren, Th., 1992, The oxidation of bronze, in Abstracts volume of 28th International Symposium
on Archaeometry (eds. P. Meyers, D. A. Scott and T. K. Earle), 75, Los Angeles, CA.
Pernicka, E., 1999, Trace element fingerprinting of ancient copper: a guide to technology or provenance? In Metals in
antiquity (eds. M. M. Pollard, P. Budd and R. A. Ixer), 163–71, BAR International Series 792, Archaeopress, Oxford.
Peters, T., 2000, Formation and evolution of the western Indian Ocean as evidenced by Masirah ophiolite: a review,
525–36, Geological Society of America, Special Paper.
Platel, J. P., Berthiaux, A., Roger, J., Chevel, S., and Ishag Khalifa, M., 1998, Geological map of Salalah sheet NE 40-9D
scale 1:100,000, Directorate General of Minerals Oman Ministry of Petroleum and Minerals, Muscat.
Prange, M., and Hauptmann, A., 2001, The chemical composition of bronze objects from ‘Ibri/Selme, in The metal hoard
from ‘Ibri/Selme, Sultanate of Oman (eds. P. Yule and G. Weisgerber), 75–84, Prähistorischer BronzefundeAbteilung
XX, Band 7, Franz Steiner Verlag, Stuttgart.
Prange, M. K., Gotze, H. J., and Weisgerber, G., 1999, Is it Oman the ancient Magan? Analytical studies of copper from
Oman, in Metals in antiquity (eds. M. M. Pollard, P. Budd and R. A. Ixer), 187–92, BAR International Series 792,
Archaeopress, Oxford.
Rapp, G., Jr, 1981, Native copper and the beginning of smelting: chemical studies, in Early metallurgy of Cyprus
4000–500 BC (eds J. D. Muhly, R. Maddin and V. Karageorghis), 33–8, Pierides Foundation, Larnaca, Cyprus.
Rehren, Th., 2003, Crucibles as reaction vessels in ancient metallurgy, in Mining and metal production through the ages
(eds. P. Craddok and J. Lang), 207–15, British Museum Press, London.
Robin, Chr., 1991, ‘Amdân Bayyin Yuhaqbid, roi de Saba’ et de dhu-Raydan, in Etudes sud-arabes. Recueil offert à
Jacques Ryckmans, 167–205, Peeters, Louvain-la-Neuve.
Robin, Chr., and Ba¯faqı¯h, M., 1981, Deux nouvelles inscriptions de Radma¯ n datant du IIe siècle de l’ère chrétienne,
Rayda¯n,4, 67–90.
Scott, D. A., 2010, Ancient metals: microstructure and metallurgy, vol. 1, Conservation Science Press, LosAngeles, CA.
Sedov, A. V., 2002, The coins from Sumhuram: the 1997–2000 seasons, in Khor Rori report 1 (ed. A. Avanzini), 249–70,
Edizioni Plus, Pisa.
Sedov, A. V., 2005, Hadramawt coinage: its sequence and chronology, in Archäologische Berichte aus dem Yemen X,
161–74, Verlag Philipp von Zabern, Mainz.
Metallurgical analysis of coins excavated in Sumhuram (Khor-Rori, Oman) 649
© 2013 University of Oxford, Archaeometry 56, 4 (2014) 625–650
Sedov, A. V., 2011, The coins from Sumhuram: the 2001A–2004A seasons, in A port in Arabia between Rome and the
Indian Ocean (3rd c. BC – 5th c. AD): Khor Rori report 2 (ed. A. Avanzini), 277–316, L’Erma di Bretschneider, Rome.
Sedov, A. V., and ‘Aydarus, U., 1995, The coinage of ancient Hadramawt: the pre-Islamic coins in the al-Mukallâ
Museum, Arabian Archaeology and Epigraphy,6, 15–60.
Tylecote, R. F., 1987, The early history of metallurgy in Europe, Archaeologist Series, Longman, London.
Weeks, L. R., 1997, Prehistoric metallurgy at Tell Abraq, United Arab Emirates, Arabian Archaeology and Epigraphy,8,
11–85.
Weeks, L. R., 2003, Early metallurgy of the Persian Gulf: technology, trade, and the Bronze Age world, E. J. Brill,
Boston.
Weisgerber, G., 1978, Evidence of ancient mining sites in Oman: a preliminary report, Journal of Oman Studies,4, 15–28.
Weisgerber, G., 1980, ‘. . . und Kupfer in Oman’—Das Oman-Projekt des Deutschen Bergbau-Museums, Der Anschnitt,
32, 62–110.
Weisgerber, G., 1981, Mehr als Kupfer in Oman—Ergebnisse der Expedition 1981, Der Anschnitt,33, 174–263.
650 L. Chiarantini and M. Benvenuti
© 2013 University of Oxford, Archaeometry 56, 4 (2014) 625–650
